User Preference Modeling from Positive Contents for Personalized Recommendation

摘要: With the spread of Web, users can obtain a wide variety information, and also access novel content in real time. In this environment, finding useful information from huge amount available becomes time consuming process. paper, we focus on user modeling for personalization to recommend relevant interests. Techniques used association rules deriving profiles are exploited discovering meaningful patterns users. Each preference is presented frequent term patterns, collectively called PTP (Personalized Term Pattern) terms, PT Term). addition, content-based filtering approach employed corresponding with preferences. order evaluate performance proposed method, compare experimental results those probabilistic learning model vector space model. The evaluation NSF research award datasets demonstrates that method brings significant advantages terms improving recommendation quality comparison other methods.