Research On Users Clustering Strategy And User Interest Model Base On Marketing Database

With the rapid development of E-business based on Internet, the issue of the information overload is critical day by day. E-business recommendation system has become one of the essential components of a E-business system. Personalized recommendation technique is the core and key ports of the recommendation system. The major recommendation systems include collaborative-filtering Recommendation, Content-based-Recommendation and so on. There is a high relationship between personal interest and recommend information. So the first problem is how to mine the users' interest in any recommendations, it will influence the precision of result. The theories, methods and process of clustering analyze will be first researched on this paper.. clustering analysis is search the data object's valued relationship in a certain dataset. The advantage of clustering analysis is universality, objectivity, practicability. To guide practice based on the different features and different classes of clustering algorithmsThen research the personal recommend based on recommend system, recommend technique and recommend hot issue and so on. Compare the difference recommend Strategy between home and out. The major issues are cluster based on users and cluster based on products, bring forward the users clustering opinion of this paper. Introduce the methods and process of this recommend strategy. First classify the products, take the different to a same attribute matrix, use the same matrix to descript the different products, cluster the users based on attribute. second based on purchase dataset, cluster users, integrate the both two clustering results to get the last classed and give the advice of recommend at endAt the last part of this paper, research the improvement the interest model based on the result of clustering users, create product attribute subjection degree, build two steps interest set, get the more precise recommend, attach group recommend.