A comparison of three methods to determine the subject matter in textual data.

This study compares three different methods commonly employed for the determination and interpretation of the subject matter of large corpuses of textual data. The methods reviewed are: (1) topic modeling, (2) community or group detection, and (3) cluster analysis of semantic networks. Two different datasets related to health topics were gathered from Twitter posts to compare the methods. The first dataset includes 16,138 original tweets concerning HIV pre-exposure prophylaxis (PrEP) from April 3, 2019 to April 3, 2020. The second dataset is comprised of 12,613 tweets about childhood vaccination from July 1, 2018 to October 15, 2018. Our findings suggest that the separate "topics" suggested by semantic networks (community detection) and/or cluster analysis (Ward's method) are more clearly identified than the topic modeling results. Topic modeling produced more subjects, but these tended to overlap. This study offers a better understanding of how results may vary based on method to determine subject matter chosen.

Comparison of public discussions of gene editing on social media between the United States and China.

The world's first gene-edited babies event has stirred controversy on social media over the use of gene editing technology. Understanding public discussions about this controversy will provide important insights about opinions of science and facilitate informed policy decisions. This study compares public discussion topics about gene editing on Twitter and Weibo, as wel asthe evolution of these topics over four months. Latent Dirichlet allocation (LDA) was used to generate topics for 11,244 Weibo posts and 57,525 tweets from September 25, 2018, to January 25, 2019. Results showed a difference between the topics on Twitter versus Weibo: there were more nuanced discussions on Twitter, and the discussed topics between platforms focused on different areas. Temporal analysis showed that most discussions took place around gene-edited events. Based on our findings, suggestions were provided for policymakers and science communication practitioners to develop more effective communication strategies toward audiences in China and the U.S.

Microblog credibility indicators regarding misinformation of genetically modified food on Weibo.

The considerable amount of misinformation on social media regarding genetically modified (GM) food will not only hinder public understanding but also mislead the public to make unreasoned decisions. This study discovered a new mechanism of misinformation diffusion in the case of GM food and applied a framework of supervised machine learning to identify effective credibility indicators for the misinformation prediction of GM food. Main indicators are proposed, including user identities involved in spreading information, linguistic styles, and propagation dynamics. Results show that linguistic styles, including sentiment and topics, have the dominant predictive power. In addition, among the user identities, engagement, and extroversion are effective predictors, while reputation has almost no predictive power in this study. Finally, we provide strategies that readers should be aware of when assessing the credibility of online posts and suggest improvements that Weibo can use to avoid rumormongering and enhance the science communication of GM food.

From Network Positions to Language Use: Understanding the Effects of Brokerage and Closure Structures from a Linguistic Perspective.

We investigated cancer survivors' interactions on an online breast cancer support forum, focusing on how the network structures of brokerage and closure relate to the types of support received and to the language used in posts. Data came through the extraction of 1,443 forum members' online networks. Automated linguistic analysis was carried out on the 27,248 threads these survivors made and the 336,151 replies they received. Survivors' brokerage and closure levels were positively correlated with the use of positive affective words in their posts, a linguistic marker of well-being. Different network positions fostered different types of support in the community. Specifically, people bridging unconnected users (the broker role) were more likely to receive informational support whereas people in closely knit groups (the closure role) were more likely to receive emotional support. Theoretical, methodological, and practical implications are examined.

Semantic Network Analysis Reveals Opposing Online Representations of the Search Term "GMO".

Making sound food and agriculture decisions is important for global society and the environment. Experts tend to view crop genetic engineering, a technology that can improve yields and minimize impacts on the environment, more favorably than the public. Because there is a causal relationship between public opinion and public policy, it is important to understand how opinions about genetically engineered (GE) crops are influenced. The public increasingly seeks science information on the Internet. Here, semantic network analysis is performed to characterize the presentation of the term "GMO (genetically modified organism)," a proxy for food developed from GE crops, on the web. Texts from three sources are analyzed: U.S. federal websites, top pages from a Google search, and online news titles. We found that the framing and sentiment (positive, neutral, or negative attitudes) of "GMO" varies across these sources. It is described how differences in the portrayal of GE food by each source might affect public opinion. A current understanding of the types of information individuals may encounter online can provide insight into public opinion toward GE food. In turn, this knowledge can guide teaching and communication efforts by the scientific community to promote informed decision-making about agricultural biotechnologies.

Exploring the presentation of HPV information online: A semantic network analysis of websites.

CONTEXT: Negative vaccination-related information online leads some to opt out of recommended vaccinations. OBJECTIVE: To determine how HPV vaccine information is presented online and what concepts co-occur. METHODS: A semantic network analysis of the words in first-page Google search results was conducted using three negative, three neutral, and three positive search terms for 10 base concepts such as HPV vaccine, and HPV immunizations. In total, 223 of the 300 websites retrieved met inclusion requirements. Website information was analyzed using network statistics to determine what words most frequently appear, which words co-occur, and the sentiment of the words. RESULTS: High levels of word interconnectivity were found suggesting a rich set of semantic links and a very integrated set of concepts. Limited number of words held centrality indicating limited concept prominence. This dense network signifies concepts that are well connected. Negative words were most prevalent and were associated with describing the HPV vaccine's side-effects as well as the negative effects of HPV and cervical cancer. A smaller cluster focuses on reporting negative vaccine side-effects. Clustering shows the words women and girls closely located to the words sexually, virus, and infection. DISCUSSION: Information about the HPV vaccine online centered on a limited number of concepts. HPV vaccine benefits as well as the risks of HPV, including severity and susceptibility, were centrally presented. Word cluster results imply that HPV vaccine information for women and girls is discussed in more sexual terms than for men and boys.