BibTeX
@MISC{_,
author = {},
title = {},
year = {}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
This paper describes the method we used for the Novelty Track for the 2002 Text Retrieval Conference (TREC). We tried to adapt tools we are developing for a task closely related to the novelty part of the this track. The system we are building will scan a stream of documents and present to the user only the new information it finds. For the “relevance ” part of the TREC, we decided to test the applicability of some of these tools. Since information retrieval is not a focus of our research, we thought it would be more interesting to use something new rather than try to hurriedly catch up. The results were far from satisfactory, but it is clear from the overall results that novelty detection remains a difficult and unsolved problem.
