Báo cáo khoa học "A policy Category Analysis Model for Toursim Promotion in China During the Covid-19 Pandemic Based on Data Mining and Binary Regression".

Bài báo cáo khoa học chủ đề: A policy Category Analysis Model for Toursim Promotion in China During the Covid-19 Pandemic Based on Data Mining and Binary Regression. Mời bạn đón xem!

119 60 lượt tải Tải xuống

Chia sẻ Báo cáo tài liệu

lOMoARcPSD| 36066900

Risk Management and Healthcare Policy Dovepress

open access to scientific and medical research

Open Access Full Text Article

O R I G I N A L R E S E A R C H

A Policy Category Analysis Model for Tourism

Promotion in China During the COVID-19

Pandemic Based on Data Mining and Binary

Regression

This article was published in the following Dove Press journal:

Risk Management and Healthcare Policy

Tinggui Chen

Lijuan Peng

Xiaohua Yin

Bailu Jing

Jianjun

Yang

Guodong Cong

Gongfa Li

School of Statistics and Mathematics,

Zhejiang Gongshang University,

Hangzhou 310018, People’s Republic of

China;

School of Management and

E-Business, Zhejiang Gongshang

University, Hangzhou 310018, People’s

Republic of China;

Department of

Computer Science and Information

Systems, University of North Georgia,

Oakwood, GA 30566, USA;

School of

Tourism and Urban-Rural Planning,

Zhejiang Gongshang University,

Hangzhou 310018, People’s Republic of

China;

Hubei Key Laboratory of

Mechanical Transmission and

Manufacturing Engineering, Wuhan

University of Science and Technology,

Wuhan 430081, People’s Republic of

China

Correspondence: Guodong Cong

Email cgd@mail.zjgsu.edu.cn

lOMoARcPSD| 36066900

Background and Aim: At the

end of 2019, the outbreak of

COVID-19 had a significant

impact on China’s tourism

industry, which was almost at a

standstill in the short-term. After

reaching the preliminarily stable

state, the government and the

scenic area management

department implemented a series

of incentive policies in order to

speed up the recovery of the

tourism industry. Therefore,

analyzing all sorts of social effects

after policy implementa- tion is of

guiding significance for the

government and the scenic areas.

Methods: Targeted as the social

effect with the implementation of

tourism promotion policy during

the COVID-19 pandemic, this

paper briefly analyzes the impact

of COVID-19 on the national

cultural and tourism industry and

selects several representative types

of tourism policies, crawls the

comment data of Weibo users, analyzes users’ perception and emotional preference to the

policy, and thus mines the social effect of various policies. Subsequently, by identifying the

social effects of various policies as dependent variables, a binary logistic regression model is

constructed to obtain the best combination of tourism promotion policies and promote the

rapid revitalization of the cultural and tourism industry.

Results: The results show that from the single policy, the social effect of the “safety” policy

is the best. From the perspective of combination policies, the simultaneous release of

“safety” policies and “economy” policies have the greatest social impact, which can drama-

tically accelerate the recovery of the cultural and tourism industry. Finally, this paper

proposes suggestions for policy formulation to improve the ability of the cultural tourism

industry to cope with crisis events.

Conclusion: These results explain the perceived effects of the public on the government

policies and can be used to judge whether the policies have been released in place. Based on

the above results, corresponding suggestions are proposed as follows: 1) the combination of

economic policies and security policies can achieve better results; and 2) the role of “opinion

leaders” can be played to improve the perceived effect of policies.

Keywords: online comments, social effects, combination optimization, data mining, binary

logistic regression, COVID-19

Introduction

With the progress of science and technology, as well as the improvement of

people’s living conditions and material level, China’s tourism industry has entered

a new take-off stage after more than 20 years of development. However, the

outbreak of COVID-19 in late 2019 greatly impacted on China’s tourism industry,

submit your manuscript

www.dovepress.com

Risk Management and Healthcare Policy 2020:13 3211–3233

3211

DovePress

http://doi.org/10.2147/RMHP.S284564

and incorporate the

Creative

Commons

Attribution

–

Non

Commercial

(unported,

v3.0)

License

(http://creativecommons.org/licenses/by-nc/3.0/).

accessing

the

work

you hereby accept the Terms. Non-commercial uses of the work are permitted without any further

permission from

Dove

Medical

Press

Limited, provided

the

work is

properly

attributed. For

permission for commercial use of this work, please see paragraphs 4.2 and 5 of our Terms (https://www.dovepress.com/terms.php).

Risk Management

and Healthcare

Policy downloaded

from https://www.dovepress.com/

by 113.23.110.30

on 11-Jun-2021

For personal use only.

lOMoARcPSD| 36066900

Chen et al Dovepress

wiping out more than 450 million tourist revenues. About

2 months after the enforcement of prevention and control,

on February 25, the national culture and tourism office

issued the guidance to lead the national scenic spots to

implement the epidemic prevention and control to make

recovery steadily. Meanwhile, local governments and sce-

nic spots have issued a series of promoting policies, such

as travel coupons, current limiting, 2.5-days off, which

aims to stimulate recovery of the tourism industry while

preventing COVID-19. However, the policies have

received mixed reviews from netizens, who have focused

on the safety issues brought about by the opening of scenic

spots. For the government and scenic spots, the most

concerning question is which kind of policies have the

most obvious stimulating effect on the tourism economy.

In order to answer these questions, it is necessary to

conduct an emotional analysis of a series of policies issued

by the government and scenic spots, measure the effect of

policy implementation according to online netizens’ com-

ments, and summarize the policy combination with the

best implementation effect, so as to help the government

and scenic spots maximize the validity to promote

recovery.

At present, domestic and foreign scholars have carried

out some researches on the implementation effect of tour-

ism policies, mainly including drawing on multi-

disciplinary research methods such as consumer behavior,

public management, and journalism and communication,

designing research scales and measurement processes, and

conducting research on the impact of tourism policies

based on tourism consumption behavior intention.

However, few scholars directly use online comment data

of tourism public opinion for research, but whether to

travel is easily affected by public opinion. Therefore, it

is of great significance to carry out text mining on tourists’

online comment data on tourism policy, so as to analyze

the social effect of policy implementation.

Based on this, this paper evaluates the effects of

a series of policies issued by the government and scenic

spots against the background of COVID-19. In particular,

this paper analyzes the present situation and the existing

supporting policy to select the representative tourism poli-

cies released by government, then reviews Weibo com-

ments data to understand users’ intention and emotion

preference, as well as the social effect of all kinds of

policy implementation. Besides, this article builds

a binary logistic regression model, making social effects

influenced by various policies as the dependent variable,

and making other factors (Weibo comments, comments

time, like quantity) as independent variables. In turn, it

combines the results of the single policy and mixed policy

with the most promoting effect for tourism, providing

suggestions for the government and the scenic spot to

accelerate the revitalization of the culture and tourism

industry.

The structure is as follows: Literature Review analyzes

the literature of tourism policy; Analysis on Supporting

Policies of Cultural Tourism Industry During COVID-19

discusses the tourism support policies; Data Mining Based

on Online Comments of Travel Policies During COVID-

19 conducts data mining on the online comments on tour-

ism policies; The Social Effect Analysis of Tourism

Policies Based on Binary Logistic Regression Model con-

structs a binary logistic regression model to measure the

social effects of tourism revitalization policies and seek for

the optimal combination of policies; Conclusions and

Suggestions summarizes the whole paper and provides

policy suggestions.

Literature Review

At present, there are text mining, emotion analysis, and

combinatorial optimization methods for policy implemen-

tation effect analysis in both domestic and international

research. These methods are still practiced in the study of

tourism policy implementation effect. Therefore, this

paper draws on the methods of evaluating the implementa-

tion effect of other policies to analyze the research status

of a tourism policy implementation effect.

Tourism policies mainly refer to policies that promote

tourism activities in various aspects. In order to understand

the impact of these policies on the actual tourism industry,

scholars have conducted in-depth studies with policy mea-

surement tools, and the representative results are as fol-

lows: David et al

introduces the research method of

system theory into the political field, and puts forward

the “political system theory” for the first time. It holds

that the interaction between policy system and environ-

ment is realized through “input, transformation and out-

put” in politics through “requirements and support”

existing in the environment. Rossouw and Saayman

demonstrated the relevance and necessity of using the

tourism satellite account (TSAs) as a tool for South

African decision-makers (especially tourism decision-

makers) to improve and expand the application of the

general equilibrium (AGE) model. The reasons for the

need for economic models for policy analysis and other

3212

submit your manuscript

www.dovepress.com

DovePress

Risk Management and Healthcare Policy 2020:13

Risk Management

and Healthcare

Policy downloaded

from https://www.dovepress.com/

by 113.23.110.30

on 11-Jun-2021

For personal use only.

 
lOMoARcPSD| 36066900 
 
 
Dovepress 
Chen et al
 
 
purposes were  expounded, and the new  and old  methods 
for  tourism  policy  modeling  were  summarized.  Steve
3
 
extended and applied a non-parametric method to estimate 
the effect of heterogeneous treatment, and examined how 
the policy effect changed over time. The study showed that 
this method had potential value in evaluating the impact of 
a  series  of  environmental  policies  and  environmental 
impacts. Dong and Liu
4
 used a threshold model and quan- 
tile  regression  model  to  explore  the  threshold  effect  of 
policy  power  on  policy  implementation  effect  and  the 
transformation of policy implementation effect in different 
development stages. The results showed that the effect of 
industrial  policy  varied  greatly  in  different  development 
stages, and policy power had a significant threshold effect 
on its operation intensity. Matousek et al
5
 studied how the 
uncertainty of economic policy affected the capital short- 
age of financial companies in the new crisis of the event. 
The study found that if the response of policy-makers and 
politicians was not timely and decisive during the severe 
market  downturn,  then  there  would  be  a  price  for  the 
delay. Joseph
6
 used the classical stability theory to model 
the complex social and political system, and the causes of 
social  collapse  were  studied.  Chen  et  al
7,8
  analyzed  the 
polarization of public opinion in group behavior based on 
the SIR model, and at the same time considered the influ- 
ence of external information and individual internal char- 
acteristics on the polarization of public opinion. Alexander 
and  Yusaku
9
  took  Japan  as an  example,  and  carried  out 
a survey experiment to let citizens understand the policy 
information  of  the  US,  which  varied  according  to  the 
source,  policy  content,  and  problem  prominence.  The 
results showed that when the source signal (Trump attribu- 
tion)  led  to  negative  views  in  the  US,  policy  content 
(cooperative than uncooperative) had a greater impact on 
shaping the opinions of American citizens. Ho
10
 analyzed 
the success factors of a convalescent tourism policy, and 
discussed the priority and importance of success factors of 
the recuperation tourism policy by using an AHP analysis 
method. Yin
11
 redefined the theory of the political system 
based  on  the  agenda  setting  theory  in  communication 
science,  and  constructed  a  “network  public  opinion  deci- 
sion hypothesis model” to analyze how and to what extent 
network  public  opinion  affects  government  decision- 
making. In addition, as there are many types of policies, 
some  scholars  have  classified  the  policy  tools,  among 
which Rothwell and Zegvelk
12
 were the most representa- 
tive ones. When studying industrial innovation and public 
policy, they divided the policy tools into three types: 
Environmental policy tools, Supply-oriented policy tools, 
and Demand-oriented policy tools. 
In  spite  of  various  studies  on  tourism  policy,  even 
though some scholars explore the effects of tourism policy 
implementation, few scholars study the optimal combina- 
tion of tourism policy, namely through the data mining of 
online  reviews,  research  released  by  any  combination  of 
tourism  policy  content  can  bring  about  a  better  social 
effect. In addition, as far as the social  effects of tourism 
policies  are  concerned,  there  is  still  a  lack  of  objective 
analysis based on network text data, so the research meth- 
ods need to be innovated. In view of this, this paper takes 
the  Weibo  comments  as  the  data  source  to  find  out  the 
optimal  combination  that  can  make  tourism  promotion 
policies  exert  the  greatest  social  effect,  so  as  to  provide 
suggestions  on  the  content  form  of  policies  issued  by 
government departments. 
In order to understand the actual effect of policy imple- 
mentation, scholars have conducted relevant studies on the 
evaluation  methods  of  the  policy  implementation  effect. 
Kim et al
13
 studied the “free public transport” policy and 
the  “citizen  participation alternative  day  no  driving”  sys- 
tem implemented by Seoul government in order to reduce 
dust.  They adopted a  regression  analysis method  to ana- 
lyze the impact of traffic on fine dust, and used text mining 
technology to analyze the response of two traffic policies 
and  citizens’  petition,  and  proposed  policy  improvement 
direction according to the research results. Lee et al
14
 used 
text mining technology and emotional text analysis tech- 
nology  to  test  the  online  evaluation  of  Japanese  tourism 
websites after the 2011 Japanese tsunami, and studied the 
impact  of  the  tsunami  on  Japan’s  tourism  industry.  The 
results show that  the low exchange rate  and the  positive 
sentiment  of  online  reviews  from  tourism  websites  have 
a positive impact on the number of tourists to Japan. Chen 
et al
15,16
 used text mining and emotion analysis methods to 
implement  the  effect  of  online  education  under  the  epi- 
demic  situation.  Also,  the  experience  effect  was  studied 
from  the perspective  of user  and  platform. Bucek
17
 used 
text mining technology to investigate the twitter accounts 
of  US  President  Barack  Obama  from  March  2012  to 
January 2016, so as to study whether politicians’ behavior 
on social  networks  would  affect actual  economic policy. 
Qi  et  al
18
  analyzed  the  innovative  fiscal  policy  texts  of 
provinces  and  cities  in  China  based  on  python,  so  as  to 
understand the regulation and control of tax policies. The 
results declared that fiscal policy and regional innovation 
capability showed obvious spatial heterogeneity, and R&D 
Risk Management
 and Healthcare
 Policy downloaded
 from https://www.dovepress.com/
 by 113.23.110.30
 on 11-Jun-2021 
For personal use only.
 

lOMoARcPSD| 36066900

Chen et al Dovepress

investment and industrial structure were the main sources

of improving innovation capability. Sun

proposed an

evaluation scheme to optimize the policy process by

using sorting and clustering strategies, and proposed

a combination algorithm combining the algorithm with

policy priority evaluation. Finally, satisfactory policies

and rules were selected to improve the matching speed.

The experimental results showed that the method reduced

the matching operation and improved the evaluation effi-

ciency. Pellesova

discussed the technology of using an

econometric method to optimize economic policy, mainly

studied the target variable method, and explored the

advantages and disadvantages of the selected method and

its possible application in the formulation of optimal eco-

nomic policy. Dash and Kajiji

proposed a mixed integer

nonlinear objective program (MINLGP), which aimed to

solve the model of multi-objective portfolio optimization

decision-makers facing binary hedging decision-making

between Portfolio Rebalancing periods. It was found that

when percentile risk measurement was used, the expected

catastrophic loss of the best diversified portfolio of hed-

ging was obviously less than that of non hedging products.

Yang et al

evaluated the impact of the recently imple-

mented policy of improving gasoline quality on reducing

the concentration of fine particulate matter (PM). The

study illustrated that it was difficult to completely solve

the particulate pollution problem in China by a single

policy, and a series of policy system designs were needed

to alleviate this problem. Geng and Kamal

analyzed the

optimal policy options for two main types of price regula-

tion in China, and found that the domestic optimal external

reference pricing (ERP) policy reduced domestic prices

while maintaining export incentives for enterprises.

Maansi and Nomesh

measured the efficiency of Indian

high courts using Data Envelopment Analysis (DEA).

Secondly, they studied the impact of including pending

cases on judicial efficiency. Kelly et al

proposed

a composite index to assess home-heating energy-poverty

risk across 18,641 small area clusters in Ireland. The index

offered the capacity to analyze changes in energy-poverty

risk associated with specific policy intervention proposals,

including major contemporary environmental policy tran-

sitions. Elke and Andreas

examined the acceptance of

burden sharing rules by using multivariate binary and

ordered Probit models. The rule involved the costs of the

German energy transition, which was one of the most

challenging and disputed national climate and energy pol-

icy measures. The results declared that polluter-pays rule

had by far the highest support. Carlotta et al

used

Random Forest and Gradient Boosted Regression Trees

algorithms to predict the response of freshwater ecosys-

tems to multiple anthropogenic pressures, with the goal of

informing the definition of water policy targets and man-

agement measures to recover and protect aquatic biodiver-

sity. Ekaterina et al

confirmed a positive impact of the

zone merger on the gas trading market’s spatial equili-

brium and indicated the causes of remaining market inef-

ficiencies used by an extended parity bounds model, which

provided a tool for evaluating the efficiency of policy

decision-making. Dong and Liu

analyzed the optimal

functioning power of policies and determined the direction

of future policy implementation. This paper utilized the

COPA framework to analyze policy evolution in respect of

the new-energy vehicle industry (NEVI). Smith and

Hasan

discussed the methods and practices involved in

quantitative evaluations of implementation research stu-

dies, and analyzed available measurement methods for

common quantitative implementation outcomes involved

in such an evaluation-adoption, fidelity, implementation

cost, reach, and sustainment, and the sources of such

data for these metrics using established taxonomies and

frameworks.

According to the above literature, many scholars have

studied public policies and built many models. However,

due to the epidemic, the formulation of tourism policies

should also take the characteristics of emergencies, health,

and safety into account. At present, there are few studies

on this aspect in the literature. Therefore, this paper takes

the implementation of the promotion policy as the back-

ground, classifies policy, analyzes online reviews, uses the

binary regression model to evaluate the implementation

effect under a different combination of policies, and finally

selects policy combination ways with better effect, which

provide a decision basis and support for the release of the

government’s policy.

Analysis on Supporting Policies of

the Cultural Tourism Industry

During COVID-19

The status of the cultural and tourism industry during

COVID-19 needs to be analyzed before analyzing the

implementation effect of the tourism revitalization policy

issued by the government. In light of the update of the

epidemic, when should the government issue policies?

What policies have been issued? What topics have been

3214

submit your manuscript

www.dovepress.com

DovePress

Risk Management and Healthcare Policy 2020:13

Risk Management

and Healthcare

Policy downloaded

from https://www.dovepress.com/

by 113.23.110.30

on 11-Jun-2021

For personal use

only.

lOMoARcPSD| 36066900

Dovepress

Chen et al

discussed on Weibo? Which media does report the policy

heavily? What kind of publicity effect did it have? Based

on this, the current situation of tourism is elaborated

below, and the current implementation of decentralized

tourism revitalization policies are summarized. More

topics related to tourism policies discussed by the public

are selected through the Microblog platform, and the

online comment data of the public are retrieved, so as to

prepare for the subsequent in-depth analysis of the social

effects brought by the implementation of policies.

A Brief Analysis of the Current Situation

of the Cultural Tourism Industry in China

The outbreak of COVID-19 at the end of 2019 spread

rapidly across the country and seriously affected China’s

economic and social development and people’s livelihood.

In particular, the impact of the epidemic on the tourism

industry includes both direct losses of many tourism enter-

prises and related employees and indirect losses of related

industries in the tourism industry. Just consider the Spring

Festival, the direct economic loss caused by the shutdown

of China’s tourism industry is as high as 400 to 500 billion

Yuan, resulting in the annual expectation to change from

a “year-on-year growth of about 10% to a negative growth

of 14% to about 18%. On January 24, the General Office

of the Ministry of Culture and Tourism of China issued an

Urgent Notice on COVID-19 prevention and control to

suspend the business activities of tourism enterprises,

requiring travel agencies and the online tourism industry

of China to suspend the operation of group tourism and

“air ticket + hotel” tourism products. As of February 1,

450 million people had canceled or postponed their Spring

Festival trips.

