Based on the features of the forest products trading information, a compound keywords vector space model was utilized to represent the user interest model of the forest products trading information website. Every compound keyword in the vector space consists of three keywords, which are supply and demand classification, the name of the forest product and the production place, and some information about the specifications, price and the company name of the forest products. Every compound keyword has a user interest value. The learning and updating of the user interest model were reaiized through the browsing, registering and releasing behaviors of the users and introducing the forgetting factors for the interest values, respectively. The perssstent interests and the temporary interests of the users were distinguished by introducing a temporary interests group. Finally, the recommendation algorithms based on the contents itering were proposed based on our user interest model, and its advantage was demonstrated by comparison experiments [ABSTRACT FROM AUTHOR]