Sentiment Analysis across the Courses of a MOOC Specialization

A Massive Open Online Course (MOOC) is an effective way for a university to deliver course content that reaches a global audience. Such developments are not without substantial costs and risks [1]. On the Coursera platform, there are MOOC specializations, which package a sequence of related courses. Retaining learners is of particular interest to instructors as they progress through the courses of a long specialization. Learners that encounter delays, unfairness, or plagiarism in peer evaluations of assignments could become dissatisfied enough to withdraw, for example. We describe using trends in sentiment analysis of discussion forum postings across the courses. The idea is to detect, interpret, and address points of lower sentiment in a MOOC specialization, to avoid losing learners. We outline our findings with a MOOC specialization on software product management, which consists of six courses, involving a nominal 24 weeks of content [2,3]. Interestingly, higher sentiment in a course's content did not necessarily preserve enrollment numbers for the subsequent course.