Rank Optimization of Personalized Search

Augmenting the global ranking based on the linkage structure of the Web is one of the popular approaches in data engineering community today for enhancing the search and ranking quality of Web information systems. This is typically done through automated learning of user interests and re-ranking of search results through semantic based personalization. In this paper, we propose a query context window (QCW) based framework for Selective uTilization of search history in personalized leArning and re-Ranking (STAR). We conduct extensive experiments to compare our STAR approach with the popular directory-based search methods (e.g., Google Directory search) and the general model of most existing re-ranking schemes of personalized search. Our experimental results show that the proposed STAR framework can effectively capture user-specific query-dependent personalization and improve the accuracy of personalized search over existing approaches.