Recommendation system and method based on multi-order neighbor prediction

Abstract

The invention discloses a recommendation system and method based on multi-order neighbor prediction, and the system comprises a user score matrix building module which is used for building a user score matrix according to a user set and an article set; a similarity calculation module used for calculating the similarity between any two users; an iterative neighbor searching module used for selecting N neighbor users as first-order neighbors for a target user according to a similarity calculation result, then selecting the first-order neighbors of the N first-order neighbors as second-order neighbors, and sequentially iterating until a k-order neighbor set of the target user is found, wherein N and k are positive integers; and a prediction and recommendation module which is used for re-predicting the score of the target user on the article according to the neighbor set searched by each iteration, and recommending the article to the target user. According to the recommendation system andmethod based on multi-order neighbor prediction, the idea of iterative search is adopted, so that the prediction score of the recommendation system and method is more accurate, and the effectiveness is higher.

Description

technical field [0001] The present invention relates to the technical field of information processing, in particular to a recommendation system and method based on multi-stage neighbor prediction. Background technique [0002] The emergence of the Internet has brought us into the era of global informationization, and at the same time, it has also plunged us into the dilemma of information overload. Faced with overwhelming information, people often feel at a loss, and it is difficult to find the information that suits them, thus reducing the efficiency of information use. Therefore, how to analyze and develop massive data and maximize the effective use of information has become a hot research topic. [0003] In order to solve the problem of information overload, there have been three ways of information classification, search engines and recommendation systems. Although information classification and search engines alleviate the problem of information overload to a certain ...

AI Technical Summary

Problems solved by technology

[0004] Although the recommendation system is widely used, there are still many deficiencies in terms of recommendation accuracy and data sparsity.

With the continuous increase of the number of users and items, the sparsity of rating data is becoming more and more obvious, which directly leads to the difficulty of calculating the similarity between users.

In addition, in the selection of neighbor users, due to the intricate relationship between users, if only direct neighbors are selected, users with inconsistent tastes may be found, thereby reducing the accuracy of recommendation

Images