Collaborative filtering recommendation method based on item classification and user classification

Abstract

The present invention relates to a collaborative filtering algorithm, specifically a collaborative filtering recommendation algorithm based on item classification and user classification, characterized in that it adopts the following steps: A, item clustering and classification steps; B, user clustering and classification step; C, fusion step of item clustering and user clustering; D, sorting recommendation step. Compared with the prior art, the invention uses the improved KMEANS algorithm to complete the data clustering, the method is simple, the expandability is increased, and the sparsity problem and the cold start problem are solved at the same time.

Description

technical field [0001] The invention relates to a collaborative filtering algorithm, in particular to a collaborative filtering recommendation algorithm based on item classification and user classification. Background technique [0002] In today's society, there is a vast amount of information on the Internet, and a large number of new works are added every year. How can the audience find the information they like? Currently, search engines are one of the important means for users to find information, but this is not the answer to the problem. Because search engines have the following fatal flaws: 1. Traditional search algorithms present exactly the same search ranking results for all users, and cannot provide corresponding services for different users’ individual characteristics; 2. For a search keyword, the search engine will Tens of thousands of information items are returned, and only a very small part of them is what the user really needs or is interested in; 3. The pr...

AI Technical Summary

Problems solved by technology

Because search engines have the following fatal flaws: 1. Traditional search algorithms present exactly the same search ranking results for all users, and cannot provide corresponding services for different users’ individual characteristics; 2. For a search keyword, the search engine will Tens of thousands of information items are returned, and only a very small part of them is what the user really needs or is interested in; 3. The premise of searching is that the user knows what he / she needs, if even the user does not know what he / she can get When you have something or want something, search is useless

There are three main deficiencies in the collaborative filtering algorithm: one is the sparsity problem, that is, when the amount of data in the recommendation system is large and the user's explicit rating data is small, it is difficult to calculate the similarity and cannot recommend; the other is the cold start problem , when a new item first enters the system, there are no users to evaluate it, resulting in collaborative filtering unable to recommend the resource

The third is the scalability problem. The users and resources in the recommendation system will grow rapidly over time, while the complexity and data volume of the collaborative filtering algorithm increase linearly, which seriously affects the execution efficiency, resulting in poor scalability.

Images