Statistics show that during the Spring Festival in 2018,

the country received 386 million tourists, rising

12.1% year on year. Tourism revenue reached 475 billion

Yuan, rising 12.6% year on year. During the Spring

Festival in 2019, 415 million tourists traveled across the

country, rising 7.6% year-on-year. Tourism revenue

reached 513.9 billion Yuan, up 8.2% year on year. By

2020, more than 450 million tourism revenues have been

lost. At the same time, online travel agency (OTA) plat-

forms such as Ctrip and Tuniu have invested more than

hundreds of millions of Yuan in cancellation fees, and

more than 260,000 travel agencies are struggling.

According to STR, the hotel occupancy rate on the

Chinese mainland peaked at 70% in early January 2020,

and began to plummet a day later, plummeting to 17% on

January 26.

As can be seen from the above brief description of the

current situation, with the sudden outbreak of COVID-19,

the Chinese government responded quickly and took var-

ious measures to block the transmission channels of

COVID-19. As it turned out, China suddenly went into

a “dormant” state. However, the economic development,

especially the development of the tertiary industry, has

encountered regression this year. The cultural and tourism

industries, which are characterized by crowd gathering,

bear the brunt of the contraction. How much impact will

this “disaster” have on economic development? Can the

cultural tourism industry survive? Is tourism a fragile

industry? Experts, scholars, and ordinary people are con-

cerned about these issues.

Supporting Policies of the Cultural and

Tourism Industry During COVID-19

The cultural and tourism industry is a modern service

industry with human service targeted with human services.

Its basic feature is the movement of people, and the pursuit

of security is the primary condition for people’s needs. By

May 2020, the epidemic prevention and control situation

in China has been stable, laying the foundation and creat-

ing the basic conditions for people to travel safely during

the May 1 holiday. Since the outbreak, the industry has

acquired high attention from international organizations to

the central ministries and commissions, and from local

government, industry association to the tourism enter-

prises, and tourism-related aspects. Taking positive action

and dealing with unprecedented pressure that the tourism

industries are facing should manage well in two aspects:

the first is to provide epidemic prevention and control,

the second is to introduce all kinds of policy for supporting

all kinds of damaged industries and enterprises. As for the

various policies issued by the government, through the

Internet reports of major media, the public can express

their own opinions and cognitive emotions on the public

events they care about, thus forming mixed opinions on

the revitalization of tourism policy.

Based on the comprehensive analysis of Weibo content

and online comment related to the policy of “revitalizing

tourism”, this paper summarizes the policies that netizens

have paid close attention to and discussed enthusiastically,

striving to cover and describe the public’s response to the

government’s policies in the cyberspace to the greatest

Risk Management

and Healthcare

Policy downloaded

from https://www.dovepress.com/

by 113.23.110.30

on 11-Jun-2021

For personal use

only.

lOMoARcPSD| 36066900

Chen et al Dovepress

extent. Firstly, this paper summarizes government policy

as five major series and regards them as the first-level

subject category, collects the network comment, and

chooses the topic with most comments on each category

acknowledged by the CCTV news, the blue whale finan-

cial journalist work platform, Sina News, People’s Daily,

and the Paper. Further, it uses the octopus Weibo topic

review as a selected collection to distinguish between

policy and Weibo topics. In order to distinguish policies

and the post in Weibo, ”#” is used to annotate. The

categories of policies and online discussion topics for

revitalizing tourism are shown in Table 1. (The number

of the comments in Table 1 are all from Weibo. The

limitation search of Weibo is as follows: first, it may

produce a lot of meaningless information; second, the

search of Weibo still adopts the traditional way, and the

information content cannot be reprocessed.)

Data Mining Based on Online

Comments of Travel Policies During

COVID-19

The Internet represents an important channel for people to

express their interests and emotions. Since the COVID-19

outbreak, people have become more vocal on the Internet

due to restrictions on travel, and there has been a lot of

discussion about the travel-related topic. As a reflection of

public sentiment, the influence of online public opinion is

not only manifested in its influence on major develop-

ments, but also penetrates the political level, becoming

an important channel for the government to listen to the

voices of the people and understand public opinion. In

order to explore the public’s attitudes towards various

official tourism policies during COVID-19, this section

conducts data mining on the comments on the Weibo

topic mentioned above. Firstly, the data of netizen com-

ments were selected and cleaned, then the pre-processed

netizen comments were analyzed from the perspective of

policy perception, and the categories of policies were

divided. Finally, the visualization analysis and emotional

analysis of netizen comments were conducted based on

different policy categories.

The Selection and Cleanout of

Comments

According to the second-level microblog topics, comments

on Weibo topics at different times were selected as the

research objects. In the first category, under the

government policy “online tourism,” the topic #Travel

around China online# and the netizens’ comments from

April 25 to May 5, 2020 were selected. In the second

category, under the government policy of #Many pro-

vinces define 2.5 days off#, the netizens’ comments on

five topics on January 14, 2020, from March 19 to April 1,

from April 26 to July 24 were selected. In the third

category, under the government policy #Measures in sce-

nic spots (free tickets or restricted access or real-name

purchasing system)#, eight hot topics were selected and

the discussions were divided into two categories: because

Huangshan Mountain scenic spot was congested, the gov-

ernment introduced a series of policies for the situation

rapidly to adjust scenic spot, which arouses heat discus-

sion in the Internet. The comments with regard to

#Huangshan Scenic Spot starts emergency plan# from

April 5 to April 6 were collected. After the introduction

of policies for national scenic spots, the comments were

collected from February 19 to March 18, and from

April 13 to April 15. In the fourth category, under the

government policy #The tourism industry in many places

across the country resumes business#, the netizens’ com-

ments from February 21 to April 29, 2020 and from

July 14, 2020 to July 15 were collected. In the fifth

category, under the government policy #Travel coupons

issued in many places across the country#, the netizens’

comments from March 26, 2020 to July 1, 2020 were

selected.

Through octopus crawling to get the corresponding

Weibo topic comments, as the original data contains

some trivial comments, these trivial comments will inter-

fere with the subsequent analysis results, so the trivial

data needs to be processed. In view of the requirement of

emotion analysis, the data needs to be processed. First,

the symbols, emoticons, punctuation, and some useless

marks in the comments should be removed. The second

is to clean out comments irrelevant to the policy, such as

advertising, the word count comments, etc. The software

Python is mainly used to remove Chinese and English

symbols, emoticons, invalid texts (such as “ah”, “en”,

and other modal words, comments that are not related to

the tourism topic of this article, and irrelevant advertise-

ments), and some key codes to realize this function are

as follows: import re; the line = line. decode (“utf8”);

String = re.sub (“ [\s +\!\/_ $% ^ * (+\“\”] + | [+ -.??, ~

@ # $% and * () “+”. decode (“utf8”), ““. decode

(“utf8”),line). The data after cleanout is shown in

Table 2.

3216

submit your manuscript

www.dovepress.com

DovePress

Risk Management and Healthcare Policy 2020:13

Risk Management

and Healthcare

Policy downloaded

from https://www.dovepress.com/

by 113.23.110.30

on 11-Jun-2021

For personal use only.

lOMoARcPSD| 36066900

Dovepress

Chen et al

Table 1

Policies on Revitalizing Tourism and Classification of Online Discussion Topics

The Primary Category

(Government Policy)

The Secondary Category

(Weibo Discussion Topic)

The Number of Comments

Collected on Weibo Topics

Travel coupons issued in many places across the

country

#National version of the consumer coupons will be

launched tomorrow# Here comes the strategy

5439

#Hangzhou will issue 1.68 billion Yuan consumption

coupons#

#Wuhan will issue 500 million Yuan consumption coupons#

#Zhejiang Jiaxing issues 200 million Yuan consumption

voucher#

Many provinces define 2.5 days off

#Zhejiang encourages 2.5 days off a week#

17,731

#2.5 days of weekend vacation system will be implemented

in three places#

#Jiangxi tries out 2.5-day flexible work and rest on

weekends#

#2.5-day flexible vacation system implemented in Yichang,

Hubei Province#

#It is suggested that one of three flexible weekend vacation

systems can be implemented#

Measures in scenic spots (free tickets or

restricted access or real-name purchasing

system)

#Notice on stopping receiving tourists in Huangshan Scenic

Area#

30,961

#112 scenic spots in Sichuan are free of admission to all

tourists in April#

#Huangshan scenic spot is congested#

#Huangshan Scenic Spot starts emergency plan#

#The Ministry of culture and tourism requires that the

opening of scenic spots should strictly control the flow#

#Real-name purchasing system is required for Sichuan

scenic spots to reopen#

#Visitors are required to be 1 meter apart in the opening of

the scenic area#

#During the epidemic period, only outdoor areas are

opened#

Online travel (live travel or live commerce)

#Travel around China online#

38,216

The tourism industry in many places across the

country resumes business

#The West Lake in Hangzhou will open up orderly from

today#

5306

#Ministry of culture and tourism issues notice to resume

inter-provincial team Tourism#

#Yunnan tourism industry resumed business#

Risk Management

and Healthcare

Policy downloaded

from https://www.dovepress.com/

by 113.23.110.30

on 11-Jun-2021

For personal use only.

lOMoARcPSD| 36066900

Chen et al Dovepress

Table 2

Selected Data After Cleanout

Classification of

Policy

Perspectives

Economy

Safety

Idle

Feasible

Specific policies

Consumption

coupon

Scenic

spot

policy

Huangshan

crowded

Travel around

China online

2.5 days

off

a week

Resuming

tourism

industry

Resuming inter-

provincial travel

Number of

comments

4,473

8,286

20,802

26,808

17,156

3,251

1,710

Percentage of

comments

5.42%

67.76%

20.80%

6.01%

Analysis of Netizens’ Perception of the

Policy

The Classification of Comments

Netizen perception refers to whether the netizen’s under-

standing of the policy is consistent with policy-makers’

desired goals. From the perspective of policy, this paper

divides the above five policy categories into four cate-

gories, namely, economy, safety, idle, and feasible cate-

gory. The economy category refers to the policy-makers’

expectation to achieve the purpose of promoting the econ-

omy through the policy, such as “consumption coupon”.

The safety category refers to the policies formulated by

policy-makers to avoid crowd gathering and virus cross-

infection in scenic spots from the perspective of safety,

such as “Scenic spot policy”, “Huangshan crowded”, and

“Travel around China online”. The idle category refers to

the policy made by the policymaker from the perspective

of whether tourists have free time to travel, such as “2.5

days off a week”. The feasible category refers to whether

the scenic spot is open, such as “Resuming tourism indus-

try”, “Resuming inter-provincial travel”.

Although policymakers formulate from the above four

perspectives, different netizens have different understand-

ing of policy. Taking coupon for example, some netizens

consider it from an economy perspective, and point out

that “in order to stimulate consumption, various places

think about the different ways”. From a safety perspective,

some netizens think “it will lead to offline congestion,

which is not safe”. Therefore, in order to understand

netizens’ perception of the four types of policies, we

divided netizen comments into five types according to

their comment content and designated tag numbers for

subsequent data processing. The specific division criteria

are shown in Table 3.

According to the above criteria, all the comments made

by netizens should be classified. Since the total number of

comments on all policies is as huge as 82,486, TextCNN

convolutional neural network is adopted in this paper to

Table 3

Policy Division Criteria from the Perspective of Netizens

Classification of

Netizens’

Perspective

Tag

Number

Meaning

Comment Examples

Economy

When it comes to money, the economy, etc

“Just back to work, no money”

Safety

When it comes to epidemic situation, safety, etc

“Is it safe? The epidemic is not over”

Idle

When it comes to holidays, whether you have time to

travel, etc

“Oh, where did you get your vacation when you just

went to work?” “school is not allowed to go out”

Feasible

From the perspective of whether the scenic spot is

open or not, we can see whether the tour is feasible

“Many scenic spots are not open. How can I get

there?”

Others

Other netizens’ comments except for the above four

Categories

The online comments of the public on the policies during

the epidemic period were obtained from the websites. As

they were relatively redundant and dispersed, the software

ROST CM5.8.0 was used to classify the emotional ten-

dency of user comments, and analyze the policies that

aroused better public perception and the emotional ten-

dency of the public on such policies. Then, according to

the visual analysis technology of the semantic network, the

hot public concern about the travel during the epidemic

period is mined, which can be used as the basis for the

follow-up test of whether the public perceives the policy.

Sentiment Analysis

ROST CM

software is a digital research platform for

humanities and social sciences based on content mining. It

is a group of digital academic research platforms with close

functional connections, which can collaborate intelligently

with each other, and finally conduct an intelligent analysis of

humanities and social sciences according to a certain para-

digm. ROST CM software is capable of semantic network

and emotion analysis. In this paper, ROST CM5.8 is used for

emotion analysis, and the analysis results are used to inte-

grate the proportion of positive, neutral, and negative com-

ments brought by the implementation of four kinds of

policies during COVID-19, as shown in Table 5.

According to the proportion of positive, neutral, and

negative comments in Table 5, 37.90% hold a positive

attitude, 54.68% hold a neutral attitude, and only 7.42%

hold a negative attitude towards the online tourism policy

issued by the government, indicating that most netizens

have a favorable impression and support this policy. For

Table 5

Sentiment Analysis of Various Weibo Topics

Classification

Policy Category

Topic

The Proportion of

Positive Emotions

The Proportion of

Negative Emotions

The Proportion of

Neural Emotions

Safety

Online travel

Travel around China online

37.90%

7.42%

54.68%

Scenic spot measures

Measures for Huangshan

Scenic Area

31.53%

37.96%

30.51%

National scenic spot

measures

48.99%

16.21%

34.79%

Idle

Many provinces define 2.5

days off

Many provinces define 2.5

days off

37.49%

23.18%

39.33%

Economy

Coupons issued in many

places across the country

Coupons issued in many

places across the country

46.78%

15.04%

38.18%

Feasible

National Tourism

recovery

Resuming tourism industry

throughout the country

49.91%

18.78%

31.31%

Resuming inter-provincial

travel

47.31%

15.60%

37.09%

3220

submit your manuscript

www.dovepress.com

DovePress

Risk Management and Healthcare Policy 2020:13

Risk Management

and Healthcare

Policy downloaded

from https://www.dovepress.com/

by 113.23.110.30

on 11-Jun-2021

For personal use only.

lOMoARcPSD| 36066900

Dovepress

Chen et al

tourists to gather together after Huangshan scenic spot

opened, although the government issued a series of mea-

sures to control the flow, 37.96% of the population held

a negative mood for this event, which is more than the

number of people with positive emotions. It indicates that

government departments should prepare for emergency

measures and security measures before opening a scenic

spot, rather than wait and solve problems when they arise.

Measures of the scenic spots in the country mainly include

“online booking”, “flow restriction”, etc. Then, 48.99% of

the people show positive emotions, which are far more

than negative emotions, indicating that the implementation

of scenic spot measures in various places has a significant

effect, which can give people enough sense of safety

during the trip. The negative comments on the “2.5 days

off” policy reached 23.18%, when the positive comments

reached 37.49%. Although the government called on

every medium enterprise to extend rest time, under

COVID-19, employees hope to recover economic loss as

soon as possible, so more people heold a neutral attitude.

Under the “economy” policy, 46.78% of netizens held

a positive attitude, while only 15.06% hold a negative

attitude, indicating that most netizens held a positive atti-

tude in support of the policy. It can be seen from the

“feasible” policy that the number of people holding posi-

tive emotions in this policy is much higher than the

number of people holding negative emotions, by more

than 30%, indicating that this policy is supported by the

masses.

Visual Analysis Based on Semantic Network Under

Different Policy Categories

According to ROST CM analysis of the proportion of

positive and negative opinions in people’s online com-

ments, those posts under the four categories of policies,

namely “safety”, “idle”, “economy”, and “feasible”, are

generally positive. Here, in order to obtain the major

concerns of the public under each type of policy, semantic

network visual analysis was adopted to analyze the online

comment data of the four policies under the background of

COVID-19. The semantic network is one of the represen-

tations of an artificial intelligence program, which

expresses human knowledge construction in the form of

a network. It consists of arcs between nodes, where nodes

represent concepts (events or things), and arcs represent

relationships between them. The semantic network dia-

gram is used to represent the degree of association

between words and reflect the most concerned words in

user comments. The main purpose is to find the words that

are mentioned most in netizens’ comments, so as to judge

whether people perceive the content of the policy. By

ROST CM5.8.0 analysis, a relevant semantic network

diagram can be obtained. Figures 1–7 represent the seman-

tic network diagram of user comment content under each

travel topic, and represent the relationship between words

in each user comment content.

The Analysis of Safety Policies

The semantic graph of the Internet is obtained by analyz-

ing the comments made by netizens under the topic of

#Travel around China Online#.

From Figure 1, netizen focus more on the reporters and

anchors of online travel live-streaming. “Travel around

China online” is broadcasted at a fixed time every day.

Generally speaking, beautiful anchors or journalists attract

more audience, especially web celebrity “Weiya”.

Affected by the COVID-19 epidemic, people can neither

travel abroad nor buy travel products. Through online

mode, people can be personally involved and promote

the consumption of tourism products in remote areas.

According to Figures 3 and 4, Xinjiang, Sanya, Shanxi

Pingyao, Changbai Mountain, Guilin, and other places

have attracted more netizens’ attention. Through live

streaming of local food and scenic spots, people want to

travel in person, which accelerates the sales of tourism

products.

In addition, the Internet semantic graph of Figures 2

and 3 is obtained by analyzing the comments made by

netizens under the topics #Huangshan Scenic Spot mea-

sures# and #National Scenic spot measures# on Weibo.

Since announcing it was open, Huangshan Mountain

scenic area has been crowded, indicating that, after stabi-

lizing of the epidemic, the majority of people are chasing

for tourism, but congestions cause certain difficulties for

epidemic control. Huangshan Mountain scenic area started

the emergency plan, stopped serving tourists, preventing a

widespread infection epidemic. It can be seen from Figure

2 that Tomb-sweeping Day (the first holiday after the

epidemic), has become the relaxation of travel for people

all over the country after they have been “confined” for

a long time, which also increases the difficulty for the

government to prevent and control the epidemic. “Wear

Mask” has become important and necessary. In Figure 3,

the nodes of “Anhui”, “Huangshan”, and “local” indicate

that most tourists to Huangshan scenic spot are residents

of Anhui province. Many tourism enterprises, such as

Risk Management

and Healthcare

Policy downloaded

from https://www.dovepress.com/

by 113.23.110.30

on 11-Jun-2021

For personal use only.

lOMoARcPSD| 36066900

Chen et al Dovepress

Figure 1

Semantic graph of #Travel around China Online#.

Figure 2

Semantic graph of #Huang shan Scenic Spot measures#.

Ctrip and Feizhu, have launched local and provincial tours

to narrow the traveling scope and make the public feel

more assured. As can be seen from the “current-limiting”,

“making an appointment”, “free”, and “time” nodes, the

government has taken measures such as online booking,

limiting the visiting number of scenic spots, offering free

tickets, and extending the opening hours of scenic spots to

provide people with a safe place to travel, and at the same

time actively controlling the mass influx of people into

scenic spots. Despite a series of measures to control flow,

3222

submit your manuscript

www.dovepress.com

DovePress

Risk Management and Healthcare Policy 2020:13

Risk Management

and Healthcare

Policy downloaded

from https://www.dovepress.com/

by 113.23.110.30

on 11-Jun-2021

For personal use only.

lOMoARcPSD| 36066900

Dovepress

Chen et al

Figure 3

Semantic graph of #National Scenic spot measures#.

Figure 4

Semantic graph of #Many provinces define 2.5 days off#.

for the emergence measures taken by Huangshan, 37.96%

of people hold a negative mood for this event, which is

more than the number of positive emotions. It indicates

while recovering tourism, government should consider

unexpected circumstances, optimize emergency measures,

and avoid crowding in advance.

Risk Management

and Healthcare

Policy downloaded

from https://www.dovepress.com/

by 113.23.110.30

on 11-Jun-2021

For personal use only.

lOMoARcPSD| 36066900

Chen et al Dovepress

Figure 5

Semantic graph of #Coupons issued in many places across the country#.

