This data mining project investigates the changing and contested narratives of the BRI in social media, focusing on sentimental and networking characteristics in online communication platforms. In an interim outcome, this work has firstly framed a systematic approach to analyse the discourse on the BRI in social networking platforms; with further studies to focus on conversations related to the BRI’s transnational circulation of personnel, religion and cosmological values.
The targeted data period spans seven years between 1st Jan 2013 to 31st April 2020, exemplified by the worldwide social media platform Twitter. This work has retrieved more than 400,000 tweets with BRI related keywords mentioned in content or hashtags, applying the Boolean logic “OR” for following terms in Twitter’s advanced search mode: Belt and road, one belt one road, New silk road, silk road economic belt, maritime silk road, OBOR. Tweets are then documented with the name of the user, content, hyperlink, posted time, number of likes, retweets, replies, sentiment level of tweet, and geographical location provided by the user.
The methodology for data mining and sentiment analysis is based on the Python programming language with data scraping module to retrieve tweets from Twitter and VADER sentiment module (Hutto & Gilbert, 2014) to further classify the emotional level of each tweet within the range of -1 to +1 from most negative to most positive. The search language of terms is in English, which can be further expanded to other languages such as Chinese, Arab, Urdu, to cover a wider context of the BRI in different languages.
Until now, this sub-project has worked on five aspects of the BRI discourse in the Twittersphere:
- Infrastructural and religious related postings
- Trending topics/conversations of the BRI at different times
- Sentimental dynamics of the BRI’s dialogue in global and local frames
- Key opinion leaders and their networking characteristics
- Comparative studies on the opinion field: China, United States, Pakistan, Saudi Arabia, Hong Kong
(1). Postings about the BRI with terms related to religions, sects or sacred scriptures in Twitter reveal the general focuses on religious issues relating to the BRI.
(2). Topic modelling was employed to summarize tweets of the BRI from 2013-2020 into ten topics. which are generated by the Latent Dirichlet Allocation (LDA) algorithm in Python topic modelling coding (Blei et al., 2003).
(3). Visualizing the sentiment level of tweets on the BRI in monthly average.
(4). Ranking and networking the top 20 opinion leaders in a given year and in different countries by the number of retweets. Accounts include state media, research institutes, think tanks, news organizations, and individuals
(5). Comparison between users’ postings on the BRI in different countries.
The Belt and Road Initiative in Social Media: Investigating the Discursive Landscape of International Relations through Sentiment and Social Network Analysis
Edward Man, Qian Junxi and David A. Palmer
Since the Belt and Road Initiative (BRI) was launched in 2013, diverse arguments and sentiments are growing online, concerning a promising but latently coercive Sino-centric order in future. In 2020, the discourse of the BRI has been prominently associated with Sinophobic opinions, correlated to the deterioration in Sino-American relations and suspicious transmission of COVID-19 through China backed overseas infrastructures. This brings into question how the shifting sentiments are driven by stakeholders’ engagements in dissemination of information online. This article quantified the communal cyberspace of the BRI as found on the social media platform, Twitter since 2013 to the third quarter of 2020. Text mining analytics are employed to spotlight the evolving discourse on the BRI in trending topics and sentiments; and to build upon the shifts in narrative to investigate Key Opinion Leaders (KOLs)’ online behaviours and influence by mapping out their social networking patterns. Using computational approaches to identify the evolution of the BRI narrative on social media, this work offers a vantage point to understand the online public opinion field in temporal dynamics and how different actors behave in reacting and adapting to changes in international affairs.