Wikimedia Research/Showcase/Archive/2018/06

June 2018

18 June 2018 Video: YouTube

Conversations Gone Awry. Detecting Early Signs of Conversational Failure

slides

By Justine Zhang and Jonathan Chang, Cornell University

One of the main challenges online social systems face is the prevalence of antisocial behavior, such as harassment and personal attacks. In this work, we introduce the task of predicting from the very start of a conversation whether it will get out of hand. As opposed to detecting undesirable behavior after the fact, this task aims to enable early, actionable prediction at a time when the conversation might still be salvaged. To this end, we develop a framework for capturing pragmatic devices—such as politeness strategies and rhetorical prompts—used to start a conversation, and analyze their relation to its future trajectory. Applying this framework in a controlled setting, we demonstrate the feasibility of detecting early warning signs of antisocial behavior in online discussions.

Building a rich conversation corpus from Wikipedia Talk pages

By TBA

We present a corpus of conversations that encompasses the complete history of interactions between contributors to English Wikipedia's Talk Pages. This captures a new view of these interactions by containing not only the final form of each conversation but also detailed information on all the actions that led to it: new comments, as well as modifications, deletions and restorations. This level of detail supports new research questions pertaining to the process (and challenges) of large-scale online collaboration. As an example, we present a small study of removed comments highlighting that contributors successfully take action on more toxic behavior than was previously estimated.