Research On Collaborative Filtering Algorithm Based On Items Attributes And Perference Comparison

With the popularity of the internet and the rapid development of e-commerce, overload of internet information has become a serious problem which the users are facing. It is difficult for the users to find the goods they needed in the vast amounts of products information, so e-commerce recommendation system came into with the need of time. In this paper, we have studied deeply in the area of collaborative filtering recommendation techniques on the basis of the comprehensive analysis of the status quo of the domestic and foreign research.Firstly, since the rating data of users-items rating are extremely sparse in the traditional situation, it is difficult to guarantee the recommendation quality of collaborative filtering recommendation algorithm. To solve this problem, we designed a new user clustering collaborative filtering algorithm based on the items attributes. It maps rating to the value of corresponding items attributes, and then gets users'similarity of different rated items. It clusters processing to the users that have common interests on certain attributes, and then build the users'similarity of different rated items, finally, achieved for the cluster analysis of different user groups which have different taste.Secondly, the traditional items rating recommendation systems is difficult to find suitable rating criteria, and it has some deficiency in mining of large number of rating medium sets, so in this paper, we introduced the related concepts of items preference comparison, and designed a new collaborative filtering recommendation algorithm which is based on multiple ranked choosing domains, which uses the sliding match of the choosing domains to search the correlation of items, and then calculate the value of preference comparison. According to the users'profile matrix, the algorithm can make pre-rate to the unrated items. Finally, on the basis of mentioned study executed simulation analysis. The experimental results show that collaborative filtering recommendation algorithm based on items attributes and preference comparison reducd the error rate, improved the accuracy of pre-rated, reduced the negative influence by sparse rating data and obtained better recommendation quality.