Recommendation system false information injection method and system in big data scene

Abstract

The invention discloses a recommendation system false information injection method and system in a big data scene. The method comprises the following steps: constructing a Movielens data set as a real data set of a recommendation system; using cosine similarity to construct a similarity matrix to represent the similarity between the target user a and other users; k users with the highest similarity are obtained through sorting, all items preferred by the K users are listed, items scored by a target user a are removed, first N items are calculated, and a formed recommendation result page is recommended to the target user a; constructing an injection model, determining injection strength, filling rate and selection rate parameters to form the number of false users, and determining score data of the false users; and injecting the obtained false data set into a real data set of a recommendation system, wherein the recommendation system is interfered by false information to form a new recommendation page result. According to the method and the device, a recommendation function is effectively realized, and a normal recommendation result page can be displayed when injection is not carried out.

Description

technical field [0001] The invention belongs to the field of network technology, and in particular relates to a method and system for injecting false information into a recommendation system in a big data scenario. Background technique [0002] With the development of mobile networks, the scale of information data is growing explosively, and the concept of big data has attracted widespread attention. While big data enriches human life, it also brings a lot of problems. Information overload is one of the most serious problems in the current big data environment. This problem will greatly reduce the utilization rate of information, and users have to spend a lot of energy to filter redundant information. Big data has five characteristics: huge data volume, fast speed, many modes, difficult to distinguish, low value, and low density. Due to these characteristics, traditional recommendation systems cannot meet the requirements for data processing capabilities. At the same time, ...

AI Technical Summary

Problems solved by technology

The injection of false information seriously interferes with the recommendation results of the recommendation system, seriously affects the real user experience, causes users to gradually lose trust in the recommendation system, damages the legitimate interests of users and the platform, and hinders the popularization and improvement of collaborative filtering recommendation technology in various fields

Images