Ontology-driven personalized query refinement

  • Authors:

  • Sofia Stamou;Lefteris Kozanidis;Paraskevi Tzekou;Nikos Zotos

  • Affiliations:

  • Computer Engineering and Informatics Department, Patras University, Greece;Computer Engineering and Informatics Department, Patras University, Greece;Computer Engineering and Informatics Department, Patras University, Greece;Computer Engineering and Informatics Department, Patras University, Greece

  • Venue:
  • Journal of Web Engineering
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

The most popular way for finding information on the Web is go to a search engine, submit a query that describes an information need and receive a list of results that relate to the information sought. As more and more topics are being discussed over the Web and our vocabulary remains relatively stable, it is increasingly difficult for Web users to select queries that express their varying information needs in a distinguishable by the engine manner. Query refinement is the process of providing information seekers with alternative wordings for expressing their search intentions. Although refined queries may contribute to the improvement of retrieval results, nevertheless their realization is intrinsically limited in that they consider nothing about the preferences of the user issuing that query. One way to go about selecting suitable query alternatives is to account for the user interests in the query refinement process. This task involves two great challenges. First we need to be able to effectively identify the user preferences and build a profile for every user. Second, once such a profile is available, we need to identify among a set of candidate query alternatives those that match the user interests. In this article, we present our work towards a personalized query refinement technique and we discuss how we address both of these challenges. Since Web users are reluctant to provide explicit information on their personal preferences, for the first challenge we attempt to determine them based on the analysis of the users' click history. In particular, we leverage a topical ontology for estimating the user's topic preferences based on her past searches. For the second challenge, we have developed a query refinement mechanism that uses the learnt user preferences in order to disambiguate the user's current query and thereafter identify alternative query wordings that match both the initial query semantics and the user preferences. Our experiments show that user preferences can be learnt accurately through the use of the topical ontology and refined queries based on the user preferences yield significant improvements in the search quality over existing query improvement techniques.