Exploring repost features of police-generated microblogs through topic and sentiment analysis
| DOI | https://doi.org/10.1108/EL-02-2019-0044 |
| Date | 05 August 2019 |
| Pages | 607-623 |
| Published date | 05 August 2019 |
| Author | XiaoBo Tang,Shixuan Li,Na Gu,MingLiang Tan |
| Subject Matter | Information & knowledge management,Information & communications technology,Internet |
Exploring repost features of
police-generated microblogs
through topic and
sentiment analysis
XiaoBo Tang,Shixuan Li,Na Gu and MingLiang Tan
Department of Information Management, Wuhan University, Wuhan, China
Abstract
Purpose –This study aims to explore the repostfeatures of microblogs acting to promote the information
diffusionof government-generated content on social media.
Design/methodology/approach –This study proposes a topicsentiment analysis using a mixed
social media analytics framework to analyse the microblogs collected from the Sina Weibo accounts of 30
Chinese provincial police departments.On the basis of this analysis, this study presents the distributionof
repostedmicroblogs and reveals the reposting characteristicsof police-generated microblogs(PGMs).
Findings –The experimentalresults indicate that children’s safety and crime-relatedPGMs with a positive
sentimentcan achieve a high level of online informationdiffusion.
Originality/value –This study is novel,as itreveals the reposting features of PGMs from both a topic and
sentimentperspectives, and provides new findings that can inspire users’repostingbehaviour.
Keywords Sentiment analysis, Social media, Topic modelling, Information diffusion,
Government-generated content, Repost features
Paper type Research paper
1. Introduction
Social media has become an important platform for acquiring and disseminating
information (Stieglitz et al.,2018). Users of social media platforms, such as Twitter,
Facebook and Sina Weibo, can acquire knowledge and news from different domains, share
their opinions about trending topics and build connections with others online (Ozturkcan
et al.,2017). In comparison with traditional mass media, a unique feature of social media is
that it produces a largeamount of user-generated content (Kavanaugh et al.,2012).On social
media platforms, users post messagesthat can be promoted to their followers. If some of the
followers repost the same message,this can then be viewed by several other users, thereby
propagating throughout the platform. In other words, reposting, which also refers to
retweeting or sharing on some platforms,is the main method used on social media platforms
to achieve informationdiffusion (Luo et al., 2017).
Because of the characteristics enabling social media to disseminate information rapidly,
organisations ranging from businesses to political parties have created social media
accounts to spread information and foster interactions (Zhang and Xiao, 2017;Zheng and
Zheng, 2014). However,the potential value of social media has not been completely achieved.
Many government social media accounts receive insufficient attention from the public,
which means that many government-generated online posts failto generate citizens’interest
in reading, commenting on orreposting them. This, therefore, makes it difficult to maintain
efficient online citizen–government communication and participation (Hao et al.,2016;
Topic and
sentiment
analysis
607
Received18 February 2019
Revised18 April 2019
Accepted28 April 2019
TheElectronic Library
Vol.37 No. 4, 2019
pp. 607-623
© Emerald Publishing Limited
0264-0473
DOI 10.1108/EL-02-2019-0044
The current issue and full text archive of this journal is available on Emerald Insight at:
www.emeraldinsight.com/0264-0473.htm
Hong and Nadler, 2012;Strauß et al., 2015). Existing studies have typically explored
government social media-related questions from a single perspective, such as presenting the
social network and opinion leaders of onetrending topic (Choi et al.,2014;Kim and Hastak,
2018), discovering the most popular topics among online political posts (Hofmann et al.,
2013;Bons
on et al., 2015) and revealing the sentiments behind politicians’online comments
(Yaqub et al., 2017;Zavattaro et al.,2015). However, to date, little research has combined
different social media analysismethods to explore how to promote the information diffusion
of government-generated online posts and improve online citizen–government
communication,as well as participation.
This study focuses on the use of social media in the Chinese Government, choosing the
Sina Weibo accounts of police departmentsas research objects, given that Sina Weibo is the
largest social media platform in China (Liu et al., 2018), and because these police
departments’accounts are among the most active politicalWeibo accounts (Lu et al.,2016).
There exist 30 provincial police departments with Weibo accounts; this study collected
content from these accounts for one year (from June 2017 to May 2018) as raw data. To
explore citizens’semantic preferences and motivate them to spread police-generated posts,
thereby engaging in online political participation, this study attempts to answer the
following researchquestions:
RQ1. What are the main topics posted by provincial policedepartments on their Weibo
accounts?
RQ2. What are the sentimentsexpressed in the different topics of the posts?
RQ3. What content can promote the information diffusion of police-generated
microblogs (PGMs)in terms of topic and sentiment?
To contribute to these research objectives, this study applied content analysis to mine
PGMs. To the best of our knowledge, this is the firststudy to explore the repost features of
PGMs by combining the methods of topic and sentimentanalysis. The analysis results show
that repost features, as measuredin relation to topic and sentiment, are meaningful, and that
the findings can provideinsights on social media for the government.
The main contributions of this study are as follows. First, it provides a novel method of
exploring the repost characteristics of PGMs through semantic analysis. Second, it
highlights the discovery that a specific topic and sentiment can trigger a high repost rate.
Finally, the method described in this paper may be seen as straightforward to be
implemented in other governmentagencies.
The rest of this paper is organizedas follows. Section 2 outlines the relevant literature. In
Section 3, the paper explains the study methodology and describes the analysis approaches
in detail, followed by a presentationof the main research results and answers to the research
questions in Section 4. Section 5 provides a discussion of the results. The final section
concludes this study and providessome future research directions.
2. Related works
2.1 Information diffusion and reposting behaviour on social media
Information diffusion can be regarded as the process by which information is spread and
received by individuals via online interactions. There are three essential factors in this
process: sender, receiver and medium(Zafarani et al., 2014). Social media plays a significant
role as the medium of online information diffusion, propagating information at a higher
speed in comparison with other channels (Yoo et al.,2016). Reposting is the main process
EL
37,4
608
To continue reading
Request your trial