Anton Leuski
Center for Intelligence Information Retrieval
University of Massachusetts
Amherst, MA , USA

Interactive Information Retrieval: Combining Ranked List and Clustering

Abstract:

Locating interesting information on the World Wide Web is the main task of on-line search engines. Such an engine accepts a query from a user and responds with a ranked list of web pages ordered by their likelihood of being relevant to the user's request. The user is expected to follow the ranked list while examining the retrieved documents. In practice, browsing the ranked list is rather tedious and often unproductive. In the talk I present a study of an alternative document organization technique which can help the user to find the relevant information in the retrieved data much more quickly than the ranked list.

The performance of an interactive system is determined by both the system components and the user of the system. We introduce a novel evaluation approach, called the Strategy Based Evaluation (SBE) framework. The SBE method is based in part on modeling the system-user interaction. It allows us to separate the user's effect on the overall performance from the system qualities and study document organization as an information retrieval problem.

The SBE method is applied to a visualization approach that presents documents as spheres in space arranged in proportion to the inter-document similarity. We show that a projection of the high-dimensional document representations on 2 and 3 dimensions can be accurate enough for the retrieval task. A small user study reveals that the users' can navigate the visualization effectively. We define the problem of browsing the retrieved documents in terms of reinforcement learning and develop and effective browsing strategy after a modest amount of training.

Speaker info:

Anton Leuski is a post-doctoral fellow at the Center for Intelligent Information Retrieval at the University of Massachusetts at Amherst. He received his M.S. with distinction in Physics from the Moscow State University in 1993, followed by M.S. and Ph.D. degrees in Computer Science from the University of Massachusetts at Amherst in 1995 and 2001 respectively. His Ph.D. work with James Allan and Bruce Croft focused on interactive information organization, retrieval, and visualization. His current research interests include interactive information access, human-computer interaction and machine learning. he is presently working on interactive news topic tracking and visualization of concept hierarchies. More information is available at http://www-ciir.cs.umass.edu/~leouski/.

Click here to view Anton Leuski's Schedule