Personalized Web Search Based on Dynamic User Profile

Yuan Xiaojie,Dou Zhicheng,Liu Fang,Zhang Lu
2008-01-01
Abstract:Personalized Web search is an effective way to solve the problem of query ambiguity because it is able to return different search results to different users based on their preferences. Because user search history involves much noisy information that is irrelevant to user interests, static user profile built upon all historical user queries is usually inaccurate and may harm search performance sometimes. To solve this problem, this paper proposes a personalized Web search algorithm based on a dynamic user profile. When generating this user profile, historical queries that are related to the current query are given high weights and other unrelated queries are ignored. Experimental results show that personalized search algorithm which uses dynamic user profile outperforms the one which uses static user profile, and could improve search performance significantly.
What problem does this paper attempt to address?