Khan, Atif, et al. "Machine learning approach for answer detection in discussion forums: an application of big data analytics." Scientific Programming 2020.1 (2020): 4621196.

This research addresses the challenge of automatically identifying the most relevant answers in online forum threads. By treating the task as a classification problem, the study uses lexical and nonlexical features to classify replies as relevant, partially relevant, or irrelevant. A LinearSVC classifier, enhanced with chi-square and univariate selection techniques, demonstrated superior accuracy in identifying relevant answers compared to other classifiers, based on datasets from Ubuntu and TripAdvisor forums.