Roles and Success in Wikipedia Talk Pages: Identifying Latent Patterns of Behavior

Proceedings of the The 8th International Joint Conference on Natural Language Processing, pages 1026–1035, Taipei, Taiwan, November 27 – December 1, 2017 c⃝2017 AFNLP Roles and Success in Wikipedia Talk Pages: Identifying Latent Patterns of Behavior Keith Maki, Michael Miller Yoder, Yohan Jo and Carolyn Penstein Ros´e Language Technologies Institute Carnegie Mellon University Pittsburgh, PA, USA {kmaki,yoder,yohanj,cprose}@cs.cmu.edu Abstract In this work we investigate how role-based behavior profiles of a Wikipedia editor, considered against the backdrop of roles taken up by other editors in discussions, predict the success of the editor at achiev- ing an impact on the associated article. We first contribute a new public dataset in- cluding a task predicting the success of Wikipedia editors involved in discussion, measured by an operationalization of the lasting impact of their edits in the article. We then propose a probabilistic graphical model that advances earlier work induc- ing latent discussion roles using the light supervision of success in the negotiation task. We evaluate the performance of the model and interpret findings of roles and group configurations that lead to certain outcomes on Wikipedia. 1 Introduction In this paper we explore the discussion strategies and configurations of conversational roles that al- low Wikipedia editors to influence the content of articles. In so doing, we contribute both a new public dataset and proposed model that advance work towards induction of latent conversational roles using light supervision. Online production communities like Wikipedia, an online encyclopedia which anyone can edit, have the potential to bring disparate perspectives together in producing a valuable public resource. Individual Wikipedia editors unavoidably carry their own perspectives; these voices can explic- itly or subtly influence the jointly produced article content even when editors strive for neutrality1. 1https://en.wikipedia.org/wiki/ Wikipedia:Neutral_point_of_view This work explores the interaction between indi- vidual editors and the collaborative process that supervises the development of a Wikipedia article. Wikipedia editors discuss article improvements, coordinate work and resolve disagreements on talk pages associated with each article (Ferschke, 2014). Pairing talk page discussions with simulta- neous edits in shared content, we introduce a task predicting the success of a particular editor’s arti- cle edits based on the corresponding discussion. We propose a lightly supervised probabilistic graphical model of discussion roles and behav- iors that offers advances over the prior discus- sion role modeling work of Yang et al. (2015), which employs a more restricted conceptualiza- tion of role taking. While the earlier model only allowed each role to be played by one editor, our extended model learns a distribution over roles for each editor. Furthermore, it can assign roles to an arbitrary number of editors rather than being re- stricted to a specific number. This model allows the interpretation of config- urations of roles that are conducive or detrimental to the success of individual editors. We find that the greatest success is achieved by detail-oriented editors working in cooperation with editors who play more abstract organizational roles. 2 Related Work This work investigates influence in discussion as part of the collaborative editing …

Content truncated to 500 words