Chaturvedi, Snigdha, Dan Goldwasser, and Hal Daumé III. "Predicting instructor’s intervention in mooc forums." Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2014.

This paper addresses the challenge of predicting instructor interventions in student discussion forums within Massive Open Online Courses (MOOCs), where personalized interaction is limited. The authors propose several prediction models that incorporate course information, forum structure, and post content to capture unique aspects of MOOCs. These models use latent categories to abstract the content of individual posts in threads, which are learned jointly with the binary intervention prediction problem. Experiments using data from two Coursera MOOCs demonstrate that including thread structure in the learning process improves predictive performance.