Robust and Web Retrieval with Document-Centric Integral Impacts

Vo Ngoc Anh
Department of Computer Science and Software Engineering, The University of Melbourne, Victoria 3010, Australia.

Alistair Moffat
Department of Computer Science and Software Engineering, The University of Melbourne, Victoria 3010, Australia.

Status

To appear in the 2003 TREC Notebook Proceedings.

Abstract

This paper reports the experiments done at The University of Melbourne for the Robust and Web tracks of TREC-2003. We explore the idea of determining the impact of a term locally within the document and in a qualitative manner instead of a quantitative one. The impact of each distinct term in a document or query text is defined to be a small integer. The scalar product of the impact vector for a document and the impact vector for a query is taken to be the similarity score between them, an arrangement that allows very fast query evaluation.