Matching far-right voices: A computational exploration of user identity linkage strategies across social media platforms

Far-right actors use multiple platforms to communicate, mobilize, and recruit. This observation aligned with broader transformations in digital information media environments captured by notions such as the 'hybrid media system' or 'networked public sphere'. Still, most studies on the digital far right focus on single platforms, thereby not taking into account how the same actors use various platforms. Establishing a framework to link actors across platforms carries the potential to understand how the same actors form cross-platform communities by exploiting different platform’s affordances. One existing framework that offers a solution to that challenge is User-Identity-Linkage (UIL). Developed in the field of computer science, UIL aims to identify social media accounts belonging to the same user—natural persons or organizations–enabling researchers to conduct comparative and cross-platform studies on the actor level (Chen & Chen, 2022). Thus, while largely unknown in political and communication science, this approach can accommodate moves from single- to more robust multi-platform studies. Individual features of UIL, such as measures of textual similarity or shared URLs, are already familiar to communication science researchers, making UIL a scalable and efficient method that might advance and reduce the manual labor of multi-platform research. Given the multiple different approaches to performing UIL (Shu et al., 2017), we ask (RQ1): How well do UIL approaches from computer science can be successfully applied to link user accounts by far-right political actors across platforms? With far-right as an umbrella term for the radical and extreme right, our case contains heterogeneous actor types, e.g., politicians, journalists, or organizations (Pirro, 2023). The social media usage of these actors likely varies in terms of activity level, sharing patterns, and messaging (use of visuals, texts, hyperlinks). Reflecting on these differences, we ask (RQ2): Which UIL features work better to link which actor types? Lastly, we turn towards a possible application of UIL in multi-platform research. Based on different platform affordances, we expect far-right actors on Twitter (now X) to highlight more person-centric antagonists such as individual politicians or other elites that can be tagged. In contrast, their Facebook posts should contain more generalized out-group statements e.g., antisemitic or Islamophobic comments (Puschmann et al., 2022). We ask (RQ3): How do far-right actors’ antagonists differ on Facebook and Twitter? To address our research questions, we evaluate the performance of different UIL approaches to link far-right actors across different platforms, namely Twitter and Facebook. We use data from three iteration snowball samples on both platforms from April to September 2022. The seed lists for these samples consist of a matched sample of around 40 German far-right politicians, activists, and right-wing alternative media. Data collection resulted in approximately 1,000 Facebook pages and 40,000 Twitter accounts. For ethical and privacy reasons, we focus exclusively on actors of public interest (e.g., politicians, parties, and media). Our main objective is to integrate UIL into research on political communication, showcasing its potential by evaluating the performances of different features and providing a comparative application that studies the strategies of far-right actors in mentioning antagonists.