Prediction for the Newsroom: Which Articles Will Get the Most Comments?

Abstract

The overwhelming success of the Web and mobile technologies has enabled millions to share their opinions publicly at any time. But the same success also endangers this freedom of speech due to closing down of participatory sites misused by individuals or interest groups. We propose to support manual moderation by proactively drawing the attention of our moderators to article discussions that most likely need their intervention. To this end, we predict which articles will receive a high number of comments. In contrast to existing work, we enrich the article with metadata, extract semantic and linguistic features, and exploit annotated data from a foreign language corpus. Our logistic regression model improves F1-scores by over 80% in comparison to state-of-the-art approaches.

Top Comment or Flop Comment? Predicting and Explaining User Engagement in Online News Discussions

“…A growing body of research aims to foster respectful and fruitful discussions on the Web. Applications of this research manifest in real-world system implementations that support moderators and community managers, for example, by predicting how many comments a news article will receive (Ambroselli et al 2018), identifying comments that require moderation (Schabus and Skowron 2018;Risch and Krestel 2018) or highlighting comments that are worth reading (Park et al 2016). To this end, there are two primary directions of related work on comment classification: identifying either toxic or high-quality comments.…”

“…Chronological ranking in discussion threads is an essential difference in news comments compared to, for example, posts on Twitter or Facebook that can stand alone without a conversational context. In contrast to related work that predicts the popularity of a news article and the number of received user comments (Ambroselli et al 2018), we predict the users' interactions with a comment. To this end, we neglect the news article text and focus on the comment text, upvotes, and replies.…”

A Dataset of Journalists' Interactions with Their Readership

“…Forecasting the popularity of news articles helps moderation teams to schedule their workload [2,33]. Popularity is measured in terms of the expected number of received comments because content moderators at the considered platforms need to check each and every comment for toxic content.…”