Document Type


Publication Date



Most search engines, indispensable tools for finding information on the Web, do not take advantage of a user's personal preferences in creating result sets from search queries. In particular, collaborative filtering, an effective personalization technique that uses peer opinions to recommend items of interest, has not been widely used in Web search engines nor have the benefits of collaborative filtering to search engine technology been thoroughly evaluated. We have designed and implemented a search engine called Foible that personalizes Web searches based on user preferences and uses collaborative filtering to enhance the result sets returned from user queries. Through a carefully designed user study, we evaluate the effectiveness of Web search with personalization results and collaborative filtering compared with a traditional Web search engine. We find Web search results based on personalization and collaborative filtering provides result sets more closely related to user interests that result sets returned by traditional search engines. Moreover, users overwhelmingly prefer results returned by a personalized filter with collaborative filtering to those returned by traditional search engines.