Modeling the Dynamics of Online Learning Activity

15 Nov
Tuesday, 11/15/2016 4:00pm to 5:00pm
Computer Science Building, Room 150/151
Data Science Tea


Abstract: People are increasingly relying on social media and the Web to find solutions to their problems in a wide range of domains. Closely related problems often lead people to the same characteristic learning pattern, which is expressed by visiting related pieces of information, performing almost identical sequences of queries or, more generally, taking a series of similar actions. In this talk, we introduce a novel modeling framework for clustering continuous-time grouped streaming data, the Hierarchical Dirichlet Hawkes process (HDHP). This model allows us to automatically uncover a wide variety of learning patterns from detailed traces of online learning activity. Experiments on real data gathered from Stack Overflow reveal that our framework can recover meaningful learning patterns in terms of both content and temporal dynamics, as well as track users' interests and goals over time. This is a joint work with Isabel Valera and Manuel Gomez Rodriguez.


Bio: Harry is a PhD student at the Department of Computer Science at Boston University. There, he is member of the Data Management Lab, advised by Evimaria Terzi and George Kollios. His research focuses on Graph Mining, Algorithms on Big Data, Machine Learning, and Database Security.