Figure 6

Semantic graph of #Resuming tourism industry throughout the country#.

The Analysis of Idle Policies

The semantic graph is obtained by analyzing the com-

ments made by netizens under five topics.

In order to stimulate the recovery of the tourism indus-

try, the government proposed the “2.5 days off” policy.

Hubei, Zhejiang, Jiangxi, Hunan, and other provinces

responded positively. Figure 4 shows that around the

“rest” node, most people mention “implement”, “work

overtime”, “double rest”, “civil servants”, and “private

enterprise”. It shows that most of the private personnel

3224

submit your manuscript

www.dovepress.com

DovePress

Risk Management and Healthcare Policy 2020:13

Risk Management

and Healthcare

Policy downloaded

from https://www.dovepress.com/

by 113.23.110.30

on 11-Jun-2021

For personal use only.

lOMoARcPSD| 36066900

Dovepress

Chen et al

Figure 7

Semantic graph of #Resuming inter-provincial travel#.

hope to implement 2.5 days off as early as possible, and

guarantee their overtime pay. For state-owned enterprises,

organizations and institutions, employees rarely work

overtime compared with people from private sectors, so

they do not extravagantly hope for 2.5 days off. It can be

seen from the nodes “economy” and “development”, this

policy is mainly aimed at stimulating people’s consump-

tion in holidays and developing the economy. At present,

most enterprises have responded positively to this policy

by changing the former bonus payment into “complimen-

tary travel products” or “company incentive group travel”,

which promotes the recovery of the tourism industry. The

sentiment analysis results of Table 5 show that negative

emotions reached 23.18%, and the number of positive

emotion accounts for about 10%. Some people paid hourly

or daily encounter huge economic pressure, so they reduce

the demand for tourism and extend working hours to

afford their family expenses.

The Analysis of Economy Policies

The semantic graph is obtained by analyzing the

comments made by netizens under four sub-category

topics.

In Figure 5, the above analysis result shows that the

national policy of issuing coupons aims to “stimulate

economy”, “benefit the people”, “provide public welfare”,

and “stimulate the market”. The coupon platforms include

“Alipay”, “Dianping”, and “Mafengwo”. As the most

common application in the national people’s life,

“Alipay” has spread to all industries. Hangzhou, the city

where Alibaba is located, also responded positively to the

government’s policies and issued 1.68 billion Yuan of

consumption coupons. “Wuhan”, as an important node, is

the city hit the hardest by the epidemic. For this reason,

Wuhan issued 500 million Yuan of consumption coupons

to stimulate Wuhan’s economy, encouraging small and

middle enterprises to resume work and production, and

increasing people’s welfare. According to the

“Consumption” node in Figure 5, the government’s pur-

pose of promoting tourism consumption by issuing con-

sumption vouchers can be well accepted by the public.

People can buy travel vouchers in advance, which can be

used after the epidemic stabilizes, and it is also a kind of

advance consumption. At the same time, the government

also advocates daily consumption and issues a series of

coupons. “Supermarket”, “the Mall”, and “Market” are all

perceptions of the coupon policy. If the government wants

to emphasize tourism vouchers, the content of the infor-

mation should be clear when the policy is issued. The node

“Positive” indicates that netizens are optimistic about the

epidemic situation in China and are eager to travel after

the epidemic stabilizes.

Risk Management

and Healthcare

Policy downloaded

from https://www.dovepress.com/

by 113.23.110.30

on 11-Jun-2021

For personal use only.

lOMoARcPSD| 36066900

—

Chen et al Dovepress

The Analysis of Feasible Policies

The semantic graph of the network is obtained by integrating

and analyzing the comments of the netizens under three sub-

categories of Weibo topics, as shown in Figures 6 and 7.

During April and the Tomb-Sweeping Day, that is, when

the COVID-19 was gradually controlled, the tourism industry

across the country resumed business one after another.

Yunnan, as one of the major winter tourism destinations in

China, relies mainly on the cultural tourism industry sup-

ported by the floating population economy. Cultural and

tourism authorities at all levels are restoring confidence,

optimizing the modern tourism governance system, and

opening several scenic spots while strengthening of the epi-

demic prevention and control. In July, when the epidemic

was controlled, the ministry of Cultural and Tourism

announced the resumption of inter-provincial travel. At the

same time, cinemas began operating. From the node “entry

and exit”, “unemployment”, “half a year”, and so on, the

issued policies for returning to work make the unemployed

encouraged, ignite people’s inspiration to travel abroad, and

solve the problems for employees with the need to leave the

country for work or enterprises with foreign business. The

node “nucleic acid”, “testing”, and “ protect” show that

although the epidemic is in a stable state at present, the

country has not taken the prevention and control of the

epidemic carelessly.

The Social Effect Analysis of

Tourism Policies Based on Binary

Logistic Regression Model

The logistic regression model mainly studies the probabil-

ity P of some phenomena and discusses the factors related

to the probability P. In this article, studying whether peo-

ple perceive the government policy belongs to the 0–1

binary classification variables. Therefore, by constructing

a strictly monotone function Logistic (P) to study the

model between P and the independent variables, this

nodes after the visual characteristics of tourism revitaliza-

tion policies under COVID-19, the netizen-related vari-

ables are selected from the perspective of netizens,

including: total number of Microblogs, the timeliness of

comments posted, the division of policy in the view of

user, users’ emotional score (opinion tendency), thumb up

number in netizens comments, the netizen comments, the

gender of the netizens, the number of fans, the number of

the followers, the degree of activity, and the level of

development of COVID-19. These 11 variables obtained

by the network data after the expansion of crawl can fully

show the impact of netizens personal influence, individual

opinion on the effects of the policy implementation. The

meanings of 11 variables are shown in Table 6.

The Construction of Binary Logistic

Regression Model

The dependent variables set in this paper are binary classifi-

cation variables (ie, the Boolean variables), therefore, the

binary classification logical model (Binary Logistic

Regression) is adopted to study the factors affecting netizens’

perception to policy, as well as set an optimal combination of

policy and explain the influence of factors on perception

effect. In the regression model, the independent variable is

, and the dependent variable Y represents whether the

policy content is perceived from the perspective of Internet

users. ε is the error term, assuming that it is independent of

other variables; β

is the regression coefficient in logistic

regression; In

represents the logarithmic change value

of the ratio for the probability of occurrence to non-

occurrence when X

changes a unit. By referring to the

definition of logistic model in literature,

the influencing

factor model of netizens’ perceived effects on tourism poli-

cies is constructed, as shown in Equation (1).

Þ ¼ β X

þ β X

—

1 2 3 4 5

paper selects the binary logistic regression model.

(1)

Variable Selection and Data Definition

From the above analysis, various policies to revitalize

tourism during COVID-19 arise people’s attention and

discussion on the Internet. As the main channel of major

government policy, the new media must consider the guid-

ing force of influential people, who are represented by

a large quantity of Weibo and fans. Combining the analy-

sis of the emotional distribution and semantic network

The combination of different types of policies cause diverse

social effects, leading to different public perception of the

degree as well as the influencing factors of netizens’ percep-

tion. In order to find the best combination of tourism promo-

tion policy, by dividing tourism policy into four categories

and combining them, 15 kinds of policy combinations can be

acquired (economy, idle, feasible, safety, economy + safety,

economy + feasible, economy + idle, idle + safety, idle +

3226

submit your manuscript

www.dovepress.com

DovePress

Risk Management and Healthcare Policy 2020:13

Risk Management

and Healthcare

Policy downloaded

from https://www.dovepress.com/

by 113.23.110.30

on 11-Jun-2021

For personal use only.

lOMoARcPSD| 36066900

Dovepress

Chen et al

Table 6

Variables and Definitions

Table 6

(Continued).

Variables

Code

Definitions

Level of development of

COVID-19

According to the level of

response to major public health

emergencies, The first level

response was from January 24

to April 30, the second level

response was from April 30 to

June 12, the level 3 response

was from June 12 to July 30.

feasible, safety + feasible, economy + idle + feasible, econ-

omy + idle + feasible + safety). With regard to the policy

combinations, the binary logistic regression model is set up.

In the binary logistics regression model, how many indepen-

dent variables are introduced needs to be studied. If fewer

independent variables are introduced, the regression equation

will not be able to explain the changes of dependent variables

in an accurate manner, but it does not mean that more

independent variables are absolutely better. Therefore, it is

necessary to adopt some strategies to control the independent

variables by introducing regression equations. The Stepwise

Selection method is adopted here, that is, the introduction

threshold of P-value is tested according to the significance of

the set regression coefficient, independent variables are intro-

duced into the model one by one, then P-values of all coeffi-

cients in the model are recalculated, and variables are

screened according to the set elimination threshold. The

Stepwise Selection method includes forward selection

method and backward selection method. The forward selec-

tion method is relatively simple, but the biggest disadvantage

is that if there is multicollinearity, the final model may be

mixed with less important independent variables. The back-

ward selection method is more conservative in terms of

information. The backward selection method is chosen in

this article. Data processing was conducted in SPSS25.0

software, and the significance level of entering the model

was set at 0.05, and the significance level of removing or

retaining variables was also set at 0.05.

The main steps of regression analysis are as follows:

using LR Likelihood Method to select the independent

variables with a significant relationship between public

perception effect, conduct significance test for the variable

regression coefficient and model to get binary logistics

regression equation between each variable and dependent

Risk Management

and Healthcare

Policy downloaded

from https://www.dovepress.com/

by 113.23.110.30

on 11-Jun-2021

For personal use only.

Variables

Code

Definitions

Dependent variable

(Policy implementation

effect):

The division of policies

from the perspective of

netizens

Judge whether the classification

of the netizen’s perspective and

policy perspective is consistent,

if it’s 1, otherwise it’s 0 (If it is

consistent, it indicates that this

policy is effective and produces

better social effect).

The policy concerns the

total number of

Microblogs

The number of netizens’

comments on Microblog topics

under each policy category.

Timeliness of comments

posted

The Microblog topics involved

under each policy category, the

average difference between

netizens’ comment time and

Weibo release time.

Opinion tendency

ROST was used to score each

netizen’s comments

emotionally, a number less than

zero indicates a negative

emotion, A number greater

than zero is a positive emotion,

equals zero is neutral.

Thumb up number

The number of thumb ups each

netizen comments on the

policy from other netizens.

Comment number

The number of thumb ups each

netizen comments on the

policy from other netizens.

Netizens gender

Gender of netizens

participating in comments

under each policy category, it’s

1 for women and 0 for men.

Number of fans

The number of followers of

netizens participating in

comments under each policy

category.

Number of followers

Netizens participating in

comments under each policy

category, the number of

followers of other netizens on

eibo

The degree of active

The number of original

Microblogs posted by netizens

participating in comments

under each policy category.

(Continued)

lOMoARcPSD| 36066900

Chen et al Dovepress

Table 7

Significance Test of Independent Variables Selected by “Economic” Policies

Variable

Model Log Likelihood

Change in-2 Log Likelihood

Degrees of Freedom

Significance of the Change

Step1

−3031.611

437.468

0.000

−2812.878

0.003

0.960

−2813.892

2.031

0.154

−2814.068

2.383

0.123

−2813.835

1.916

0.166

−2813.162

0.571

0.450

−2814.975

4.197

0.041

−2813.964

2.175

0.140

−2848.757

71.761

0.000

Note:

Means backward stepwise selection regression method is adopted in regression analysis.

Table 8

The Variable Coefficients in the “Economic” Policy Equation

Standard Error

Wald

Degrees of Freedom

Significance

Exp(B)

AIC

Step7

0.028

0.002

310.135

0.000

1.028

0.000

4.090

0.043

1.000

0.309

0.037

70.078

0.000

1.361

Constant

−0.688

0.086

63.620

0.000

0.502

519.426

Step1

540.505

Note:

Means backward stepwise selection regression method is adopted in regression analysis.

variable, and test the prediction accuracy of the overall

model. The following is a detailed introduction to the

regression analysis of the public perception effect under

the “economy” tourism policy, and the regression analysis

process of the remaining 13 tourism policy combinations

can be similarly obtained.

The regression model construction process of the pub-

lic perception effect under the “economy” tourism policy

is illustrated as an example. Based on the comment data of

“economy” policy obtained above, SPSS was used for

binary logistics regression. After eliminating the indepen-

dent variable X

which could not be introduced, the inde-

pendent variable could be selected and the results were

shown in Table 7, the value of regression coefficient of

each variable was shown in Table 8, and the model regres-

sion statistical results were shown in Tables 9 and 10.

1. Generally speaking, the significance level is to estimate

the probability of wrong parameter within a certain

interval. When the original hypothesis is true and repre-

sented by α, the probability of rejecting the original

hypothesis is usually 0.05. When the decision to accept

the original hypothesis is made, the probability of its

correctness is 95%. It can be seen from Table 7 that in

the selection of independent variables, the significance

level Sig.<0.05 is X

for the timeliness of comment

release, the number of attention is X

, and the develop-

ment level of COVID-19 is X

. Therefore, there is

a significant relationship between the public perception

of “economy” policies and these three variables, so

Table 9

Omnibus Tests of Model Coefficients Under the Policy

of “Economic”

Chi-Square

Degrees of

Freedom

Significance

Step 7

Step

−2.673

0.102

Block

526.017

0.000

Model

526.017

0.000

Note:

Means backward stepwise selection regression method is adopted in

regression analysis.

3228

submit your manuscript

www.dovepress.com

DovePress

Risk Management and Healthcare Policy 2020:13

Risk Management

and Healthcare

Policy downloaded

from https://www.dovepress.com/

by 113.23.110.30

on 11-Jun-2021

For personal use only.

lOMoARcPSD| 36066900

Dovepress

Chen et al

Table 10

The Final Model of “Economic” Policy Predicts Rates

Observed

Predicted

Netizens

Have a Sense

of the Policy

Netizens Did

Not Perceive

the Policy

Percentage

Correct

Step

Netizen’s

perception of

policy: Y

1341

680

66.4

759

1693

69.0

Overall

percentage

67.8

Note:

Means backward stepwise selection regression method is adopted in

regression analysis.

independent variables of the binary logistics model

under such policies can be screened out.

2. According to the selected independent variable, the

coefficient of its independent variable is determined

by the value of Exp(B) in Table 4. Exp(B) is also

known as the preponderance ratio, which means

that the preponderance ratio is twice that of the

original Exp(B) when the other independent vari-

ables are fixed and unchanged.

3. The regression equation of public perception under

the “economy” policy can be obtained as follows:

Y ¼ 0:465 þ 1:029X

þ X

þ 1:38X

(2)

Due to the Sig. <0.05 in Table 9, it indicates that model is

significant under 95% significance level, so the X

, X

, and

can be used as factors that influence the public perception

of policies. According to the model prediction accuracy in

Table 10, the probability of the model in formula (2) that can

predict the public perception of policy is 67.8%.

Similarly, according to the above policy categories, the

basic four kinds of policies are combined. When policy

content contains one, two, three, or four kinds of information,

15 class policy combinations can be obtained. The construc-

tion of regression model under 15 combinations all meet the

test of significance level. The results are shown in Table 11.

Analyzing the Implementation Effect of

Tourism Policy

1) After the release of the 15 policy combinations, the

binary logistic regression model was established based on

public perception of policy. From Table 11, the

conclusions can be made that the release of “safety” policy

brings better social effect, whose social policy implemen-

tation effect is 93.2%. If the “safety policies + feasible

policies” and “economy policies + safety policies” can be

released together, people’s perceptions are 92.2%, 89.3%,

followed by the “economy + safety + feasible policies”,

which is 88.5%.

As can be seen from Table 5, the main topic of “safety”

policy on Weibo is “Travel around China Online”. By

means of online travel, people lower their travel frequency

while keeping their enthusiasm for travel consumption

after COVID-19. Moreover, during the epidemic period,

the safety of the tourism environment is the most con-

cerned and worried issue. The most fundamental problem

for the revitalization of the tourism industry is to control

the development of the domestic epidemic. After the

security is guaranteed, the “economy” and “feasible” poli-

cies will be released at the same time, which can promote

the revitalization of the tourism industry. As for the “econ-

omy” policy, it is mainly to issue tourism coupon, while

the “feasible” policy is to open major tourist attractions or

to resume inter-provincial travel. Only when the tourist

attractions start to operate normally, can people use tour-

ism coupon to recover the tourism industry through scenic

spot consumption.

2) The combinations of “economy + feasible”, “economy +

idle + feasible”, “economy + idle” bring poorer social effect

than other policy combinations. This is because if the govern-

ment does not release the information of epidemic develop-

ment situation nor the safety measures for travel, people are not

interested in traveling. Before safety is guaranteed, the effect of

policy is poor. Based on this, it is suggested that the govern-

ment should report the development of epidemic situation at

home and abroad in real time, as well as the relevant safety

measures taken in tourist attractions or the process of tourism

in the content of the revitalization of tourism policy, so as to

ensure more obvious social effects after the release of the

policy.

3) From Table 11, the conclusions can be made that the

netizens’ activeness X

greatly impacts on public’s perception

effect. In the top five optimized policy combinations, the

greater the X

of Internet users is, the stronger the perception

about policy is, the greater the effect of policy implementation

brings. The activeness of netizens is the original Weibo

released by netizens. Generally speaking, the more active

a netizen is on Weibo, the greater his personal influence will

be, and the stronger the effect of network public opinion will be

caused. For government departments, if they want to expand

Risk Management

and Healthcare

Policy downloaded

from https://www.dovepress.com/

by 113.23.110.30

on 11-Jun-2021

For personal use only.

lOMoARcPSD| 36066900

Chen et al Dovepress

Table 11

Results of Binary Logistic Regression Model for 15 Policy Portfolios

Ordinal

Number

of Policy

Portfolios

Ranking of

Different

Policy

Preferences

Different Policy

Mix Categories

Binary Logistic

Regression Model

Factors Influencing the Social Effect

of Policies

Public

Perception

of the Effects

of Policy

Safety policies

Y ¼ 4:881 þ 0:988X

Opinion tendency\Netizens gender\The

93.2%

0:789X

þ X

þ 0:03X

degree of active\Level of development of

AIC=2,869.762

COVID-19

Safety policies +

Feasible policies

Y ¼ 0:891 þ X

þ 0:996X

0:983X

0:857X

0:399X

The policy concerns the total number of

Microblogs\Timeliness of comments

posted\Opinion tendency\Netizens gender

92.2%

AIC=1,641.991

\Level of development of COVID-19

Economy policies

+ Safety policies

Y ¼ 1:114 þ X

þ 0:999X

0:747X

The policy concerns the total number of

Microblogs\Timeliness of comments

89.3%

AIC=4,768.212

posted\Netizens gender\The degree of

active

Economy policies

+ Safety policies +

Y ¼ 0:992 þ 0:995X

0:78X

0:788X

Opinion tendency\Netizens gender\The

degree of active\Level of development of

88.5%

AIC=229.69

Feasible policies

COVID-19

Idle policies +

Y ¼ 2:927 þ X

þ 0:998X

The policy concerns the total number of

85.5%

Safety policies

0:987X

0:946X

Microblogs\Timeliness of comments

AIC=2,148.01

þ X

þ 0:059X

þ 0:187X

posted\Opinion tendency\Netizens gender

\Number of followers\The degree of active

\Level of development of COVID-19

Economy policies

+ Idle policies +

Safety policies

Y ¼ 2:927 þ X

þ 0:998X

0:987X

0:946X

0:187X

85.5%

AIC=218.008

Safety policies +

Y ¼ 1:651 þ X

þ 0:996X

The policy concerns the total number of

85.2%

Idle policies +

Feasible policies

0:985X

0:997X

þ 0:951X

þ X

þ 0:354X

Microblogs\Timeliness of comments

posted\Opinion tendency\Thumb up

AIC=2,746.003

number\Comment

number\Netizens

gender\Number of followers\Level of

development of COVID-19

Economy policies

+ Safety policies +

Feasible policies +

Y ¼ 1:59 þ X

þ 0:999X

0:991X

0:892X

0:627X

The policy concerns the total number of

Microblogs\Timeliness of comments

posted\Opinion tendency\Netizens gender

82.8%

AIC=4,731.289

Idle policies

\Number of followers\The degree of active

\Level of development of COVID-19

Feasible policies

Y ¼ 0:975X

þ 0:972X

The policy concerns the total number of

79.8%

1:363X

Microblogs\Opinion

tendency\Netizens

AIC=60.456

gender

Economy policies

Y ¼ 0:465 þ 1:029X

Timeliness of comments posted\Number

67.8%

1:38X

of followers\Level of development of

AIC=519.426

COVID-19

Idle policies +

Feasible policies

Y ¼ 0:214 þ X

þ 0:986X

1:277X

0:93X

The policy concerns the total number of

Microblogs\Opinion tendency\Netizens

66.9%

AIC=514.627

gender\Number of followers\Level of

development of COVID-19

(Continued)

3230

submit your manuscript

www.dovepress.com

DovePress

Risk Management and Healthcare Policy 2020:13

Risk Management

and Healthcare

Policy downloaded

from https://www.dovepress.com/

by 113.23.110.30

on 11-Jun-2021

For personal use only.

lOMoARcPSD| 36066900

Dovepress

Chen et al

Table 11

(Continued).

