Recommendation system and method

Abstract

The invention relates to a recommendation system and a method. The system comprises a determination module of a set to be recommended, a score forecasting module, a recommendation generation module and a recommendation control module. The method comprises the steps: a neighbor project set of scored project of target users and the scored project set of the neighbor users of the target users are obtained, and then the intersection of the neighbor project set and the project set is calculated so as to obtain the project set to be recommended; a forecasting score, made by the target users, of each project in the project set to be recommended is acquired; relevant project is recommended to the target users according to the forecasting scores. By taking projects and users into comprehensive consideration, various factors are considered more comprehensively, thus improving the accuracy of project recommendation.

Description

technical field [0001] The invention relates to network technology, in particular to a recommendation system and method. Background technique [0002] With the popularization of the Internet and the rapid development of e-commerce, the recommendation system has been widely used in various fields, and has become an important research content of information technology (IT) technology, and has received more and more attention. At present, almost all large-scale e-commerce systems, such as Amazon, CDNOW, eBay, Dangdang online bookstore, etc., use various forms of recommendation systems to varying degrees. [0003] A recommendation system is based on user-based collaborative recommendation, that is, according to the rating data of the nearest neighbors with similar ratings, recommendations are generated to target users. Another recommendation system is item-based collaborative recommendation, that is, item-based collaborative filtering recommendation, which relies on the similar...

AI Technical Summary

Problems solved by technology

[0004] In the process of realizing the present invention, the inventor found that the existing technology has at least the following defects: the two recommendation systems have the problem of low recommendation quality when the scoring data is sparse

Usually in an e-commerce website, the items purchased or rated by users account for only a limited percentage of the total number of items, resulting in a sparse data set of user ratings for items.

In the case of such a large amount of item data and sparse rating data, on the one hand, it is difficult for the former recommendation system to successfully locate the neighbor user set, which affects the recommendation accuracy; The process of calculating similar user groups on the Internet inevitably becomes the bottleneck of the algorithm

When the latter recommendation system pre-calculates the item similarity, the similarity between most items is zero, which directly affects the recommendation effect, and the recommendation accuracy is not high

Images