Study On Recommender System Based On Interest Measure

In the developed market economy, customer has more choices, so as to losing customer is usual to proprietor. How to promote the customer's loyalism and holding them, and how to realize cross-marketing is a key in the marketing, especially in E-commerce mode. It is necessary for proprietor to use active marketing strategy, mine customer's taste, analysis customer's purchasing motivation, accordingly abstracting customer's attention effectively, satisfying customer's favor, and make the customer tastes absolute individual shopping experience. Recommender System became the most effective tool.The paper introduces data-mining theory, which is the condition of Interest measure bring forward and the basic recommender system carries out. To satisfy with customers, as the same time, recommends products to customers pertinently, therefore, Interest measure is a focus of attention. The paper expatiates multi-factors about objective and subjective sides of the Interest measure, and then offers some methods to evaluate these factors, the M-FIM is based on these factors. The model properly measure each evaluating index in the multi-factors about Interest measure, then find various person's taste to product, in order that expand the Interest measure's adaptable scale. Collaborative filtering is a successful technology that is implemented in recommender systems today. But, when the system scale (such as the number of customers or the types of products) is very large, collaborative filtering faces great challenges. In order to make the problems solved and improve the quality and efficient of recommender systems, a recommender system based on collaborative filtering and Interest measure was brought forward in this paper. The model use Interest measure pruning the large numbers of redundant rules from data mining, the recommend efficiency and accuracy is improved in a large degree, at the same time, reduces the computer system's cost.This paper resolves the Interest measure-evaluating problem by a multi-criteriondecision making problems, which can bring a more similar customer's taste, and then recommend products individually. Finally, we summarize on the paper, point out defects and the directions that will be further studied in the future.