Helping users learn about social processes while learning from users : developing a positive feedback in social computing

Advisors: Philippe J. Giabbanelli.Committee members: Reva K. Freedman; Jie Zhou.Includes bibliographical references.Includes illustrations.Social computing is concerned with the interaction of social behavior and computational systems. From its early days, social computing has had two foci. One was the development of technology and interfaces to support online communities. The other was to use computational techniques to study society and assess the expected impact of policies. This thesis seeks to develop systems for social computing, both in the context of online communities and the study of societal processes, that allow users to learn while in turn learning from users. Communities are approached through the problem of Massive Open Online Courses (MOOCs), via a complementary use of network analysis and text mining. In particular, we show that an efficient system can be designed such that instructors do not need to categorize the interactions of all students to assess their learning experience. This thesis explores the study of societal processes by showing how Text Analytics, Visual Analytics, and Fuzzy Cognitive Map (FCM) can collectively help an analyst to understand complex scenarios such as obesity. Overall, this work had two key limitations. One was in the dataset we used, as it was small and didn't show all possible interactions, and the other is in the scalability of our systems. Future work can include the use of non n-gram features to improve our MOOC system, and the use of graph layouts for our visualization system.M.S. (Master of Science