Design And Research Of E-Commerce Ecommendation System Based On Users Behavior

At present,there are a large number of e-commerce websites, and fast growth rate information resources, which lead to information overload on the whole commodity space, and make the lower use efficient of information. However, existing e-commerce website search system and search engine works mainly based on the user's search terms to provide information filtering ,lack of information available to the user's relevance, convenient and effectiveness and accuracy, in particular, can not meet the search needs of professional information. We have to use personalized recommendation techniques so as to adapt model that E-commerce development into the form of user-centric and the high-speed development pace. Recommended System can not only provide users with personalized service, but also to establish a long-term stable relation with the users, and increase customer loyalty and ultimately improve the sales of e-commerce sites.Recommendation system can simulate sellers who can recommend products sold to customers based on user preferences.Recommendation Algorithm is an important part in personalized recommendation system. Collaborative Filtering Algorithms being as one of the most successful algorithm still have data sparsely and user rating authenticity problems, eventually leading to a sharp decline of recommends quality.Based on this situation, a recommendation algorithm which learns from the users'actions is designed and implemented, by using web usage mining and collaborative filtering techniques. It avoids the use of collaborative filtering techniques. The experimental results show that our algorithm provides better recommendation results and decreasing the density of sparse data sets. This section describes the recommended system's ideas and the implementation process and proofed that improves the accuracy and decrease the data sparsely of the recommendation. The algorithm is composed of three modules: user clustering sub-module, personalized recommendations sub-module, recommended feedback sub-module. According to the algorithm, design and implementation of instance - "Lego net". This article describes the system requirements analysis, database design and the design of each functional module, mainly introducing the design and realization of personalized recommendation system.