The Student Workshops in Information Retrieval and Language Presents
(http://ciir.cs.umass.edu/~hema/swirl/swirl.htm): 
-------------------------------------------------------------------

Razvan Stefan Bot
New Jersey Institute of Technology
Tuesday, February 15th, 2005
Computer Science Building, Room 151
10:30 AM

Faculty Host: Bruce Croft

Title: Improving Document Representations using Relevance Feedback - The
RFA Algorithm

Abstract: 
In this talk the speaker will present a document representation
improvement algorithm named Relevance Feedback Accumulation (RFA). RFA's
learning function uses the history of relevance assessments to improve
document representations, over time and across users. Document
representation improvements are evaluated across two axes: average
number of index terms per document, and the quality of the index terms.
Results show that: (a) the number of index terms used to represent the
documents affected by the algorithm is reduced, while preserving
retrieval effectiveness; (b) the perceived quality of the index terms
generated by using the RFA algorithm is higher when compared to other
systems. The algorithm is efficient and scalable, suited for retrieval
systems managing larger document collections.

Razvan Stefan Bot is currently a doctoral candidate at New
Jersey Institute of Technology within the Information Systems
Department. He has a BSc in Computer Science from the
Technical University of Cluj-Napoca in Romania, and a MS in Information
Systems from New Jersey Institute of Technology. His research interests
are information retrieval, knowledge management and information systems.
------------------------------------------------------------------