Font Size: a A A

A B2c Mode Model Recommendation System

Posted on:2007-09-11Degree:MasterType:Thesis
Country:ChinaCandidate:Y Z CuiFull Text:PDF
GTID:2209360185956224Subject:Management Science and Engineering
Abstract/Summary:PDF Full Text Request
The electronic commerce recommendation system simulates server to lead people to buy, putting forward proposal in the user's purchase process, and provides user's most possibly interested commodity according to his browsing and purchase information when he will visit the website next time. The system can: turn the visitor into the buyer; so it can enhance sells as well as improve the users'satisfaction.There are several problems in traditional systems from the current B2C website electronic commerce personalization recommendation system: data sparsity, the commodities which are purchased or rated by users only occupy the total commodity number about 1%; new project problem, the new user and the new commodity which doesn't be purchased or rated can't be recommended; solely recommend means, long data processing and low recommendation precision. The article proposes a new multi-models recommendation system's (MMRS) design and realization based on B2C. The system deals with server's log, purchasing history, Web data and user's registration information, uses the associate rule, cluster and a new collaborative filtering method, recommending the result in different ways such as commodities, user's comment and email marketing advertisement. The plan can give effective recommendation and new kinds of commodities to new and old users according to the different information gathered. The system makes progress in data disposing, recommender precision and scope.It's usually very difficult to recognize users in Web Logs. This article discerns users by IP address with browser classification and combines Cookies. For common anonymous user, it may recognize user through the Cookies, obtain the user's browse information, and give the certain commodity recommendation.Because the traditional collaborative filtering recommendation has certain insufficiency such as recommendation precision, the data processing efficiency, this article proposes a collaborative filtering method based on cluster and project forecast in coordination. After the users and the commodities are carried into gathers, the people of the same kind and the commodity of the same sort should be constructed the...
Keywords/Search Tags:Electronic Commerce, Recommendation Systems, Cross-sell, Collaborative Filtering
PDF Full Text Request
Related items