Research And Implementation Of An E-commerce Recommender System Based On Improved Collaborative Filtering

With the rapid development of electronic commerce in the recent ten years, the traditional consumption concept is also in constant change. In the environment of vast amounts of network resources and information overload, how to meet the consumer demand, to allow the user to find the goods they want quickly and accurately, to enhance the consumer experience and satisfaction, and to improve customer loyalty, is particularly important Through the analysis of commodity or the user’s information, recommender systems could provide highly personalized recommendation for the user, and better explore the items of the Long Tail to play a great value of non-hot but potential commodities, so it becomes an important research direction of data mining. As behavioral factors which influence users’interest are numerous, and each factor’s influence to the user is not the same, this thesis which based on the analysis of the recommended system and commonly used algorithms, utilizes the Genetic Algorithm and Collaborative Filtering to do research and implementation on the recommender system.The main researches of the thesis are given as follows.Above all, this thesis proposes a method of extracting user’s interest by optimized weight of behavior based on the Genetic Algorithm, and combines it with the Item-based Collaborative Filtering algorithm to recommend. Due to the customer’s behalf of adding to shopping cart, purchasing, scoring, searching, collecting and other acts affectting on the commodity level of interest are not the same, so, in order to achieve a more accurate prediction of the user’s interest, it’s necessary to give a weight for each user behavior. The thesis utilizes the Genetic Algorithm to learn and optimize the behavior weight, combines it with the weighted behavior extraction of user interest and the Item-based Collaborative Filtering algorithms to recommend products for consumer, and achieves more accurate recommendations through predicting or evaluating the user’s preference.Furthermore, the thesis analyzes the demand of E-commerce recommender system and the overall design of function module and database, and uses the JSP technology and MYSQL, MyEclipse tools to implement a simple electronic commerce recommendation protosystem.