Ordinal

Number

of Policy

Portfolios

Ranking of

Different

Policy

Preferences

Different Policy

Mix Categories

Binary Logistic

Regression Model

Factors Influencing the Social Effect

of Policies

Public

Perception

of the Effects

of Policy

Idle policies

Y ¼ 0:53 þ 0:99X

1:261X

Opinion tendency\Netizens gender\Number

of followers\The degree of active

63.2%

AIC=91.277

Economy policies

+ Feasible policies

The implementation effect of the three policy combinations is not as good

as that of the single policy release (This regression model is meaningless,

AIC value is meaningless)

63.4%

Economy policies

+ Idle policies +

Feasible policies

63.2%

Economy policies

+ Idle policies

59.4%

the social effects brought by the policies, they can forward the

policies through “opinion leaders” on Weibo to let more people

know the contents of the policies, promote tourism consump-

tion, and drive the tourism economy.

4) From Table 11, except for the “idle + feasible”, “econ-

omy”, “economy + safety” policy, opinion tendency degree X

impacts on the effect of different policy combination. X

regression equation coefficient shows that each type of policy

combination regression coefficients is above 0.9, indicating

that X

had a greater influence on the effect of the policy

implementation. This is because after the release of policy,

netizens hold positive, negative, or neutral attitudes, which

affects the tendency of public opinion. If the policy is released,

“opinion leaders” hold negative attitudes, which will not be

conducive to policy implementation and cause rejection on the

network platform. Therefore, after the release of policy, gov-

ernment should control trend of public opinion, timely stop

bad happens so as to play the positive impact of policies.

Conclusions and Suggestions

This paper selects Weibo comments, comments time, gender,

the original Weibo number, and other aspects from January 11,

2020 to July 24 on the revitalization of the tourism policy, and

divides the revitalization of tourism policy into “economy”,

“safety”, “idle”, “feasible” four major categories. From the

perspective of users, by constructing binary logistic regression

model, the combination of all kinds of policy is analyzed.

Based on the above analysis, this paper provides the following

suggestions for the government and scenic spots to cope with

public health emergencies and improve the perceived effect of

policies:

Tourism Policies Under the Epidemic

Situation Should Highlight Safety

Measures

The epidemic brings great challenge to China’s governance

system and capacity. For the tourism industry, it is neces-

sary to carry out the corresponding assessment, prevention,

and treatment, and handle the tourism crisis properly, which

not only requires scientific decision-making and precise

measures by the government, but also requires the joint

efforts of scenic spot practitioners. According to the analy-

sis of this paper, the public’s perception of “safety” policies

is the strongest. Only when tourists perceive that it is safe to

travel can the cultural and tourism industry gradually

resume work and production. Based on this, the government

should timely issue a “safety” policy after the initial stabi-

lity of the epidemic. For example, on February 25, the

Ministry of Culture and Tourism issued a “Guide to

Prevention and Control Measures for Reopening of Tourist

Attractions”, which enables tourists to perceive that the

government encourages the cultural and tourism industry

to resume work and production, and it is relatively safe to

travel in the current environment. In addition, the govern-

ment can issue a series of advocacy policies, such as advo-

cating industry associations to strengthen the safety and

supervision of epidemic prevention and control, and guiding

scenic spots to actively participate in relevant work;

Secondly, the scenic spot should issue a series of effective

and feasible prevention and control policies to ensure safety,

such as implementing a series of policies in terms of limit-

ing the capacity of tourists, keeping social distance between

Risk Management

and Healthcare

Policy downloaded

from https://www.dovepress.com/

by 113.23.110.30

on 11-Jun-2021

For personal use only.

 
lOMoARcPSD| 36066900 
 
 
Chen et al  Dovepress 
 
tourists, regularly disinfecting all scenic spots, and requiring 
tourists to wear face masks. 
 
The Combination of Economic Policies 
and Security Policies Can Achieve Better 
Results 
The  most  fundamental  problem  of  reviving  the  tourism 
industry  is  to  control  the  development  of  the  epidemic. 
When safety is guaranteed, “economy” and “safety” poli- 
cies  will  be  issued  at  the  same  time,  which  will  greatly 
accelerate  the recovery of  the tourism industry.  The rea- 
lity  of  the  difficulties  in  view  of  the  current  tourism 
industry development, in addition to the policy of issuing 
coupons  to attract  tourists, governments at  all levels can 
introduce  more  perfect  tourism  industry  revitalization 
policy,  especially  the  release  and  enforcement  of  fiscal 
policy, tax policy, credit policy, and social security policy 
for  the  troubled  tourism-related  businesses.  Providing 
financial subsidies for tourism services and related enter- 
prises  to  resume  operation  and  production.  For  tourism 
enterprises  that  have  special  difficulties  and  fail  to  pay 
tax  on  time,  tax  payment  shall  be  reduced  or  postponed 
appropriately. Tourist attractions can also release a series 
of  “economy”  policies  to  attract  tourists,  such  as  free 
tickets  or  appropriate  discounts,  multi-scenic  joint  ticket 
discounts  and  other  marketing  policies.  In  addition,  “fea- 
sible”  policies  will  be  issued  at  the  same  time,  such  as 
enforcing  the  paid  leave  system  for  employees  and  2.5 
days off policy. 
 
Play the Role of “Opinion Leaders” to 
Improve the Perceived Effect of Policies
 
Through analysis, this paper finds that “opinion leaders” have 
a significant impact on the perceived effect of the public after 
the release of policies. The greater the influence of “opinion 
leaders” is, the stronger the perception of policies will be, and 
the greater the social effect brought by the implementation of 
policies will be. Therefore, the government can advocate 
“opinion  leaders”  on  Weibo  to  forward  and  expand  the 
exposure of the policy, so as to improve the perceived effect 
of the policy, let more people know the content of the policy, 
boost tourism consumption and drive the tourism economy. 
 
Use Big Data of Tourism to Improve the 
Effect of Policies 
Since the outbreak of the COVID-19, big data of tourism 
has played an important role. However, in the face of the 
sudden  outbreak,  the  release  of  policy  information  and 
monitoring of public perception is still lagging behind. In 
this regard, local governments and scenic spots can use the 
big  data platform  to  accurately locate the spread path  of 
the epidemic, quickly track the flow of tourists and their 
movements, and establish a tourist relationship map, so as 
to provide data protection for “safety” policies and reduce 
tourists’ concerns about safety. In addition, big data can be 
used  to  speed  up  the  connectivity  of  all  kinds  of  policy 
information and expand the exposure of policies, so as to 
increase the public’s perception of the effect of policies. 
 
Funding 
This research is supported by the National Social Science 
Foundation of China (Grant No. 20BTQ059), Hubei Key 
Laboratory of Mechanical Transmission and Manufacturing 
Engineering  (MECOF2020B04),  Contemporary  Business 
and  Trade Research Center  and Center  for Collaborative 
Innovation  Studies  of  Modern  Business  of  Zhejiang 
Gongshang  University  of  China  (Grant  No. 
14SMXY05YB),  as  well  as  First  Class  Discipline  of 
Zhejiang-A (Zhejiang Gongshang University- Statistics). 
 
Disclosure 
The authors declare that they have no conflicts of interest 
for this work. 
 
References 
1. Dave K,  Steve L,  Pennock DM. Mining the peanut gallery: opinion 
extraction and semantic classification of product reviews. Proceedings 
of the 12th International Conference on World Wide Web. 2003:519–
528. 
2. Rossouw R,  Saayman M.  Assimilation of  tourism  satellite  accounts 
and  applied  general  equilibrium  models  to  inform  tourism  policy 
analysis. Tour Econ. 2011;17(4):753–783. doi:10.5367/te.2011.0058 
3. Miller S. Causal forest estimation of heterogeneous and time-varying 
environmental  policy  effects.  J  Environ  Econ  Manage. 
2020;103:102337. doi:10.1016/j.jeem.2020.102337 
4. Dong F, Liu Y. Policy evolution and effect evaluation of new-energy 
vehicle  industry  in  China[J].  Resour  Polic.  2020;67:101655. 
doi:10.1016/j.resourpol.2020.101655 
5. Matousek R, Panopoulou E, Papachristopoulou A. Policy uncertainty 
and  the  capital  shortfall  of  global  financial  firms.  J  Corp  Finance. 
2020;62:101558. doi:10.1016/j.jcorpfin.2020.101558 
6. Livni J.  Investigation of collapse of complex socio-political systems 
using  classical  stability  theory.  Physica  A.  2019;524:553–562. 
doi:10.1016/j.physa.2019.04.167 
7. Chen T, Shi J, Yang J, Cong G, Li G. Modeling public opinion polariza- 
tion in group behavior by integrating SIRS-based information diffusion 
process. Complexity. 2020;2020:1–20. doi:10.1155/2020/4791527 
8. Chen T, Wang Y, Yang J, Cong G. Modeling public opinion reversal 
process  with  the  considerations  of  external  intervention  information 
and  individual  internal  characteristics.  Healthcare.  2020;8(2):160. 
doi:10.3390/healthcare8020160 
 
 
 
3232 
submit your manuscript 
| 
www.dovepress.com
 
DovePress
 
Risk Management and Healthcare Policy 2020:13
 
Risk Management
 and Healthcare
 Policy downloaded
 from https://www.dovepress.com/
 by 113.23.110.30
 on 11-Jun-2021 
For personal use only.
 

 
lOMoARcPSD| 36066900 
 
 
Dovepress 
Chen et al
 
 
9. Alexander A, Yusaku H. Has trump damaged the U.S. image abroad? 
Decomposing  the  effects  of  policy  messages  on  foreign  public 
opinion. Polit Behav. 2020;42(3):581–602. doi:10.1007/s11109-018- 
9511-3 
10. Ho M. An analysis on the critical success factors of healing tourism 
policy by AHP. J Humanit Soc Sci. 2019;10(1):1473–1486. 
11. Yin D. Citizen network participation: an analytical framework of the 
new  political  system  theory.  J  Party  School  Tianjin  Municipal 
Committee Communist Party China. 2010;2:44–52. 
12. Rothwell R, Zegveld W. Reindusdalization and Technology. Logman 
Group Limited; 1985:83–104. 
13. Kim  A,  Jeong  SH,  Choi  HB,  Kim  HH.  An  analysis  of  effects  of 
emergency fine dust reduction measures and national petition using 
regression analysis and text mining. KIPS Trans Software Data Eng. 
2018;7(11):427–434. 
14. Gunhee L, Sutherland I, Park S. A study of factors affecting tourists 
to  Japan  using  text  mining  approach.  J  Tour  Leis  Res.  2017;29 
(11):181–199. 
15. Chen T, Peng L, Yin X, Rong J, Yang J, Cong G. Analysis of user 
satisfaction  with  online  education  platforms  in  China  during  the 
COVID-19  pandemic.  Healthcare.  2020;8(3):200.  doi:10.3390/ 
healthcare8030200 
16. Chen T, Peng L, Jing B, Wu C, Yang J, Cong G. The Impact of the 
COVID-19 pandemic on user experience with online education plat- 
forms  in  China.  Sustainability.  2020;12(18):7329.  doi:10.3390/ 
su12187329 
17. Bucek J. Fiscal policy for the 21st century: does Barack Obama effect 
the  real  economic  policy?  34th  International  Conference 
Mathematical Methods in Economics (MME 2016). 2016:97–102. 
18. Qi Y, Peng W, Xiong N. The effects of fiscal and tax incentives on 
regional  innovation  capability:  text  extraction  based  on  python. 
Mathematics. 2020;8(7):1193. doi:10.3390/math8071193 
19. Sun PJ.  XACML policy evaluation optimization  research  based  on 
attribute  weighted  clustering  and  statistics  reordering.  2017  IEEE 
International  Conference  on  Information  and  Automation  (IEEE 
ICIA 2017). 2017:1190–1195. 
20. Pellesova P. Selected econometric methods of optimization of eco- 
nomic  policy.  Proceedings  of  30th  International  Conference 
Mathematical Methods in Economics, PTS I AND II. 2012:692–696. 
21. Dash  GH,  Kajiji  N.  On  multiobjective  combinatorial  optimization 
and dynamic interim hedging of efficient portfolios. Int Trans Oper 
Res. 2014;21(6):899–918. doi:10.1111/itor.12067 
22. Yang  G,  Zhang  Y,  Li  X.  Impact  of  gasoline  upgrade  policy  on 
particulate  matter  pollution  in  China.  J  Clean  Prod. 
2020;262:121336. doi:10.1016/j.jclepro.2020.121336 
23. Geng D, Saggi K. Optimal price regulations in international pharma- 
ceutical  markets  with  generic  competition.  J  Health  Econ. 
2020;71:102315. doi:10.1016/j.jhealeco.2020.102315 
24. Gupta M, Bolia NB. Efficiency measurement of Indian high courts 
using DEA: a policy perspective. J Policy Model. 2020. 
25. Andrew Kelly J, Peter Clinch J, Kelleher L, Shahab S. Enabling a just 
transition: a composite indicator for assessing home-heating energy-
poverty  risk  and  the  impact  of  environmental  policy  measures. 
Energy Policy. 2020;146:111791. doi:10.1016/j. enpol.2020.111791 
26. Groh ED, Ziegler A. On self-interested preferences for burden shar- 
ing  rules:  an  econometric  analysis  for  the  costs  of  energy  policy 
measures.  Energy  Econ.  2018;74:417–426.  doi:10.1016/j. 
eneco.2018.06.026 
27. Valerio C, De Stefano L, Martínez-Muñoz G, Garrido A. A machine 
learning  model  to  assess  the  ecosystem  response  to  water  policy 
measures  in  the  Tagus  River  Basin  (Spain).  Sci  Total  Environ. 
2021;750:141252. doi:10.1016/j.scitotenv.2020.141252 
28. Dukhanina  E,  Massol  O,  Lévêque  F.  Policy  measures  targeting 
a  more  integrated  gas  market:  impact  of  a  merger  of  two  trading 
zones  on  prices  and  arbitrage  activity  in  France.  Energy  Policy. 
2019;132:583–593. doi:10.1016/j.enpol.2019.05.044 
29. Dong F, Liu Y. Policy evolution and effect evaluation of new-energy 
vehicle industry in China. Resour Polic. 2020;67:1–13. 
30. Smith JD,  Hasan M.  Quantitative approaches for  the evaluation  of 
implementation  research studies.  Psychiatry  Res.  2019;283:112521. 
doi:10.1016/j.psychres.2019.112521 
31. Xu Y, You X, Wong Y. Evaluation of tourist satisfaction based on 
ROSTCM method: a case study of five domestic terrace scenic spot. 
Tour Forum. 2018;11(5):22–34. 
32. Wang J, Guo Z. Logistic regression model: method and application. 
Beijing: Higher Education Press; 2001:13–26. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Risk Management and Healthcare Policy 
Dovepress
 
Publish your work in this journal 
Risk Management and  Healthcare  Policy is an  international, peer- 
reviewed,  open  access  journal  focusing  on  all  aspects  of  public 
health,  policy,  and  preventative  measures  to  promote  good  health 
and improve morbidity and mortality in the population. The journal 
welcomes  submitted  papers  covering  original  research,  basic 
science, clinical & epidemiological studies, reviews and evaluations, 
guidelines,  expert  opinion  and  commentary,  case  reports  and 
extended reports. The manuscript management system is completely 
online and includes a very quick and fair peer-review system, which 
is all easy to use. Visit http://www.dovepress.com/testimonials.php 
to read real quotes from published authors. 
 
Submit  your  manuscript  here: 
https://www.dovepress.com/risk-management-and-healthcare-policy-journal
 
Risk Management
 and Healthcare
 Policy downloaded
 from https://www.dovepress.com/
 by 113.23.110.30
 on 11-Jun-2021 
For personal use only.
 

