Jaime Arguello will present, "Aggregated Search: Motivations, Methods and Milestones" on Tuesday, January 11th at 2:30 p.m. in room 208 Manning Hall.
In its early stages, information retrieval (IR) research focused on retrieving full-text documents within a single collection in response to a user's query. Since then, however, IR tasks have become increasingly more diverse and specialized. There are two trends driving this change. The first is the growing diversity of data available for search in different environments. The second is the ubiquity of computers and the diversity of information-seeking tasks that occur in daily life. A common finding in empirical IR research is that different searchable media and different retrieval tasks require customized solutions. This gives rise to a new challenge: how do we provide users with integrated access to all these diverse search services within a single search interface? This is the goal of aggregated search.
In this talk, Arguello will highlight his work on aggregated search within the context of Web search. In addition to retrieving Web pages, commercial search engines also function as a single point of access to many back-end search services (called verticals) that focus, for example, on news, blogs, images, local business listings, items for sale, driving directions and weather forecasts. He will first describe methods for automatically predicting that a vertical is relevant to a query and should be presented to the user. He will then discuss challenges that occur in aggregated search evaluation and describe a new methodology for evaluating aggregated search.
About Jaime Arguello
Jaime Arguello is a Ph.D. candidate at the Language Technologies Institute within the School of Computer Science at Carnegie Mellon University. His research focuses on information retrieval (IR). In his thesis work, he investigates aggregated search---the task of providing integrated access to multiple services within a single search interface. Prior to this, he conducted research in blog retrieval, information extraction, topic-segmentation of transcribed dialogue, and on developing visualizations and interactions for corpus exploration. In addition to IR, Jaime is interested in machine-learning, human-computer interaction, and text-mining. He has published on these subjects at conference such as SIGIR, CIKM, ECIR, TREC, CHI, ICWSM, HLT, and DG.O. He was a recipient of the SIGIR 2009 Best Paper Award and the 2009 Yahoo! Key Scientific Challenges Program Award.
