Personalizing XML text search in PIMENT

  • Authors:
  • Sihem Amer-Yahia;Irini Fundulaki;Prateek Jain;Laks Lakshmanan

  • Affiliations:
  • AT&T Labs Research;Bell Labs;Bell Labs;University of British Columbia, Canada

  • Venue:
  • VLDB '05 Proceedings of the 31st international conference on Very large data bases
  • Year:
  • 2005

Quantified Score

Hi-index 0.00

Visualization

Abstract

A growing number of text-rich XML repositories are being made available. As a result, more efforts have been deployed to provide XML full-text search that combines querying structure with complex conditions on text ranging from simple keyword search to sophisticated proximity search composed with stemming and thesaurus. However, one of the key challenges in full-text search is to match users' expectations and determine the most relevant answers to a full-text query. In this context, we propose query personalization as a way to take user profiles into account in order to customize query answers based on individual users' needs.We present PIMENT, a system that enables query personalization by query rewriting and answer ranking. PIMENT is composed of a profile repository that stores user profiles, a query customizer that rewrites user queries based on user profiles and, a ranking module to rank query answers.