Linguistically guided community discovery. Issue 1 (May 2019)

• Record Type: Journal Article
• Authors: Gawron, Jean M; Dodge, Alex; Tsou, Ming-Hsiang; Spitzberg, Brian; An, Li
• Is Part Of: Big data & society. Volume 6:Issue 1(2019)
• Publication Date: 2019-05
• Subjects: Community discovery -- Gamergate -- linguistic features -- Puppygate -- social networks -- unsupervised clustering; Big data -- Social aspects -- Periodicals; Social sciences -- Research -- Data processing -- Periodicals; Social sciences -- Research -- Methodology -- Periodicals; Data mining -- Periodicals; 300.28557
• DOI: 10.1177/2053951719846634
• Languages: English
• ISSNs: 2053-9517

Abstract

Within some online communities, discussion often centers on issues on which writers take sides, and within some subset of those debate-prone communities, we find over time that particular sets of writers almost always end up on the same side of an issue. These sets we call factions. In this paper, we describe a tool to perform what we call faction discovery on online communities. Generalizing methods developed in the bibliometrics and information retrieval literature, we define a network determined by similarities of content in a community of users and add in direct evidence of online ties between users (e.g., link information such as mention-links). We then perform community detection on the network to find factions. Using a set of data collected from science and fantasy blogs, we show that the discovered factions accurately reflect an active conflict in the community leading to significant, politically related social fracture.