Bấm Tải xuống để xem toàn bộ.

| 1/24

| | Tải xuống

Preview text:

lOMoAR cPSD| 36066900 lOMoAR cPSD| 36066900
Risk Management and Healthcare Policy Dovepress
open access to scientific and medical research Open Access Full Text Article
O R I G I N A L R E S E A R C H
A Policy Category Analysis Model for Tourism
Promotion in China During the COVID-19
Pandemic Based on Data Mining and Binary Regression
This article was published in the following Dove Press journal:
Risk Management and Healthcare Policy Tinggui Chen1 Email cgd@mail.zjgsu.edu.cn Lijuan Peng1 Xiaohua Yin1 Bailu Jing2 Jianjun Yang 3 Guodong Cong4 Gongfa Li5
1School of Statistics and Mathematics,
Zhejiang Gongshang University,
Hangzhou 310018, People’s Republic of
China; 2School of Management and
E-Business, Zhejiang Gongshang
University, Hangzhou 310018, People’s
Republic of China; 3Department of
Computer Science and Information
Systems, University of North Georgia,
Oakwood, GA 30566, USA; 4School of
Tourism and Urban-Rural Planning,
Zhejiang Gongshang University,
Hangzhou 310018, People’s Republic of
China; 5Hubei Key Laboratory of Mechanical Transmission and
Manufacturing Engineering, Wuhan
University of Science and Technology,
Wuhan 430081, People’s Republic of China Correspondence: Guodong Cong lOMoAR cPSD| 36066900 Background and Aim:
comment data of Weibo users, analyzes users’ perception and emotional preference to the At the
policy, and thus mines the social effect of various policies. Subsequently, by identifying the end of 2019, the outbreak of
social effects of various policies as dependent variables, a binary logistic regression model is COVID-19 had a significant impact on China’s tourism
constructed to obtain the best combination of tourism promotion policies and promote the
rapid revitalization of the cultural and tourism industry.
industry, which was almost at a
Results: The results show that from the single policy, the social effect of the “safety” policy
standstill in the short-term. After
is the best. From the perspective of combination policies, the simultaneous release of
reaching the preliminarily stable
“safety” policies and “economy” policies have the greatest social impact, which can drama- state, the government and the 021 2
tically accelerate the recovery of the cultural and tourism industry. Finally, this paper - scenic area management un
proposes suggestions for policy formulation to improve the ability of the cultural tourism J-
department implemented a series 11
industry to cope with crisis events.
of incentive policies in order to on
Conclusion: These results explain the perceived effects of the public on the government 0 speed up the recovery of the 3
policies and can be used to judge whether the policies have been released in place. Based on 0. tourism industry. Therefore, 11.
the above results, corresponding suggestions are proposed as follows: 1) the combination of
analyzing all sorts of social effects 23.
economic policies and security policies can achieve better results; and 2) the role of “opinion
after policy implementa- tion is of 113
leaders” can be played to improve the perceived effect of policies. y guiding significance for the b /
Keywords: online comments, social effects, combination optimization, data mining, binary
government and the scenic areas. om c.s Methods: logistic regression, COVID-19 Targeted as the social s re
effect with the implementation of ep ov
tourism promotion policy during d. . y Introduction w the COVID-19 pandemic, this w onl
With the progress of science and technology, as well as the improvement of w/ e
paper briefly analyzes the impact /: s s u people’s pt of COVID-19 on the national
living conditions and material level, China’s tourism industry has entered ht onal
cultural and tourism industry and
a new take-off stage after more than 20 years of development. However, the rs romf pe
selects several representative types
outbreak of COVID-19 in late 2019 greatly impacted on China’s tourism industry, or ded F
of tourism policies, crawls the nloa ow
submit your manuscript | www.dovepress.com
Risk Management and Healthcare Policy 2020:13 3211–3233 d 3211 y DovePress
© 2020 Chen et al. This work is published and licensed by Dove Medical Press Limited. The ful terms of this license are available at https://www.dovepress.com/terms.php
and incorporate the Creative Commons Attribution – Non Commercial (unported, v3.0) License (http://creativecommons.org/licenses/by-nc/3.0/). By accessing the work
http://doi.org/10.2147/RMHP.S284564 olic
you hereby accept the Terms. Non-commercial uses of the work are permitted without any further permission from Dove Medical Press Limited, provided the work is properly attributed. For P
permission for commercial use of this work, please see paragraphs 4.2 and 5 of our Terms (https://www.dovepress.com/terms.php). are hc ealt H and nt e nagem a M kisR lOMoAR cPSD| 36066900 Chen et al Dovepress
wiping out more than 450 million tourist revenues. About
and making other factors (Weibo comments, comments
2 months after the enforcement of prevention and control,
time, like quantity) as independent variables. In turn, it
on February 25, the national culture and tourism office
combines the results of the single policy and mixed policy
issued the guidance to lead the national scenic spots to
with the most promoting effect for tourism, providing
implement the epidemic prevention and control to make
suggestions for the government and the scenic spot to
recovery steadily. Meanwhile, local governments and sce-
accelerate the revitalization of the culture and tourism
nic spots have issued a series of promoting policies, such industry. 021 2-
as travel coupons, current limiting, 2.5-days off, which
The structure is as follows: Literature Review analyzes un J-
aims to stimulate recovery of the tourism industry while
the literature of tourism policy; Analysis on Supporting 11 on
preventing COVID-19. However, the policies have
Policies of Cultural Tourism Industry During COVID-19 0 3
received mixed reviews from netizens, who have focused
discusses the tourism support policies; Data Mining Based 0. 11.
on the safety issues brought about by the opening of scenic
on Online Comments of Travel Policies During COVID- 23.
spots. For the government and scenic spots, the most
19 conducts data mining on the online comments on tour- 113 y
concerning question is which kind of policies have the
ism policies; The Social Effect Analysis of Tourism b /
most obvious stimulating effect on the tourism economy.
Policies Based on Binary Logistic Regression Model con- om c.s
In order to answer these questions, it is necessary to
structs a binary logistic regression model to measure the s re
conduct an emotional analysis of a series of policies issued
social effects of tourism revitalization policies and seek for ep ov
by the government and scenic spots, measure the effect of
the optimal combination of policies; Conclusions and d. . y w
policy implementation according to online netizens’ com-
Suggestions summarizes the whole paper and provides w onl w // e : s
ments, and summarize the policy combination with the policy suggestions. s u p t
best implementation effect, so as to help the government
ht onalrs and scenic spots maximize the validity to promote Literature Review romf pe recovery.
At present, there are text mining, emotion analysis, and or ded F
At present, domestic and foreign scholars have carried
combinatorial optimization methods for policy implemen- nloa
out some researches on the implementation effect of tour-
tation effect analysis in both domestic and international ow d y
ism policies, mainly including drawing on multi-
research. These methods are still practiced in the study of olic P
disciplinary research methods such as consumer behavior,
tourism policy implementation effect. Therefore, this are
public management, and journalism and communication,
paper draws on the methods of evaluating the implementa- hc
designing research scales and measurement processes, and
tion effect of other policies to analyze the research status ealt H
conducting research on the impact of tourism policies
of a tourism policy implementation effect. and
based on tourism consumption behavior intention.
Tourism policies mainly refer to policies that promote nt e
However, few scholars directly use online comment data
tourism activities in various aspects. In order to understand
of tourism public opinion for research, but whether to
the impact of these policies on the actual tourism industry, nagem a
travel is easily affected by public opinion. Therefore, it
scholars have conducted in-depth studies with policy mea- M kis
is of great significance to carry out text mining on tourists’
surement tools, and the representative results are as fol- R
online comment data on tourism policy, so as to analyze
lows: David et al1 introduces the research method of
the social effect of policy implementation.
system theory into the political field, and puts forward
Based on this, this paper evaluates the effects of
the “political system theory” for the first time. It holds
a series of policies issued by the government and scenic
that the interaction between policy system and environ-
spots against the background of COVID-19. In particular,
ment is realized through “input, transformation and out-
this paper analyzes the present situation and the existing
put” in politics through “requirements and support”
supporting policy to select the representative tourism poli-
existing in the environment. Rossouw and Saayman2
cies released by government, then reviews Weibo com-
demonstrated the relevance and necessity of using the
ments data to understand users’ intention and emotion
tourism satellite account (TSAs) as a tool for South
preference, as well as the social effect of all kinds of
African decision-makers (especially tourism decision-
policy implementation. Besides, this article builds
makers) to improve and expand the application of the
a binary logistic regression model, making social effects
general equilibrium (AGE) model. The reasons for the
influenced by various policies as the dependent variable,
need for economic models for policy analysis and other
3212 submit your manuscript | www.dovepress.com
Risk Management and Healthcare Policy 2020:13 DovePress lOMoAR cPSD| 36066900 Dovepress Chen et al
purposes were expounded, and the new and old methods
Environmental policy tools, Supply-oriented policy tools,
for tourism policy modeling were summarized. Steve3
and Demand-oriented policy tools.
extended and applied a non-parametric method to estimate
In spite of various studies on tourism policy, even
the effect of heterogeneous treatment, and examined how
though some scholars explore the effects of tourism policy
the policy effect changed over time. The study showed that
implementation, few scholars study the optimal combina-
this method had potential value in evaluating the impact of
tion of tourism policy, namely through the data mining of
a series of environmental policies and environmental
online reviews, research released by any combination of 021 2-
impacts. Dong and Liu4 used a threshold model and quan-
tourism policy content can bring about a better social un J-
tile regression model to explore the threshold effect of
effect. In addition, as far as the social effects of tourism 11 on
policy power on policy implementation effect and the
policies are concerned, there is still a lack of objective 0 3
transformation of policy implementation effect in different
analysis based on network text data, so the research meth- 0. 11.
development stages. The results showed that the effect of
ods need to be innovated. In view of this, this paper takes 23.
industrial policy varied greatly in different development
the Weibo comments as the data source to find out the 113 y
stages, and policy power had a significant threshold effect
optimal combination that can make tourism promotion b /
on its operation intensity. Matousek et al5 studied how the
policies exert the greatest social effect, so as to provide om c.s
uncertainty of economic policy affected the capital short-
suggestions on the content form of policies issued by s re
age of financial companies in the new crisis of the event. government departments. ep ov
The study found that if the response of policy-makers and
In order to understand the actual effect of policy imple- d. . y w
politicians was not timely and decisive during the severe
mentation, scholars have conducted relevant studies on the w onl w // e : s
market downturn, then there would be a price for the
evaluation methods of the policy implementation effect. s u p t
delay. Joseph6 used the classical stability theory to model
Kim et al13 studied the “free public transport” policy and ht onalrs
the complex social and political system, and the causes of
the “citizen participation alternative day no driving” sys- romf pe
social collapse were studied. Chen et al7,8 analyzed the
tem implemented by Seoul government in order to reduce or ded F
polarization of public opinion in group behavior based on
dust. They adopted a regression analysis method to ana- nloa
the SIR model, and at the same time considered the influ-
lyze the impact of traffic on fine dust, and used text mining ow d y
ence of external information and individual internal char-
technology to analyze the response of two traffic policies olic P
acteristics on the polarization of public opinion. Alexander
and citizens’ petition, and proposed policy improvement are
and Yusaku9 took Japan as an example, and carried out
direction according to the research results. Lee et al14 used hc
a survey experiment to let citizens understand the policy
text mining technology and emotional text analysis tech- ealt H
information of the US, which varied according to the
nology to test the online evaluation of Japanese tourism and
source, policy content, and problem prominence. The
websites after the 2011 Japanese tsunami, and studied the nt e
results showed that when the source signal (Trump attribu-
impact of the tsunami on Japan’s tourism industry. The
tion) led to negative views in the US, policy content
results show that the low exchange rate and the positive nagem a
(cooperative than uncooperative) had a greater impact on
sentiment of online reviews from tourism websites have M kis
shaping the opinions of American citizens. Ho10 analyzed
a positive impact on the number of tourists to Japan. Chen R
the success factors of a convalescent tourism policy, and
et al15,16 used text mining and emotion analysis methods to
discussed the priority and importance of success factors of
implement the effect of online education under the epi-
the recuperation tourism policy by using an AHP analysis
demic situation. Also, the experience effect was studied
method. Yin11 redefined the theory of the political system
from the perspective of user and platform. Bucek17 used
based on the agenda setting theory in communication
text mining technology to investigate the twitter accounts
science, and constructed a “network public opinion deci-
of US President Barack Obama from March 2012 to
sion hypothesis model” to analyze how and to what extent
January 2016, so as to study whether politicians’ behavior
network public opinion affects government decision-
on social networks would affect actual economic policy.
making. In addition, as there are many types of policies,
Qi et al18 analyzed the innovative fiscal policy texts of
some scholars have classified the policy tools, among
provinces and cities in China based on python, so as to
which Rothwell and Zegvelk12 were the most representa-
understand the regulation and control of tax policies. The
tive ones. When studying industrial innovation and public
results declared that fiscal policy and regional innovation
policy, they divided the policy tools into three types:
capability showed obvious spatial heterogeneity, and R&D lOMoAR cPSD| 36066900 Chen et al Dovepress
investment and industrial structure were the main sources
had by far the highest support. Carlotta et al27 used
of improving innovation capability. Sun19 proposed an
Random Forest and Gradient Boosted Regression Trees
evaluation scheme to optimize the policy process by
algorithms to predict the response of freshwater ecosys-
using sorting and clustering strategies, and proposed
tems to multiple anthropogenic pressures, with the goal of
a combination algorithm combining the algorithm with
informing the definition of water policy targets and man-
policy priority evaluation. Finally, satisfactory policies
agement measures to recover and protect aquatic biodiver-
and rules were selected to improve the matching speed.
sity. Ekaterina et al28 confirmed a positive impact of the 021 2-
The experimental results showed that the method reduced
zone merger on the gas trading market’s spatial equili- un J-
the matching operation and improved the evaluation effi-
brium and indicated the causes of remaining market inef- 11 on
ciency. Pellesova20 discussed the technology of using an
ficiencies used by an extended parity bounds model, which 0 3
econometric method to optimize economic policy, mainly
provided a tool for evaluating the efficiency of policy 0. 11.
studied the target variable method, and explored the
decision-making. Dong and Liu29 analyzed the optimal 23.
advantages and disadvantages of the selected method and
functioning power of policies and determined the direction 113 y
its possible application in the formulation of optimal eco-
of future policy implementation. This paper utilized the b /
nomic policy. Dash and Kajiji21 proposed a mixed integer
COPA framework to analyze policy evolution in respect of om c.s
nonlinear objective program (MINLGP), which aimed to
the new-energy vehicle industry (NEVI). Smith and s re
solve the model of multi-objective portfolio optimization
Hasan30 discussed the methods and practices involved in ep ov
decision-makers facing binary hedging decision-making
quantitative evaluations of implementation research stu- d. . y w
between Portfolio Rebalancing periods. It was found that
dies, and analyzed available measurement methods for w onl w // e : s
when percentile risk measurement was used, the expected
common quantitative implementation outcomes involved s u p t
catastrophic loss of the best diversified portfolio of hed-
in such an evaluation-adoption, fidelity, implementation
ht onalrs ging was obviously less than that of non hedging products. cost, reach, and sustainment, and the sources of such romf pe
Yang et al22 evaluated the impact of the recently imple-
data for these metrics using established taxonomies and or ded F
mented policy of improving gasoline quality on reducing frameworks. nloa
the concentration of fine particulate matter (PM). The
According to the above literature, many scholars have ow d y
study illustrated that it was difficult to completely solve
studied public policies and built many models. However, olic P
the particulate pollution problem in China by a single
due to the epidemic, the formulation of tourism policies are
policy, and a series of policy system designs were needed
should also take the characteristics of emergencies, health, hc
to alleviate this problem. Geng and Kamal23 analyzed the
and safety into account. At present, there are few studies ealt H
optimal policy options for two main types of price regula-
on this aspect in the literature. Therefore, this paper takes and
tion in China, and found that the domestic optimal external
the implementation of the promotion policy as the back- nt e
reference pricing (ERP) policy reduced domestic prices
ground, classifies policy, analyzes online reviews, uses the
while maintaining export incentives for enterprises.
binary regression model to evaluate the implementation nagem a
Maansi and Nomesh24 measured the efficiency of Indian M
effect under a different combination of policies, and finally k is
high courts using Data Envelopment Analysis (DEA).
selects policy combination ways with better effect, which R
Secondly, they studied the impact of including pending
provide a decision basis and support for the release of the
cases on judicial efficiency. Kelly et al25 proposed government’s policy.
a composite index to assess home-heating energy-poverty
risk across 18,641 small area clusters in Ireland. The index
Analysis on Supporting Policies of
offered the capacity to analyze changes in energy-poverty
the Cultural Tourism Industry
risk associated with specific policy intervention proposals,
including major contemporary environmental policy tran- During COVID-19
sitions. Elke and Andreas26 examined the acceptance of
The status of the cultural and tourism industry during
burden sharing rules by using multivariate binary and
COVID-19 needs to be analyzed before analyzing the
ordered Probit models. The rule involved the costs of the
implementation effect of the tourism revitalization policy
German energy transition, which was one of the most
issued by the government. In light of the update of the
challenging and disputed national climate and energy pol-
epidemic, when should the government issue policies?
icy measures. The results declared that polluter-pays rule
What policies have been issued? What topics have been
3214 submit your manuscript | www.dovepress.com
Risk Management and Healthcare Policy 2020:13 DovePress lOMoAR cPSD| 36066900 Dovepress Chen et al
discussed on Weibo? Which media does report the policy
and began to plummet a day later, plummeting to 17% on
heavily? What kind of publicity effect did it have? Based January 26.
on this, the current situation of tourism is elaborated
As can be seen from the above brief description of the
below, and the current implementation of decentralized
current situation, with the sudden outbreak of COVID-19,
tourism revitalization policies are summarized. More
the Chinese government responded quickly and took var-
topics related to tourism policies discussed by the public
ious measures to block the transmission channels of
are selected through the Microblog platform, and the
COVID-19. As it turned out, China suddenly went into 021 2-
online comment data of the public are retrieved, so as to
a “dormant” state. However, the economic development, un J-
prepare for the subsequent in-depth analysis of the social
especially the development of the tertiary industry, has 11 on
effects brought by the implementation of policies.
encountered regression this year. The cultural and tourism 0 3 0.
industries, which are characterized by crowd gathering, 11. 23
A Brief Analysis of the Current Situation bear the brunt of the contraction. How much impact will . 113
of the Cultural Tourism Industry in China this “disaster” have on economic development? Can the y
cultural tourism industry survive? Is tourism a fragile b /
The outbreak of COVID-19 at the end of 2019 spread
industry? Experts, scholars, and ordinary people are con- om c.
rapidly across the country and seriously affected China’s s cerned about these issues. s re
economic and social development and people’s livelihood. ep
In particular, the impact of the epidemic on the tourism ov d. . y
Supporting Policies of the Cultural and w
industry includes both direct losses of many tourism enter- w onl w / e
