Probabilistic models for personalizing web search

David Sontag,Kevyn Collins-Thompson,Paul N. Bennett,Ryen W. White,Susan Dumais,Bodo Billerbeck

DOI: https://doi.org/10.1145/2124295.2124348

2012-01-01

Abstract:We present a new approach for personalizing Web search results to a specific user. Ranking functions for Web search engines are typically trained by machine learning algorithms using either direct human relevance judgments or indirect judgments obtained from click-through data from millions of users. The rankings are thus optimized to this generic population of users, not to any specific user. We propose a generative model of relevance which can be used to infer the relevance of a document to a specific user for a search query. The user-specific parameters of this generative model constitute a compact user profile. We show how to learn these profiles from a user's long-term search history. Our algorithm for computing the personalized ranking is simple and has little computational overhead. We evaluate our personalization approach using historical search data from thousands of users of a major Web search engine. Our findings demonstrate gains in retrieval performance for queries with high ambiguity, with particularly large improvements for acronym queries.