Personalizing Search via Automated Analysis of Interests and Activities

Jaime Teevan, Susan T. Dumais, Eric Horvitz

Access pdf file.

Abstract:

We formulate and study search algorithms that consider a userís prior interactions with a wide variety of content to personalize that userís current Web search. Rather than relying on the unrealistic assumption that people will precisely specify their intent when searching, we pursue techniques that leverage implicit information about the userís interests. This information is used to re-rank Web search results within a relevance feedback framework. We explore rich models of user interests, built from both search-related information, such as previously issued queries and previously visited Web pages, and other information about the user such as documents and email the user has read and created. Our research suggests that rich representations of the user and the corpus are important for personalization, but that it is possible to approximate these representations and provide efficient client-side algorithms for personalizing search. We show that such personalization algorithms can significantly improve on current Web search.

Keywords: Personalized search, personalization, relevance feedback.

In: Proceedings of 28th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, August 2005. ACM Press.

Author Email: teevan@mit.edu,sdumais@microsoft.com, horvitz@microsoft.com