prises and related employees and indirect losses of related /
Tourism Industry During COVID-19 : s s u p t
industries in the tourism industry. Just consider the Spring
The cultural and tourism industry is a modern service ht onalrs
Festival, the direct economic loss caused by the shutdown
industry with human service targeted with human services. romf pe
of China’s tourism industry is as high as 400 to 500 billion
Its basic feature is the movement of people, and the pursuit or ded F
Yuan, resulting in the annual expectation to change from
of security is the primary condition for people’s needs. By nloa
a “year-on-year growth of about 10% to a negative growth
May 2020, the epidemic prevention and control situation ow d y
of 14% to about 18%. On January 24, the General Office
in China has been stable, laying the foundation and creat- olic
of the Ministry of Culture and Tourism of China issued an
ing the basic conditions for people to travel safely during P
Urgent Notice on COVID-19 prevention and control to
the May 1 holiday. Since the outbreak, the industry has are hc
suspend the business activities of tourism enterprises,
acquired high attention from international organizations to ealt H
requiring travel agencies and the online tourism industry
the central ministries and commissions, and from local and
of China to suspend the operation of group tourism and
government, industry association to the tourism enter- nt e
“air ticket + hotel” tourism products. As of February 1,
prises, and tourism-related aspects. Taking positive action
450 million people had canceled or postponed their Spring
and dealing with unprecedented pressure that the tourism nagem a Festival trips.
industries are facing should manage well in two aspects: M kis
Statistics show that during the Spring Festival in 2018,
the first is to provide epidemic prevention and control, R
the country received 386 million tourists, rising
the second is to introduce all kinds of policy for supporting
12.1% year on year. Tourism revenue reached 475 billion
all kinds of damaged industries and enterprises. As for the
Yuan, rising 12.6% year on year. During the Spring
various policies issued by the government, through the
Festival in 2019, 415 million tourists traveled across the
Internet reports of major media, the public can express
country, rising 7.6% year-on-year. Tourism revenue
their own opinions and cognitive emotions on the public
reached 513.9 billion Yuan, up 8.2% year on year. By
events they care about, thus forming mixed opinions on
2020, more than 450 million tourism revenues have been
the revitalization of tourism policy.
lost. At the same time, online travel agency (OTA) plat-
Based on the comprehensive analysis of Weibo content
forms such as Ctrip and Tuniu have invested more than
and online comment related to the policy of “revitalizing
hundreds of millions of Yuan in cancellation fees, and
tourism”, this paper summarizes the policies that netizens
more than 260,000 travel agencies are struggling.
have paid close attention to and discussed enthusiastically,
According to STR, the hotel occupancy rate on the
striving to cover and describe the public’s response to the
Chinese mainland peaked at 70% in early January 2020,
government’s policies in the cyberspace to the greatest lOMoAR cPSD| 36066900 Chen et al Dovepress
extent. Firstly, this paper summarizes government policy
government policy “online tourism,” the topic #Travel
as five major series and regards them as the first-level
around China online# and the netizens’ comments from
subject category, collects the network comment, and
April 25 to May 5, 2020 were selected. In the second
chooses the topic with most comments on each category
category, under the government policy of #Many pro-
acknowledged by the CCTV news, the blue whale finan-
vinces define 2.5 days off#, the netizens’ comments on
cial journalist work platform, Sina News, People’s Daily,
five topics on January 14, 2020, from March 19 to April 1,
and the Paper. Further, it uses the octopus Weibo topic
from April 26 to July 24 were selected. In the third 021 2-
review as a selected collection to distinguish between
category, under the government policy #Measures in sce- un J-
policy and Weibo topics. In order to distinguish policies
nic spots (free tickets or restricted access or real-name 11 on
and the post in Weibo, ”#” is used to annotate. The
purchasing system)#, eight hot topics were selected and 0 3
categories of policies and online discussion topics for
the discussions were divided into two categories: because 0. 11.
revitalizing tourism are shown in Table 1. (The number
Huangshan Mountain scenic spot was congested, the gov- 23.
of the comments in Table 1 are all from Weibo. The
ernment introduced a series of policies for the situation 113 y
limitation search of Weibo is as follows: first, it may
rapidly to adjust scenic spot, which arouses heat discus- b /
produce a lot of meaningless information; second, the
sion in the Internet. The comments with regard to om c.s
search of Weibo still adopts the traditional way, and the
#Huangshan Scenic Spot starts emergency plan# from s re
information content cannot be reprocessed.)
April 5 to April 6 were collected. After the introduction ep ov
of policies for national scenic spots, the comments were d. . y w w onl
Data Mining Based on Online
collected from February 19 to March 18, and from w // e : s s u
Comments of Travel Policies During April 13 to April 15. In the fourth category, under the pt ht onal COVID-19
government policy #The tourism industry in many places rs
across the country resumes business#, the netizens’ com- romf pe
The Internet represents an important channel for people to
ments from February 21 to April 29, 2020 and from or ded F
express their interests and emotions. Since the COVID-19
July 14, 2020 to July 15 were collected. In the fifth nloa
outbreak, people have become more vocal on the Internet
category, under the government policy #Travel coupons ow d
due to restrictions on travel, and there has been a lot of y
issued in many places across the country#, the netizens’
discussion about the travel-related topic. As a reflection of olic P
comments from March 26, 2020 to July 1, 2020 were
public sentiment, the influence of online public opinion is are selected. hc
not only manifested in its influence on major develop-
Through octopus crawling to get the corresponding ealt
ments, but also penetrates the political level, becoming H
Weibo topic comments, as the original data contains
an important channel for the government to listen to the and
some trivial comments, these trivial comments will inter- nt
voices of the people and understand public opinion. In e
fere with the subsequent analysis results, so the trivial
order to explore the public’s attitudes towards various
data needs to be processed. In view of the requirement of nagem a
official tourism policies during COVID-19, this section
emotion analysis, the data needs to be processed. First, M k
conducts data mining on the comments on the Weibo is
the symbols, emoticons, punctuation, and some useless R
topic mentioned above. Firstly, the data of netizen com-
marks in the comments should be removed. The second
ments were selected and cleaned, then the pre-processed
is to clean out comments irrelevant to the policy, such as
netizen comments were analyzed from the perspective of
advertising, the word count comments, etc. The software
policy perception, and the categories of policies were
Python is mainly used to remove Chinese and English
divided. Finally, the visualization analysis and emotional
symbols, emoticons, invalid texts (such as “ah”, “en”,
analysis of netizen comments were conducted based on
and other modal words, comments that are not related to different policy categories.
the tourism topic of this article, and irrelevant advertise-
ments), and some key codes to realize this function are The Selection and Cleanout of
as follows: import re; the line = line. decode (“utf8”); Comments
String = re.sub (“ [\s +\!\/_ $% ^ * (+\“\”] + | [+ -.??, ~
According to the second-level microblog topics, comments
@ # $% and * () “+”. decode (“utf8”), ““. decode
on Weibo topics at different times were selected as the
(“utf8”),line). The data after cleanout is shown in
research objects. In the first category, under the Table 2.
3216 submit your manuscript | www.dovepress.com
Risk Management and Healthcare Policy 2020:13 DovePress lOMoAR cPSD| 36066900 Dovepress Chen et al
Table 1 Policies on Revitalizing Tourism and Classification of Online Discussion Topics The Primary Category The Secondary Category The Number of Comments (Government Policy)
(Weibo Discussion Topic)
Collected on Weibo Topics
Travel coupons issued in many places across the #National version of the consumer coupons will be 5439 country
launched tomorrow# Here comes the strategy
#Hangzhou will issue 1.68 billion Yuan consumption 021 coupons# 2- un J-
#Wuhan will issue 500 million Yuan consumption coupons# 11
#Zhejiang Jiaxing issues 200 million Yuan consumption on 03 voucher# 0. 11.
Many provinces define 2.5 days off
#Zhejiang encourages 2.5 days off a week# 17,731 23.
#2.5 days of weekend vacation system will be implemented 113 y in three places# b / om
#Jiangxi tries out 2.5-day flexible work and rest on c.ss weekends# re ep
#2.5-day flexible vacation system implemented in Yichang, ov d. . y Hubei Province# w w onl w // e
#It is suggested that one of three flexible weekend vacation : s s u p systems can be implemented# t ht onalrs
Measures in scenic spots (free tickets or
#Notice on stopping receiving tourists in Huangshan Scenic 30,961 romf pe
restricted access or real-name purchasing Area# or ded F system)
#112 scenic spots in Sichuan are free of admission to all nloa tourists in April# ow d y
#Huangshan scenic spot is congested# olic P
#Huangshan Scenic Spot starts emergency plan# are hc
#The Ministry of culture and tourism requires that the ealt
opening of scenic spots should strictly control the flow# H and
#Real-name purchasing system is required for Sichuan nt e scenic spots to reopen#
#Visitors are required to be 1 meter apart in the opening of nagem a the scenic area# M kisR
#During the epidemic period, only outdoor areas are opened#
Online travel (live travel or live commerce) #Travel around China online# 38,216
The tourism industry in many places across the #The West Lake in Hangzhou will open up orderly from 5306 country resumes business today#
#Ministry of culture and tourism issues notice to resume
inter-provincial team Tourism#
#Yunnan tourism industry resumed business# lOMoAR cPSD| 36066900 Chen et al Dovepress
Table 2 Selected Data After Cleanout Classification of Economy Safety Idle Feasible Policy Perspectives Specific policies Consumption Scenic Huangshan Travel around 2.5 days Resuming Resuming inter- coupon spot crowded China online off tourism provincial travel policy a week industry 021 2- un Number of 4,473 8,286 20,802 26,808 17,156 3,251 1,710 J- comments 11 on 0 Percentage of 5.42% 67.76% 20.80% 6.01% 3 0. comments 11. 23. 113 y
Analysis of Netizens’ Perception of the
the scenic spot is open, such as “Resuming tourism indus- b / Policy
try”, “Resuming inter-provincial travel”. om c.s
The Classification of Comments
Although policymakers formulate from the above four s re
Netizen perception refers to whether the netizen’s under-
perspectives, different netizens have different understand- ep ov
standing of the policy is consistent with policy-makers’
ing of policy. Taking coupon for example, some netizens d. . y w w
desired goals. From the perspective of policy, this paper
consider it from an economy perspective, and point out onl w // e : s u
divides the above five policy categories into four cate-
that “in order to stimulate consumption, various places s p t ht
gories, namely, economy, safety, idle, and feasible cate-
think about the different ways”. From a safety perspective, onal rs rom
gory. The economy category refers to the policy-makers’
some netizens think “it will lead to offline congestion, f pe or
expectation to achieve the purpose of promoting the econ-
which is not safe”. Therefore, in order to understand ded F
omy through the policy, such as “consumption coupon”.
netizens’ perception of the four types of policies, we nloa ow
The safety category refers to the policies formulated by d
divided netizen comments into five types according to y
policy-makers to avoid crowd gathering and virus cross-
their comment content and designated tag numbers for olic P
infection in scenic spots from the perspective of safety,
subsequent data processing. The specific division criteria are hc
such as “Scenic spot policy”, “Huangshan crowded”, and are shown in Table 3. ealt
“Travel around China online”. The idle category refers to
According to the above criteria, all the comments made H
the policy made by the policymaker from the perspective
by netizens should be classified. Since the total number of and nt
of whether tourists have free time to travel, such as “2.5 e
comments on all policies is as huge as 82,486, TextCNN
days off a week”. The feasible category refers to whether
convolutional neural network is adopted in this paper to nagem a M kis
Table 3 Policy Division Criteria from the Perspective of Netizens R Classification of Tag Meaning Comment Examples Netizens’ Number Perspective Economy 0
When it comes to money, the economy, etc
“Just back to work, no money” Safety 1
When it comes to epidemic situation, safety, etc
“Is it safe? The epidemic is not over” Idle 2
When it comes to holidays, whether you have time to “Oh, where did you get your vacation when you just travel, etc
went to work?” “school is not allowed to go out” Feasible 3
From the perspective of whether the scenic spot is “Many scenic spots are not open. How can I get
open or not, we can see whether the tour is feasible there?” Others 4
Other netizens’ comments except for the above four “Like” “support” categories
3218 submit your manuscript | www.dovepress.com
Risk Management and Healthcare Policy 2020:13 DovePress lOMoAR cPSD| 36066900 Dovepress Chen et al
automatically classify the comments. TextCNN is a deep
5) Prediction by model: The model trained in the pre-
learning algorithm. By inputting a training set and
vious step is used to automatically classify the remaining
a verification set with classification tags, the computer
more than 60,000 pieces of data.
can automatically learn the classification method, so as to The Analysis of Results
classify and predict other data. The algorithm includes five
All comments were classified by the above steps, and the
parts: word list construction, word vector construction,
classification results are shown in Table 4. 021
convolution, maximum pooling, and K classification. The 2-
As can be seen from Table 4, netizens’ comments are un
specific steps for classifying netizen comments are as J-
mixed with many contents irrelevant to the policy perspec- 11 follows:
tive, that is, comments of other categories take up a large on
1) Data cleaning: emoji, spaces, blank lines, and other 0 3
proportion. In addition to these unrelated perspectives, neti- 0.
contents should be cleaned up to make the message con- 11
zens have different perceptions of various policies. .
tent more concise. Remove invalid messages, such as 23.
Overall, the perception of policies is ranked as econ- duplicate 1, 11, etc. 113
omy > idle > feasible > safety. y b
2) Training data selection: 20,000 pieces of data were /
In terms of economy, “consumption coupon” policies om
randomly selected from more than 80,000 comments for c.
were the most popular, with 54.8% of netizens expressing s s
model training, accounting for about 25% of the total data. re
their views from an economic perspective, which may be ep
A better model training effect will be achieved. The policies
related to the timing and characteristics of the coupon ov d. . y
of each category were selected in proportion, and finally 1,240 w
policy. Consumption coupon aims to stimulate the social w onl w
economic categories, 12,600 security categories, 4,780 free // e
economy after the epidemic turned around, when the epi- : s s u p
categories, and 1,380 feasible categories were selected. t
demic was not serious and people were less worried about ht onal
3) Manual labeling of the selected training data: the rs
safety. At the same time, the policy also has low require- romf pe
selected netizen comment data of a total of 20,000 pieces
ments for travel. People can use coupon for offline dining or ded F
were manually classified according to the classification
and shopping without worrying about no free time. nloa
labels of five types of netizens, and each data was labeled
In terms of safety, the netizens’ perception of the three ow d (label number is 0–4). y
policies was relatively low, with 6.7%, 18.5%, and 0.3% of olic
4) Training neural network model: 20,000 pieces of data
the netizens respectively expressing their opinions from the P
were randomly divided into a training set, validation set, and
perspective of safety. Under the “scenic spot policy”, most are hc
test set according to a proportion of 80%, 10%, and 10%,
people think about the travel problem. Under the topic ealt H
and put into the model for training. The test set was used to
#Huangshan crowded#, people have considered both econ- and
test the classification accuracy of the final trained model, and
omy and safety issues. Some people mentioned the eco- nt e
the results showed that the accuracy of the training set
nomic help of the free ticket policy, but people are still
reached 88.6%, indicating that the model training effect
concerned about the virus infection caused by cluster beha- nagem a
was good and could be used for classification prediction.
viors during COVID-19. Under the topic of #Travel around M kis R
Table 4 Classification Results of Netizens’ Comments
Classification of Policy Perspectives Specific Policies
Classification of Netizens’ Perspective Economy Safety Idle Feasible Others Economy Consumption coupon 54.80% 0.30% 0.10% 0.40% 44.30% Safety Scenic spot policy 13.40% 6.70% 0.50% 17.20% 61.90% Huangshan crowded 18.90% 18.50% 0.20% 2.20% 60.00% Travel around China online 0.80% 0.30% 3.80% 0.10% 94.70% Idle 2.5 days off a week 11.70% 0.60% 36.80% 0.30% 50.40% Feasible Resuming tourism industry 7.10% 2.70% 0.40% 15.90% 73.80%
Resuming inter-provincial travel 9.00% 8.70% 0.10% 22.20% 59.80% lOMoAR cPSD| 36066900 Chen et al Dovepress
China online#, people expressed their opinions from a more
they were relatively redundant and dispersed, the software
free perspective, such as “Travel around China online every
ROST CM5.8.0 was used to classify the emotional ten-
weekend at 9:30 am”. This policy allows people to enjoy the
dency of user comments, and analyze the policies that
beautiful scenery of different places without leaving home,
aroused better public perception and the emotional ten-
which disperses the risk of crowd gathering brought by
dency of the public on such policies. Then, according to
offline travel, and there is no time limit.
the visual analysis technology of the semantic network, the
In terms of idle, netizens with the #Many provinces
hot public concern about the travel during the epidemic 021 2-
define 2.5 days off# policy perceived better, with 36.8% of
period is mined, which can be used as the basis for the un J-
netizens expressing their opinions from the perspective of
follow-up test of whether the public perceives the policy. 11 on
idle. Some netizens were skeptical of the policy, saying 0 3 0. “we don’t even have a two Sentiment Analysis -day weekend anyway.” Some 11.
netizens understood the purpose of the proposal, saying
ROST CM31 software is a digital research platform for 23.
that “tourism is an important pillar of future development,
humanities and social sciences based on content mining. It 113 y
is a group of digital academic research platforms with close b
so we can take a few more days off.” /
functional connections, which can collaborate intelligently om
In terms of feasibility, netizens’ perception of the two c.ss
policies was general, 15.9% and 22.2% of netizens, respec-
with each other, and finally conduct an intelligent analysis of re
humanities and social sciences according to a certain para- ep
tively, expressed their opinions from the perspective of feasi- ov
digm. ROST CM software is capable of semantic network d .
bility. In early February, netizens commented on the . y w
and emotion analysis. In this paper, ROST CM5.8 is used for w onl
restoration of the scenic spot, questioning whether it should w// e : s
emotion analysis, and the analysis results are used to inte- s u
be open to the public. In April, when the epidemic was greatly pt
grate the proportion of positive, neutral, and negative com- ht onal
controlled in China, most netizens expressed their support for rs
ments brought by the implementation of four kinds of romf
the opening of the Yellow Crane Tower scenic spot, saying pe
policies during COVID-19, as shown in Table 5. or ded F
that it was getting better and better after a long time.
According to the proportion of positive, neutral, and nloa
negative comments in Table 5, 37.90% hold a positive ow d y
Visual Analysis Based on Different Policy attitude, 54.68% hold a neutral attitude, and only 7.42% olic P Categories
hold a negative attitude towards the online tourism policy are
The online comments of the public on the policies during
issued by the government, indicating that most netizens hc
the epidemic period were obtained from the websites. As
have a favorable impression and support this policy. For ealt H and
Table 5 Sentiment Analysis of Various Weibo Topics nt e
Classification Policy Category Topic The Proportion of The Proportion of The Proportion of Positive Emotions
Negative Emotions Neural Emotions nagem a M k Safety Online travel
Travel around China online 37.90% 7.42% 54.68% is R Scenic spot measures Measures for Huangshan 31.53% 37.96% 30.51% Scenic Area National scenic spot 48.99% 16.21% 34.79% measures Idle
Many provinces define 2.5 Many provinces define 2.5 37.49% 23.18% 39.33% days off days off Economy Coupons issued in many Coupons issued in many 46.78% 15.04% 38.18% places across the country places across the country Feasible National Tourism
Resuming tourism industry 49.91% 18.78% 31.31% recovery throughout the country Resuming inter-provincial 47.31% 15.60% 37.09% travel
3220 submit your manuscript | www.dovepress.com
Risk Management and Healthcare Policy 2020:13 DovePress lOMoAR cPSD| 36066900 Dovepress Chen et al
tourists to gather together after Huangshan scenic spot
user comments. The main purpose is to find the words that
opened, although the government issued a series of mea-
are mentioned most in netizens’ comments, so as to judge
sures to control the flow, 37.96% of the population held
whether people perceive the content of the policy. By
a negative mood for this event, which is more than the
ROST CM5.8.0 analysis, a relevant semantic network
number of people with positive emotions. It indicates that
diagram can be obtained. Figures 1–7 represent the seman-
government departments should prepare for emergency
