Research On Individualized Self-adaptive E-Commence Recommendation Systems Based On Implicit Feedback

With the popularization of Intemet and the development of E-Commerce, E-Commerce system offers people convenient service. But the problem of "Information Overload" and the increasingly complex structure of E-Commerce web site make it hard for people to find products and services they want. While the explosion of online information has introduced new opportunities for finding and using electronic data, it has also underscored the problem of isolating useful information and making sense of large, multidimensional information spaces. Although the progress of search engine's technology expediates information searching process, people do not always know exactly what they want, most of them browse unconsciously on line; however, they are the potential consumers of the E-Commerce. In response to these problems, recommendation systems have gradually become an important part in E-Commerce IT technologies, more and more research papers about recommendation systems in E-Commerce have appeared in various conferences and journals.Recommendation systems predict consumer's preference and provide suggested items. Accurate recommendation services on a website can attract latent customers, increase consumer's adhesiveness and at the mean time enhance website sales. Presently, E-Commerce recommendation systems have been very successful in both research and practice, but challenging research problems still remain. Aim at the main challenges of recommendation systems in E-Commerce, this thesis explored and researched some key technologies of recommendation systems in E-Commerce, analyzed several representative sites' recommendation services at present, therefore raised a framework of individualized adaptive recommendation system in E-Commerce based on implicit feedback. The main research works in this thesis included research of analyzing and processing user's implicit information behind his browsing behavior, and thus designing and realization of real-time recommendation services automatically.This thesis aimed to design individualized self-adaptive recommendation systems based on implicit feedbacks from user. Especially, gathering and processing user's individual information efficiently is the key for recommendation systems to achieve valid and successful recommendation. And experiments showed that processing user's implicit feedback information through fuzzy theory can decrease error effectively and make recommendation better.Furthermore, analyzation of user's interest preference on real-time make it possible for recommendation to be adaptive. Experiments showed that obtaining user's real-time interest preference can let us know better about its excursion and adapt recommendation for the change, which definitely raises user's satisfaction of recommendation.Based on the theory research, the SME recommendation system on movies has been developed to test and further our analyzation.