Hybrid collaborative filtering recommendation algorithm based on user clustering and commodity clustering

Abstract

The invention discloses a mixed collaborative filtering recommendation algorithm based on user clustering and commodity clustering, and aims at an Internet e-commerce platform to design an attribute preference matrix to realize system recommendation cold start. And performing noise reduction processing on a clustering result, and introducing a fusion factor alpha to obtain a Top-n recommendation set. And finally, according to the purchase records, analyzing association rules among the commodities and preference weights of the users for commodity attributes, obtaining an association rule recommendation set and a user personalized recommendation set, and jointly obtaining a user recommendation list in combination with a Top-n recommendation set, thereby completing multi-dimensional accurate recommendation. According to the method, accurate recommendation can be carried out on the user under the condition that no actual purchase data exists in the cold start stage; compared with a traditional recommendation algorithm, the recommendation accuracy of the recommendation algorithm is greatly improved, in addition, association rule analysis is utilized, ID3 is combined for empowerment, recommendation of association rules among commodities and user personality is achieved, and the recommendation result is more scientific and comprehensive.

Description

technical field [0001] The invention belongs to the field of collaborative filtering recommendation algorithms, and provides a recommendation algorithm based on user clustering and commodity clustering hybrid collaborative filtering. Background technique [0002] With the rapid development of the Internet, the application scenarios of recommendation technology in various fields of the Internet are increasing. In the process of increasing the type and quantity of data, it is a challenging problem how to process and analyze the data obtained by the e-commerce platform and finally recommend the products that users are most likely to purchase to users accurately. [0003] At present, all major Internet platforms have implemented recommendation functions to varying degrees. To ensure the accuracy of recommendations, researchers have proposed many different types of recommendation algorithms, such as content-based recommendation algorithms, association rule-based recommendation a...

AI Technical Summary

Problems solved by technology

However, there are two problems with this method. In the stage of user evaluation assignment, although it reflects a certain real situation, it also fabricates a large number of unreliable evaluations, which will have a negative effect on subsequent recommendation results.

Secondly, in the clustering stage, using this method will affect the clustering effect, and then affect the recommendation results

In addition, using this method lacks justification for the interpretation of the results

Images