tic network diagram of user comment content under each
measures and security measures before opening a scenic
travel topic, and represent the relationship between words 021 2-
spot, rather than wait and solve problems when they arise. in each user comment content. un J-
Measures of the scenic spots in the country mainly include 11 on
“online booking”, “flow restriction”, etc. Then, 48.99% of
The Analysis of Safety Policies 0 3
the people show positive emotions, which are far more
The semantic graph of the Internet is obtained by analyz- 0. 11.
than negative emotions, indicating that the implementation
ing the comments made by netizens under the topic of 23.
of scenic spot measures in various places has a significant #Travel around China Online#. 113 y
effect, which can give people enough sense of safety
From Figure 1, netizen focus more on the reporters and b /
during the trip. The negative comments on the “2.5 days
anchors of online travel live-streaming. “Travel around om c.s
off” policy reached 23.18%, when the positive comments
China online” is broadcasted at a fixed time every day. s re
reached 37.49%. Although the government called on
Generally speaking, beautiful anchors or journalists attract ep ov
every medium enterprise to extend rest time, under
more audience, especially web celebrity “Weiya”. d. . y w
COVID-19, employees hope to recover economic loss as
Affected by the COVID-19 epidemic, people can neither w onl w // e : s
soon as possible, so more people heold a neutral attitude.
travel abroad nor buy travel products. Through online s u p t
Under the “economy” policy, 46.78% of netizens held
mode, people can be personally involved and promote ht onalrs
a positive attitude, while only 15.06% hold a negative
the consumption of tourism products in remote areas. romf pe
attitude, indicating that most netizens held a positive atti-
According to Figures 3 and 4, Xinjiang, Sanya, Shanxi or ded F
tude in support of the policy. It can be seen from the
Pingyao, Changbai Mountain, Guilin, and other places nloa
“feasible” policy that the number of people holding posi-
have attracted more netizens’ attention. Through live ow d y
tive emotions in this policy is much higher than the
streaming of local food and scenic spots, people want to olic P
number of people holding negative emotions, by more
travel in person, which accelerates the sales of tourism are
than 30%, indicating that this policy is supported by the products. hc masses.
In addition, the Internet semantic graph of Figures 2 ealt H
and 3 is obtained by analyzing the comments made by and
Visual Analysis Based on Semantic Network Under
netizens under the topics #Huangshan Scenic Spot mea- nt e Different Policy Categories
sures# and #National Scenic spot measures# on Weibo.
According to ROST CM analysis of the proportion of
Since announcing it was open, Huangshan Mountain nagem a
positive and negative opinions in people’s online com-
scenic area has been crowded, indicating that, after stabi- M kis
ments, those posts under the four categories of policies,
lizing of the epidemic, the majority of people are chasing R
namely “safety”, “idle”, “economy”, and “feasible”, are
for tourism, but congestions cause certain difficulties for
generally positive. Here, in order to obtain the major
epidemic control. Huangshan Mountain scenic area started
concerns of the public under each type of policy, semantic
the emergency plan, stopped serving tourists, preventing a
network visual analysis was adopted to analyze the online
widespread infection epidemic. It can be seen from Figure
comment data of the four policies under the background of
2 that Tomb-sweeping Day (the first holiday after the
COVID-19. The semantic network is one of the represen-
epidemic), has become the relaxation of travel for people
tations of an artificial intelligence program, which
all over the country after they have been “confined” for
expresses human knowledge construction in the form of
a long time, which also increases the difficulty for the
a network. It consists of arcs between nodes, where nodes
government to prevent and control the epidemic. “Wear
represent concepts (events or things), and arcs represent
Mask” has become important and necessary. In Figure 3,
relationships between them. The semantic network dia-
the nodes of “Anhui”, “Huangshan”, and “local” indicate
gram is used to represent the degree of association
that most tourists to Huangshan scenic spot are residents
between words and reflect the most concerned words in
of Anhui province. Many tourism enterprises, such as lOMoAR cPSD| 36066900 Chen et al Dovepress 021 2- un J- 11 on 030. 11. 23. 113 yb/ om c.ssre ep ov d. . y w w onl w / e /: s s u
Figure 1 Semantic graph of #Travel around China Online#. p t ht onal rs romf pe or ded F nloa ow d y olic P are hc ealt H and nt e nagem a M kisR
Figure 2 Semantic graph of #Huang shan Scenic Spot measures#.
Ctrip and Feizhu, have launched local and provincial tours
limiting the visiting number of scenic spots, offering free
to narrow the traveling scope and make the public feel
tickets, and extending the opening hours of scenic spots to
more assured. As can be seen from the “current-limiting”,
provide people with a safe place to travel, and at the same
“making an appointment”, “free”, and “time” nodes, the
time actively controlling the mass influx of people into
government has taken measures such as online booking,
scenic spots. Despite a series of measures to control flow,
3222 submit your manuscript | www.dovepress.com
Risk Management and Healthcare Policy 2020:13 DovePress lOMoAR cPSD| 36066900 Dovepress Chen et al 021 2- un J- 11 on 030. 11. 23. 113 yb/ om c.ssre ep ov d. . y w w onl w // e : s
Figure 3 Semantic graph of #National Scenic spot measures#. s u p t ht onalrs romf pe or ded F nloa ow d y olic P are hc ealt H and nt e nagem a M kisR
Figure 4 Semantic graph of #Many provinces define 2.5 days off#.
for the emergence measures taken by Huangshan, 37.96%
while recovering tourism, government should consider
of people hold a negative mood for this event, which is
unexpected circumstances, optimize emergency measures,
more than the number of positive emotions. It indicates
and avoid crowding in advance. lOMoAR cPSD| 36066900 Chen et al Dovepress 021 2- un J- 11 on 030. 11. 23. 113 yb/ om c.ssre ep ov d. . y w w onl w // e : s u
Figure 5 Semantic graph of #Coupons issued in many places across the country#. s p t ht onal rs romf pe or ded F nloa ow d y olic P are hc ealt H and nt e nagem a M kisR
Figure 6 Semantic graph of #Resuming tourism industry throughout the country#. The Analysis of Idle Policies
Hubei, Zhejiang, Jiangxi, Hunan, and other provinces
The semantic graph is obtained by analyzing the com-
responded positively. Figure 4 shows that around the
ments made by netizens under five topics.
“rest” node, most people mention “implement”, “work
In order to stimulate the recovery of the tourism indus-
overtime”, “double rest”, “civil servants”, and “private
try, the government proposed the “2.5 days off” policy.
enterprise”. It shows that most of the private personnel
3224 submit your manuscript | www.dovepress.com
Risk Management and Healthcare Policy 2020:13 DovePress lOMoAR cPSD| 36066900 Dovepress Chen et al 021 2- un J- 11 on 030. 11. 23. 113 yb/ om c.ssre ep ov d. . y w w onl w // e : s
Figure 7 Semantic graph of #Resuming inter-provincial travel#. s u p t ht onalrs romf pe
hope to implement 2.5 days off as early as possible, and
and “stimulate the market”. The coupon platforms include or ded F
guarantee their overtime pay. For state-owned enterprises,
“Alipay”, “Dianping”, and “Mafengwo”. As the most nloa
organizations and institutions, employees rarely work
common application in the national people’s life, ow d
overtime compared with people from private sectors, so
“Alipay” has spread to all industries. Hangzhou, the city y olic
they do not extravagantly hope for 2.5 days off. It can be
where Alibaba is located, also responded positively to the P
seen from the nodes “economy” and “development”, this
government’s policies and issued 1.68 billion Yuan of are hc
policy is mainly aimed at stimulating people’s consump-
consumption coupons. “Wuhan”, as an important node, is ealt H
tion in holidays and developing the economy. At present,
the city hit the hardest by the epidemic. For this reason, and
most enterprises have responded positively to this policy
Wuhan issued 500 million Yuan of consumption coupons nt e
by changing the former bonus payment into “complimen-
to stimulate Wuhan’s economy, encouraging small and
tary travel products” or “company incentive group travel”,
middle enterprises to resume work and production, and nagem a
which promotes the recovery of the tourism industry. The increasing
people’s welfare. According to the M k
sentiment analysis results of Table 5 show that negative is
“Consumption” node in Figure 5, the government’s pur- R
emotions reached 23.18%, and the number of positive
pose of promoting tourism consumption by issuing con-
emotion accounts for about 10%. Some people paid hourly
sumption vouchers can be well accepted by the public.
or daily encounter huge economic pressure, so they reduce
People can buy travel vouchers in advance, which can be
the demand for tourism and extend working hours to
used after the epidemic stabilizes, and it is also a kind of afford their family expenses.
advance consumption. At the same time, the government
also advocates daily consumption and issues a series of
The Analysis of Economy Policies
coupons. “Supermarket”, “the Mall”, and “Market” are all
The semantic graph is obtained by analyzing the
perceptions of the coupon policy. If the government wants
comments made by netizens under four sub-category
to emphasize tourism vouchers, the content of the infor- topics.
mation should be clear when the policy is issued. The node
In Figure 5, the above analysis result shows that the
“Positive” indicates that netizens are optimistic about the
national policy of issuing coupons aims to “stimulate
epidemic situation in China and are eager to travel after
economy”, “benefit the people”, “provide public welfare”, the epidemic stabilizes. lOMoAR cPSD| 36066900 Chen et al Dovepress
The Analysis of Feasible Policies
nodes after the visual characteristics of tourism revitaliza-
The semantic graph of the network is obtained by integrating
tion policies under COVID-19, the netizen-related vari-
and analyzing the comments of the netizens under three sub-
ables are selected from the perspective of netizens,
categories of Weibo topics, as shown in Figures 6 and 7.
including: total number of Microblogs, the timeliness of
During April and the Tomb-Sweeping Day, that is, when
comments posted, the division of policy in the view of
the COVID-19 was gradually controlled, the tourism industry
user, users’ emotional score (opinion tendency), thumb up
across the country resumed business one after another.
number in netizens comments, the netizen comments, the 021 2-
Yunnan, as one of the major winter tourism destinations in
gender of the netizens, the number of fans, the number of un J-
China, relies mainly on the cultural tourism industry sup-
the followers, the degree of activity, and the level of 11 on
ported by the floating population economy. Cultural and
development of COVID-19. These 11 variables obtained 0 3
tourism authorities at all levels are restoring confidence,
by the network data after the expansion of crawl can fully 0. 11.
optimizing the modern tourism governance system, and
show the impact of netizens personal influence, individual 23.
opening several scenic spots while strengthening of the epi-
opinion on the effects of the policy implementation. The 113 y
demic prevention and control. In July, when the epidemic
meanings of 11 variables are shown in Table 6. b /
was controlled, the ministry of Cultural and Tourism om c.s
announced the resumption of inter-provincial travel. At the s
The Construction of Binary Logistic re
same time, cinemas began operating. From the node “entry ep ov
and exit”, “unemployment”, “half a year”, and so on, the Regression Model d. . y w
issued policies for returning to work make the unemployed
The dependent variables set in this paper are binary classifi- w onl w // e : s
encouraged, ignite people’s inspiration to travel abroad, and
cation variables (ie, the Boolean variables), therefore, the s u p t
solve the problems for employees with the need to leave the
binary classification logical model (Binary Logistic
ht onalrs country for work or enterprises with foreign business. The Regression) is adopted to study the factors affecting netizens’ romf pe
node “nucleic acid”, “testing”, and “ protect” show that
perception to policy, as well as set an optimal combination of or ded F
although the epidemic is in a stable state at present, the
policy and explain the influence of factors on perception nloa
country has not taken the prevention and control of the
effect. In the regression model, the independent variable is ow d y epidemic carelessly.
X1~X10, and the dependent variable Y represents whether the olic P
policy content is perceived from the perspective of Internet are
The Social Effect Analysis of
users. ε is the error term, assuming that it is independent of hc ealt
Tourism Policies Based on Binary
other variables; βi is the regression coefficient in logistic H
regression; Inð Pi Þ —
represents the logarithmic change value 1 Pi
Logistic Regression Model and
of the ratio for the probability of occurrence to non- nt e
The logistic regression model mainly studies the probabil-
occurrence when Xi changes a unit. By referring to the
ity P of some phenomena and discusses the factors related
definition of logistic model in literature,32 the influencing nagem a
to the probability P. In this article, studying whether peo- M
factor model of netizens’ perceived effects on tourism poli- k
ple perceive the government policy belongs to the 0–1 is
cies is constructed, as shown in Equation (1). R
binary classification variables. Therefore, by constructing Pi
a strictly monotone function Logistic (P) to study the Y ¼ Inð
Þ ¼ β X1 þ β X2 þ β X3 þ β X4 þ β X5 1 2 3 4 5
model between P and the independent variables, this 1 — Pi
paper selects the binary logistic regression model.
þ β6X6 þ β7X7 þ β8X8 þ β9X9 þ β10X10 þ ε (1)
Variable Selection and Data Definition
The combination of different types of policies cause diverse
From the above analysis, various policies to revitalize
social effects, leading to different public perception of the
tourism during COVID-19 arise people’s attention and
degree as well as the influencing factors of netizens’ percep-
discussion on the Internet. As the main channel of major
tion. In order to find the best combination of tourism promo-
government policy, the new media must consider the guid-
tion policy, by dividing tourism policy into four categories
ing force of influential people, who are represented by
and combining them, 15 kinds of policy combinations can be
a large quantity of Weibo and fans. Combining the analy-
acquired (economy, idle, feasible, safety, economy + safety,
sis of the emotional distribution and semantic network
economy + feasible, economy + idle, idle + safety, idle +
3226 submit your manuscript | www.dovepress.com
Risk Management and Healthcare Policy 2020:13 DovePress lOMoAR cPSD| 36066900 Dovepress Chen et al
Table 6 Variables and Definitions Table 6 (Continued). Variables Code Definitions Variables Code Definitions Dependent variable Y
Judge whether the classification Level of development of X10 According to the level of (Policy implementation
of the netizen’s perspective and COVID-19
response to major public health effect):
policy perspective is consistent, emergencies, The first level The division of policies
if it’s 1, otherwise it’s 0 (If it is response was from January 24 from the perspective of
consistent, it indicates that this 021 2 to April 30, the second level - netizens
policy is effective and produces un response was from April 30 to J- better social effect). June 12, the level 3 response 11 on The policy concerns the X1 The number of netizens’ was from June 12 to July 30. 0 3 total number of comments on Microblog topics 0. Microblogs under each policy category. 11. 23.
feasible, safety + feasible, economy + idle + feasible, econ- Timeliness of comments X2 The Microblog topics involved 113 posted
under each policy category, the
omy + idle + feasible + safety). With regard to the policy y b / average difference between
combinations, the binary logistic regression model is set up. om c netizens’ comment time and .
In the binary logistics regression model, how many indepen- s s Weibo release time. re
dent variables are introduced needs to be studied. If fewer ep Opinion tendency X3 ROST was used to score each
independent variables are introduced, the regression equation ov d. . y netizen’s comments w
will not be able to explain the changes of dependent variables w onl w
emotionally, a number less than // es
in an accurate manner, but it does not mean that more : s u zero indicates a negative p t
independent variables are absolutely better. Therefore, it is ht emotion, A number greater onal rs
than zero is a positive emotion,
necessary to adopt some strategies to control the independent romf pe equals zero is neutral. or
variables by introducing regression equations. The Stepwise ded F Thumb up number X
Selection method is adopted here, that is, the introduction 4 The number of thumb ups each nloa ow netizen comments on the
threshold of P-value is tested according to the significance of d y policy from other netizens.
the set regression coefficient, independent variables are intro- olic P Comment number X5 The number of thumb ups each
duced into the model one by one, then P-values of all coeffi- are netizen comments on the
cients in the model are recalculated, and variables are hc policy from other netizens.
screened according to the set elimination threshold. The ealt H Netizens gender X6 Gender of netizens
Stepwise Selection method includes forward selection and participating in comments
method and backward selection method. The forward selec- nt e
under each policy category, it’s
tion method is relatively simple, but the biggest disadvantage 1 for women and 0 for men. nagem
is that if there is multicollinearity, the final model may be a M Number of fans X7 The number of followers of
mixed with less important independent variables. The back- k is netizens participating in R
ward selection method is more conservative in terms of comments under each policy
information. The backward selection method is chosen in category.
this article. Data processing was conducted in SPSS25.0 Number of followers X8 Netizens participating in
software, and the significance level of entering the model comments under each policy
was set at 0.05, and the significance level of removing or category, the number of
followers of other netizens on
retaining variables was also set at 0.05. Weibo.
The main steps of regression analysis are as follows:
using LR Likelihood Method to select the independent The degree of active X9 The number of original Microblogs posted by netizens
variables with a significant relationship between public participating in comments
perception effect, conduct significance test for the variable under each policy category.
regression coefficient and model to get binary logistics (Continued)
regression equation between each variable and dependent lOMoAR cPSD| 36066900 Chen et al Dovepress
Table 7 Significance Test of Independent Variables Selected by “Economic” Policies Variable Model Log Likelihood
Change in-2 Log Likelihood Degrees of Freedom
Significance of the Change Step1a X2 −3031.611 437.468 1 0.000 X3 −2812.878 0.003 1 0.960 X4 −2813.892 2.031 1 0.154 X5 −2814.068 2.383 1 0.123 X6 −2813.835 1.916 1 0.166 021 2- X7 −2813.162 0.571 1 0.450 un J- X8 −2814.975 4.197 1 0.041 11 X9 −2813.964 2.175 1 0.140 on X10 −2848.757 71.761 1 0.000 0 3 0.
Note: aMeans backward stepwise selection regression method is adopted in regression analysis. 11. 23 . 113
Table 8 The Variable Coefficients in the “Economic” Policy Equation y b / B Standard Error Wald Degrees of Freedom Significance Exp(B) AIC om c.s Step7aX2 0.028 0.002 310.135 1 0.000 1.028 s re ep X8 0.000 0.000 4.090 1 0.043 1.000 ov d. . y w X10 0.309 0.037 70.078 1 0.000 1.361 w onl w // e : s u Constant −0.688 0.086 63.620 1 0.000 0.502 s p t ht onal 519.426 rs romf pe Step1a 540.505 or ded F
Note: aMeans backward stepwise selection regression method is adopted in regression analysis. nloa ow d y olic P
variable, and test the prediction accuracy of the overall
hypothesis is usually 0.05. When the decision to accept are
model. The following is a detailed introduction to the
the original hypothesis is made, the probability of its hc
regression analysis of the public perception effect under
correctness is 95%. It can be seen from Table 7 that in ealt H
the “economy” tourism policy, and the regression analysis
the selection of independent variables, the significance and
process of the remaining 13 tourism policy combinations
level Sig.<0.05 is X2 for the timeliness of comment nt e can be similarly obtained.
release, the number of attention is X8, and the develop- nagem
The regression model construction process of the pub-
ment level of COVID-19 is X10. Therefore, there is a M
lic perception effect under the “economy” tourism policy
a significant relationship between the public perception k is R
is illustrated as an example. Based on the comment data of
of “economy” policies and these three variables, so
“economy” policy obtained above, SPSS was used for
binary logistics regression. After eliminating the indepen-
dent variable X1 which could not be introduced, the inde-
Table 9 Omnibus Tests of Model Coefficients Under the Policy
pendent variable could be selected and the results were of “Economic”
shown in Table 7, the value of regression coefficient of Chi-Square Degrees of Significance
each variable was shown in Table 8, and the model regres- Freedom
sion statistical results were shown in Tables 9 and 10. Step 7aStep −2.673 1 0.102
1. Generally speaking, the significance level is to estimate Block 526.017 3 0.000
the probability of wrong parameter within a certain Model 526.017 3 0.000
interval. When the original hypothesis is true and repre-
Note: aMeans backward stepwise selection regression method is adopted in
sented by α, the probability of rejecting the original regression analysis.
3228 submit your manuscript | www.dovepress.com
Risk Management and Healthcare Policy 2020:13 DovePress lOMoAR cPSD| 36066900 Dovepress Chen et al
Table 10 The Final Model of “Economic” Policy Predicts Rates
conclusions can be made that the release of “safety” policy Observed Predicted
brings better social effect, whose social policy implemen-
tation effect is 93.2%. If the “safety policies + feasible Netizens Netizens Did Percentage
policies” and “economy policies + safety policies” can be
Have a Sense Not Perceive Correct of the Policy the Policy
released together, people’s perceptions are 92.2%, 89.3%,
followed by the “economy + safety + feasible policies”, Step 1341 680 66.4 021 7aNetizen’s which is 88.5%. 2- un perception of
