Replication Data for Phrase-level Pairwise Topic Modeling to Uncover Helpful Peer Responses to Online Suicidal Crises

This dataset contains a set of threads in Reddit r/SuicideWatch. The dataset can be used to understand the peer perceived helpfulness. The task is to discover and analyze the topics in original user's posts and peer responses. We use the dataset to validate our proposed pairwise LDA model for topic discovery.