Invited Talk and Panel Discussion, the World Conference of Science Journalists, Lausanne, Switzerland
One of the key challenges facing online communities, from social networks to the comment sections of news sites, is low quality contributions. To help with this, comment moderators are often employed to identify the most informative and constructive comments, and shield readers from low-quality or abusive content. For example, The New York Times employs a staff of full-time moderators to review comments. Exemplary comments representing a range of views are highlighted and tagged as NYT Picks. With vast number of online comments, and growing challenges of how social networks manage toxic language, the role of moderators is becoming much more demanding. There is thus a growing interest in developing automation to help filter and organize online comments for both moderators and readers. We are developing computational methods to identify “constructive” comments automatically. In particular, we have created an annotated corpus of constructive comments and we have been developing feature-based and deep-learning methods to identify constructive comments posted on opinion articles and editorials. Our models achieve up to 84% accuracy.