Characterizing an online, science-based affinity space using topic modelling, diversity indices, and social network analysis

Abstract

Characterizing who participates in and contributes to science communication efforts can extend our understanding of the science education ecosystem, including giving insight into what kinds of communication works for whom and under what conditions. We describe methods for characterizing an online, scientific affinity space using @TimeScavengers as our exemplar. @TimeScavengers is an educational and science communication effort centered on geosciences, and uses varied online platforms, including Twitter (now known as X). We applied an established framework to describe members who connected with @TimeScavengers’ Twitter (n = 1113) and analyzed data (e.g. tweets, replies, mentions, and re-tweets) (n = 6538) collected from the community for an annual cycle of activity (January 2020–2021) via social network analysis, topic modeling, and diversity indices. Social network analysis showed a highly dispersed network with scientists in control of information flow. Topic modeling of tweets with original content (n = 855) generated six topics related to collaboration, connection, and scientific outreach. Application of diversity indices indicated that scientists and education and outreach members were most prevalent across topics. Using these methods uncovers deeper understanding of the community and interactions within the space, which can lead to the development of better science communication efforts in the digital realm.

Keywords

• Machine learning
• science communication
• science outreach
• InfoMaps
• Netlytic
• NodeXL