Graduate Student - Ph.D.

Computer Science

University of Massachusetts, Amherst (Amherst, MA)

James Allan

I am interested in understanding how information retrieval frustration (e.g., frustration during a Web search) can be modeled and how those models can help influence retrieval system adaptation to reduce frustration.

Many well-performing retrieval algorithms have developed over the years (e.g., explicit relevance feedback, query expansion and reduction, dependency modeling). However, most are not used because of the necessity of user interaction and/or high processing time, both of which are usually unacceptable in a live search system. One hypothesis is that frustrated users---i.e., searchers who are not finding what they seek---are more apt to use these alternative retrieval models. We then ask: when should which model be presented to the user and how?