A multi-objective recommendation method based on localized differential privacy protection

Abstract

The invention discloses a multi-target recommendation method based on localized differential privacy protection, the steps of which include: 1. Obtaining scoring information of visiting users on items from a website to obtain a scoring matrix; 2. Mapping the scoring matrix into an 01 matrix; 3. , Use the random response mechanism to perturb the 01 matrix; 4. Use the probability propagation algorithm to obtain the weight matrix for the perturbed 01 matrix; 5. Initialize the population according to the weight matrix; 6. Iteratively cross, mutate and update the population; 7. Generate multiple recommendation results based on the iterative population. The present invention can effectively protect user privacy while maintaining the balance between recommendation accuracy and diversity.

Description

technical field [0001] The invention belongs to the field of multi-target recommendation, in particular to a multi-target recommendation method based on localized differential privacy protection. Background technique [0002] With the advent of the era of big data, people are faced with a huge amount and variety of information every day. This information may be text, audio, and files on social APPs; it may also be a dazzling array of products on shopping malls. Faced with this information, users often fall into the dilemma of information overload, unable to find the most valuable one from the many choices. Recommender systems are considered to be the most promising technology for alleviating information overload, because they can actively analyze users' historical behaviors and quickly recommend items that meet their preferences from massive amounts of data. However, with the gradual development of the recommendation system, people have further expectations for the recomme...

AI Technical Summary

Problems solved by technology

[0004] In order to solve the shortcomings of the above existing technologies, the present invention proposes a multi-objective recommendation method based on localized differential privacy protection, in order to effectively solve the problem of poor security of user data in the existing multi-objective recommendation scheme. In this way, the privacy of user data can be better protected and the balance between the accuracy and diversity of recommendation results can be maintained.

Images