As can be seen from Table 5, the main topic of “safety” J- policy: Y
policy on Weibo is “Travel around China Online”. By 11 759 1693 69.0 on
means of online travel, people lower their travel frequency 0 3 0. Overall 67.8
while keeping their enthusiasm for travel consumption 11. percentage
after COVID-19. Moreover, during the epidemic period, 23.
Note: aMeans backward stepwise selection regression method is adopted in
the safety of the tourism environment is the most con- 113 regression analysis. y
cerned and worried issue. The most fundamental problem b / om
for the revitalization of the tourism industry is to control c.s
independent variables of the binary logistics model
the development of the domestic epidemic. After the s re
security is guaranteed, the “economy” and “feasible” poli
under such policies can be screened out. - ep ov d .
2. According to the selected independent variable, the
cies will be released at the same time, which can promote . y w
the revitalization of the tourism industry. As for the “econ w onl
coefficient of its independent variable is determined - w // e : s
omy” policy, it is mainly to issue tourism coupon, while s u
by the value of Exp(B) in Table 4. Exp(B) is also p t ht
known as the preponderance ratio, which means
the “feasible” policy is to open major tourist attractions or onal rs rom
that the preponderance ratio is twice that of the
to resume inter-provincial travel. Only when the tourist f pe or
original Exp(B) when the other independent vari-
attractions start to operate normally, can people use tour- ded F
ables are fixed and unchanged.
ism coupon to recover the tourism industry through scenic nloa ow
3. The regression equation of public perception under spot consumption. d y
the “economy” policy can be obtained as follows:
2) The combinations of “economy + feasible”, “economy + olic P
idle + feasible”, “economy + idle” bring poorer social effect are
than other policy combinations. This is because if the govern- hc
Y ¼ 0:465 þ 1:029X2 þ X8 þ 1:38X10 (2)
ment does not release the information of epidemic develop- ealt H
Due to the Sig. <0.05 in Table 9, it indicates that model is
ment situation nor the safety measures for travel, people are not and
significant under 95% significance level, so the X2, X8, and
interested in traveling. Before safety is guaranteed, the effect of nt e
X10 can be used as factors that influence the public perception
policy is poor. Based on this, it is suggested that the govern-
of policies. According to the model prediction accuracy in
ment should report the development of epidemic situation at nagem a M
Table 10, the probability of the model in formula (2) that can
home and abroad in real time, as well as the relevant safety k is
predict the public perception of policy is 67.8%.
measures taken in tourist attractions or the process of tourism R
Similarly, according to the above policy categories, the
in the content of the revitalization of tourism policy, so as to
basic four kinds of policies are combined. When policy
ensure more obvious social effects after the release of the
content contains one, two, three, or four kinds of information, policy.
15 class policy combinations can be obtained. The construc-
3) From Table 11, the conclusions can be made that the netizens’
tion of regression model under 15 combinations all meet the
activeness X9 greatly impacts on public’s perception
effect. In the top five optimized policy combinations, the
test of significance level. The results are shown in Table 11. greater the X
9 of Internet users is, the stronger the perception
Analyzing the Implementation Effect of
about policy is, the greater the effect of policy implementation
brings. The activeness of netizens is the original Weibo Tourism Policy
released by netizens. Generally speaking, the more active
1) After the release of the 15 policy combinations, the
a netizen is on Weibo, the greater his personal influence will
binary logistic regression model was established based on
be, and the stronger the effect of network public opinion will be
public perception of policy. From Table 11, the
caused. For government departments, if they want to expand lOMoAR cPSD| 36066900 Chen et al Dovepress
Table 11 Results of Binary Logistic Regression Model for 15 Policy Portfolios Ordinal Ranking of Different Policy Binary Logistic
Factors Influencing the Social Effect Public Number Different Mix Categories Regression Model of Policies Perception of Policy Policy of the Effects Portfolios Preferences of Policy 1 1 Safety policies
Y ¼ 4:881 þ 0:988X þ
Opinion tendency\Netizens gender\The 93.2% 3 0:789X
degree of active\Level of development of AIC=2,869.762 6 þ X9 þ 0:03X 021 10 2- COVID-19 un J- 2 2 Safety policies + 11 Y ¼ 0:891 þ X
The policy concerns the total number of 92.2% 1 þ 0:996X2 on Feasible policies þ 0:983X
Microblogs\Timeliness of comments AIC=1,641.991 3 þ 0:857X6 0 3 þ 0:399X
posted\Opinion tendency\Netizens gender 10 0.
\Level of development of COVID-19 11. 23. 3 3 Economy policies Y ¼ 1:114 þ X
The policy concerns the total number of 89.3% 1 þ 0:999X2 113 + Safety policies þ 0:747X
Microblogs\Timeliness of comments AIC=4,768.212 6 þ X9 y b /
posted\Netizens gender\The degree of om active c.ssre 4 4 Economy policies
Y ¼ 0:992 þ 0:995X
Opinion tendency\Netizens gender\The 88.5% 3 ep + Safety policies + þ 0:78X
degree of active\Level of development of AIC=229.69
6 þ X9 þ 0:788X10 ov d. . y Feasible policies COVID-19 w w onl w // es 5 5 Idle policies + :
Y ¼ 2:927 þ X1 þ 0:998X2
The policy concerns the total number of 85.5% s u p t Safety policies
þ 0:987X3 þ 0:946X6
Microblogs\Timeliness of comments AIC=2,148.01 ht onal þ X þ þ
posted\Opinion tendency\Netizens gender 8 0:059X9 0:187X10 rs romf pe
\Number of followers\The degree of active or 6 6 Economy policies Y ¼ 2:927 þ X 85.5% 1 þ 0:998X2
\Level of development of COVID-19 ded F + Idle policies + þ 0:987X AIC=218.008 3 þ 0:946X6 nloa Safety policies þ X þ þ 8 X9 0:187X10 ow d y 7 6 Safety policies +
Y ¼ 1:651 þ X1 þ 0:996X2
The policy concerns the total number of 85.2% olic P Idle policies + þ 0:985X
Microblogs\Timeliness of comments AIC=2,746.003
3 þ X4 þ 0:997X5 are Feasible policies þ 0:951X þ þ
posted\Opinion tendency\Thumb up 6 X8 0:354X10 hc
number\Comment number\Netizens ealt
gender\Number of followers\Level of H development of COVID-19 and nt e 8 7 Economy policies Y ¼ 1:59 þ X
The policy concerns the total number of 82.8% 1 þ 0:999X2 + Safety policies + þ 0:991X
Microblogs\Timeliness of comments AIC=4,731.289 3 þ 0:892X6 nagem Feasible policies + þ X
posted\Opinion tendency\Netizens gender a
8 þ X9 þ 0:627X10 M Idle policies
\Number of followers\The degree of active k is R
\Level of development of COVID-19 9 8 Feasible policies
Y ¼ 0:975X1 þ 0:972X3
The policy concerns the total number of 79.8% þ 1:363X6
Microblogs\Opinion tendency\Netizens AIC=60.456 gender 10 9 Economy policies
Y ¼ 0:465 þ 1:029X2
Timeliness of comments posted\Number 67.8%
þ X8 þ 1:38X10
of followers\Level of development of AIC=519.426 COVID-19 11 10 Idle policies + Y ¼ 0:214 þ X
The policy concerns the total number of 66.9% 1 þ 0:986X3 Feasible policies þ 1:277X þ
Microblogs\Opinion tendency\Netizens AIC=514.627 6 þ X8 0:93X10
gender\Number of followers\Level of development of COVID-19 (Continued)
3230 submit your manuscript | www.dovepress.com
Risk Management and Healthcare Policy 2020:13 DovePress lOMoAR cPSD| 36066900 Dovepress Chen et al Table 11 (Continued). Ordinal Ranking of Different Policy Binary Logistic
Factors Influencing the Social Effect Public Number Different Mix Categories Regression Model of Policies Perception of Policy Policy of the Effects Portfolios Preferences of Policy 12 11 Idle policies
Y ¼ 0:53 þ 0:99X
Opinion tendency\Netizens gender\Number 63.2% 3 021 2 þ 1:261X
of followers\The degree of active AIC=91.277 - 6 þ X8 þ X9 un J- 13 12 Economy policies
The implementation effect of the three policy combinations is not as good 63.4% 11 + Feasible policies
as that of the single policy release (This regression model is meaningless, on 0 AIC value is meaningless) 3 14 13 Economy policies 63.2% 0. 11 + Idle policies + . 23. Feasible policies 113 y 15 14 Economy policies 59.4% b / + Idle policies om c. s s re
the social effects brought by the policies, they can forward the ep
Tourism Policies Under the Epidemic ov d .
policies through “opinion leaders” on Weibo to let more people . y
Situation Should Highlight Safety w w onl
know the contents of the policies, promote tourism consump- w // e Measures : s u
tion, and drive the tourism economy. s p t
The epidemic brings great challenge to China’s governance ht
4) From Table 11, except for the “idle + feasible”, “econ- onal rs rom omy”,
system and capacity. For the tourism industry, it is neces-
“economy + safety” policy, opinion tendency degree X3 f pe
sary to carry out the corresponding assessment, prevention, or
impacts on the effect of different policy combination. X3 ded F
regression equation coefficient shows that each type of policy
and treatment, and handle the tourism crisis properly, which nloa ow
combination regression coefficients is above 0.9, indicating
not only requires scientific decision-making and precise d y that X
measures by the government, but also requires the joint
3 had a greater influence on the effect of the policy olic P
implementation. This is because after the release of policy,
efforts of scenic spot practitioners. According to the analy- are
netizens hold positive, negative, or neutral attitudes, which
sis of this paper, the public’s perception of “safety” policies hc
affects the tendency of public opinion. If the policy is released,
is the strongest. Only when tourists perceive that it is safe to ealt H
“opinion leaders” hold negative attitudes, which will not be
travel can the cultural and tourism industry gradually and
conducive to policy implementation and cause rejection on the
resume work and production. Based on this, the government nt e
network platform. Therefore, after the release of policy, gov-
should timely issue a “safety” policy after the initial stabi-
ernment should control trend of public opinion, timely stop
lity of the epidemic. For example, on February 25, the nagem a M
bad happens so as to play the positive impact of policies.
Ministry of Culture and Tourism issued a “Guide to k is R
Prevention and Control Measures for Reopening of Tourist
Conclusions and Suggestions
Attractions”, which enables tourists to perceive that the
This paper selects Weibo comments, comments time, gender,
government encourages the cultural and tourism industry
the original Weibo number, and other aspects from January 11,
to resume work and production, and it is relatively safe to
2020 to July 24 on the revitalization of the tourism policy, and
travel in the current environment. In addition, the govern-
divides the revitalization of tourism policy into “economy”,
ment can issue a series of advocacy policies, such as advo-
“safety”, “idle”, “feasible” four major categories. From the
cating industry associations to strengthen the safety and
perspective of users, by constructing binary logistic regression
supervision of epidemic prevention and control, and guiding
model, the combination of all kinds of policy is analyzed.
scenic spots to actively participate in relevant work;
Based on the above analysis, this paper provides the following
Secondly, the scenic spot should issue a series of effective
suggestions for the government and scenic spots to cope with
and feasible prevention and control policies to ensure safety,
public health emergencies and improve the perceived effect of
such as implementing a series of policies in terms of limit- policies:
ing the capacity of tourists, keeping social distance between lOMoAR cPSD| 36066900 Chen et al Dovepress
tourists, regularly disinfecting all scenic spots, and requiring
sudden outbreak, the release of policy information and tourists to wear face masks.
monitoring of public perception is still lagging behind. In
this regard, local governments and scenic spots can use the
The Combination of Economic Policies
big data platform to accurately locate the spread path of
and Security Policies Can Achieve Better the epidemic, quickly track the flow of tourists and their Results
movements, and establish a tourist relationship map, so as
to provide data protection for “safety” policies and reduce 021 2
The most fundamental problem of reviving the tourism -
tourists’ concerns about safety. In addition, big data can be un J
industry is to control the development of the epidemic. -
used to speed up the connectivity of all kinds of policy 11
When safety is guaranteed, “economy” and “safety” poli- on
information and expand the exposure of policies, so as to 0
cies will be issued at the same time, which will greatly 3
increase the public’s perception of the effect of policies. 0.
accelerate the recovery of the tourism industry. The rea- 11.
lity of the difficulties in view of the current tourism 23.
industry development, in addition to the policy of issuing 113 Funding y b
coupons to attract tourists, governments at all levels can /
This research is supported by the National Social Science om
introduce more perfect tourism industry revitalization c
Foundation of China (Grant No. 20BTQ059), Hubei Key . s s
policy, especially the release and enforcement of fiscal
Laboratory of Mechanical Transmission and Manufacturing re ep
policy, tax policy, credit policy, and social security policy
Engineering (MECOF2020B04), Contemporary Business ov d. . y
for the troubled tourism-related businesses. Providing
and Trade Research Center and Center for Collaborative w w onl
financial subsidies for tourism services and related enter- w e
Innovation Studies of Modern Business of Zhejiang //: s s u
prises to resume operation and production. For tourism p Gongshang University of China (Grant No. t ht onal
enterprises that have special difficulties and fail to pay
14SMXY05YB), as well as First Class Discipline of rs romf pe
tax on time, tax payment shall be reduced or postponed
Zhejiang-A (Zhejiang Gongshang University- Statistics). or ded F
appropriately. Tourist attractions can also release a series nloa
of “economy” policies to attract tourists, such as free Disclosure ow d
tickets or appropriate discounts, multi-scenic joint ticket y
The authors declare that they have no conflicts of interest
discounts and other marketing policies. In addition, “fea- olic for this work. P
sible” policies will be issued at the same time, such as are hc
enforcing the paid leave system for employees and 2.5 References ealt days off policy. H
1. Dave K, Steve L, Pennock DM. Mining the peanut gallery: opinion and nt
Play the Role of “Opinion Leaders” to
extraction and semantic classification of product reviews. Proceedings
of the 12th International Conference on World Wide Web. 2003:519– e
Improve the Perceived Effect of Policies 528.
2. Rossouw R, Saayman M. Assimilation of tourism satellite accounts nagem a
Through analysis, this paper finds that “opinion leaders” have
and applied general equilibrium models to inform tourism policy M
analysis. Tour Econ. 2011;17(4):753–783. doi:10.5367/te.2011.0058 k
a significant impact on the perceived effect of the public after is
3. Miller S. Causal forest estimation of heterogeneous and time-varying R
the release of policies. The greater the influence of “opinion environmental policy effects. J Environ Econ Manage.
leaders” is, the stronger the perception of policies will be, and
2020;103:102337. doi:10.1016/j.jeem.2020.102337
4. Dong F, Liu Y. Policy evolution and effect evaluation of new-energy
the greater the social effect brought by the implementation of
vehicle industry in China[J]. Resour Polic. 2020;67:101655.
policies will be. Therefore, the government can advocate
doi:10.1016/j.resourpol.2020.101655
“opinion leaders” on Weibo to forward and expand
5. Matousek R, Panopoulou E, Papachristopoulou A. Policy uncertainty the
and the capital shortfall of global financial firms. J Corp Finance.
exposure of the policy, so as to improve the perceived effect
2020;62:101558. doi:10.1016/j.jcorpfin.2020.101558
of the policy, let more people know the content of the policy,
6. Livni J. Investigation of collapse of complex socio-political systems
using classical stability theory. Physica A. 2019;524:553–562.
boost tourism consumption and drive the tourism economy.
doi:10.1016/j.physa.2019.04.167
7. Chen T, Shi J, Yang J, Cong G, Li G. Modeling public opinion polariza-
Use Big Data of Tourism to Improve the
tion in group behavior by integrating SIRS-based information diffusion
process. Complexity. 2020;2020:1–20. doi:10.1155/2020/4791527 Effect of Policies
8. Chen T, Wang Y, Yang J, Cong G. Modeling public opinion reversal
process with the considerations of external intervention information
Since the outbreak of the COVID-19, big data of tourism
and individual internal characteristics. Healthcare. 2020;8(2):160.
has played an important role. However, in the face of the doi:10.3390/healthcare8020160
3232 submit your manuscript | www.dovepress.com
Risk Management and Healthcare Policy 2020:13 DovePress lOMoAR cPSD| 36066900 Dovepress Chen et al
9. Alexander A, Yusaku H. Has trump damaged the U.S. image abroad?
21. Dash GH, Kajiji N. On multiobjective combinatorial optimization
Decomposing the effects of policy messages on foreign public
and dynamic interim hedging of efficient portfolios. Int Trans Oper
opinion. Polit Behav. 2020;42(3):581–602. doi:10.1007/s11109-018-
Res. 2014;21(6):899–918. doi:10.1111/itor.12067 9511-3
22. Yang G, Zhang Y, Li X. Impact of gasoline upgrade policy on
10. Ho M. An analysis on the critical success factors of healing tourism particulate matter pollution in China. J Clean Prod.
policy by AHP. J Humanit Soc Sci. 2019;10(1):1473–1486.
2020;262:121336. doi:10.1016/j.jclepro.2020.121336
11. Yin D. Citizen network participation: an analytical framework of the
23. Geng D, Saggi K. Optimal price regulations in international pharma-
new political system theory. J Party School Tianjin Municipal
ceutical markets with generic competition. J Health Econ.
Committee Communist Party China. 2010;2:44–52.
2020;71:102315. doi:10.1016/j.jhealeco.2020.102315 021 2
12. Rothwell R, Zegveld W. Reindusdalization and Technology. Logman
24. Gupta M, Bolia NB. Efficiency measurement of Indian high courts - Group Limited; 1985:83–104.
using DEA: a policy perspective. J Policy Model. 2020. un J-
13. Kim A, Jeong SH, Choi HB, Kim HH. An analysis of effects of
25. Andrew Kelly J, Peter Clinch J, Kelleher L, Shahab S. Enabling a just 11
emergency fine dust reduction measures and national petition using
transition: a composite indicator for assessing home-heating energy- on
regression analysis and text mining. KIPS Trans Software Data Eng.
poverty risk and the impact of environmental policy measures. 0 3 2018;7(11):427–434.
Energy Policy. 2020;146:111791. doi:10.1016/j. enpol.2020.111791 0.
14. Gunhee L, Sutherland I, Park S. A study of factors affecting tourists
26. Groh ED, Ziegler A. On self-interested preferences for burden shar- 11.
to Japan using text mining approach. J Tour Leis Res. 2017;29
ing rules: an econometric analysis for the costs of energy policy 23. (11):181–199. measures. Energy Econ. 2018;74:417–426. doi:10.1016/j. 113
15. Chen T, Peng L, Yin X, Rong J, Yang J, Cong G. Analysis of user eneco.2018.06.026 y b /
satisfaction with online education platforms in China during the
27. Valerio C, De Stefano L, Martínez-Muñoz G, Garrido A. A machine om
COVID-19 pandemic. Healthcare. 2020;8(3):200. doi:10.3390/
learning model to assess the ecosystem response to water policy c.s healthcare8030200
measures in the Tagus River Basin (Spain). Sci Total Environ. s
16. Chen T, Peng L, Jing B, Wu C, Yang J, Cong G. The Impact of the
2021;750:141252. doi:10.1016/j.scitotenv.2020.141252 re ep
COVID-19 pandemic on user experience with online education plat-
28. Dukhanina E, Massol O, Lévêque F. Policy measures targeting ov
forms in China. Sustainability. 2020;12(18):7329. doi:10.3390/
a more integrated gas market: impact of a merger of two trading d. . y w su12187329
zones on prices and arbitrage activity in France. Energy Policy. w onl w
17. Bucek J. Fiscal policy for the 21st century: does Barack Obama effect
2019;132:583–593. doi:10.1016/j.enpol.2019.05.044 // e : s s u the real economic policy? 34th International Conference
29. Dong F, Liu Y. Policy evolution and effect evaluation of new-energy p t
Mathematical Methods in Economics (MME 2016). 2016:97–102.
vehicle industry in China. Resour Polic. 2020;67:1–13. ht onal
18. Qi Y, Peng W, Xiong N. The effects of fiscal and tax incentives on
30. Smith JD, Hasan M. Quantitative approaches for the evaluation of rs romf pe
regional innovation capability: text extraction based on python.
implementation research studies. Psychiatry Res. 2019;283:112521. or
Mathematics. 2020;8(7):1193. doi:10.3390/math8071193
doi:10.1016/j.psychres.2019.112521 ded F
19. Sun PJ. XACML policy evaluation optimization research based on
31. Xu Y, You X, Wong Y. Evaluation of tourist satisfaction based on nloa
attribute weighted clustering and statistics reordering. 2017 IEEE
ROSTCM method: a case study of five domestic terrace scenic spot. ow
International Conference on Information and Automation (IEEE
Tour Forum. 2018;11(5):22–34. d y ICIA 2017). 2017:1190–1195.
32. Wang J, Guo Z. Logistic regression model: method and application. olic
20. Pellesova P. Selected econometric methods of optimization of eco-
Beijing: Higher Education Press; 2001:13–26. P
nomic policy. Proceedings of 30th International Conference are
Mathematical Methods in Economics, PTS I AND II. 2012:692–696. hc alt He and nt e nagem a M k is R
Risk Management and Healthcare Policy Dovepress
Publish your work in this journal
Risk Management and Healthcare Policy is an international, peer-
guidelines, expert opinion and commentary, case reports and
reviewed, open access journal focusing on all aspects of public
extended reports. The manuscript management system is completely
health, policy, and preventative measures to promote good health
online and includes a very quick and fair peer-review system, which
and improve morbidity and mortality in the population. The journal
is all easy to use. Visit http://www.dovepress.com/testimonials.php
welcomes submitted papers covering original research, basic
to read real quotes from published authors.
science, clinical & epidemiological studies, reviews and evaluations,
Submit your manuscript here: https://www.dovepress.com/risk-management-and-healthcare-policy-journal

Báo cáo khoa học "A policy Category Analysis Model for Toursim Promotion in China During the Covid-19 Pandemic Based on Data Mining and Binary Regression".

Thông tin :

Tài liệu liên quan:

Trắc nghiệm kinh tế vĩ mô - Đại học Ngoại Thương

Tên đề tài: Tình hình cung thị trường khẩu trang y tế tại Việt Nam khi bị ảnh hưởng bởi dịch bệnh COVID – 19 | Đại học Ngoại thương

Đề kiểm tra giữa kỳ Kinh tế vi mô | Kinh tế vi mô | Đại học Ngoại thương

16 bộ đề trắc nghiệm kinh tế vi mô | Kinh tế vi mô | Đại học Ngoại thương

Chương 1: Kinh tế vi mô | Kinh tế vi mô | Đại học Ngoại thương