System and method for detecting attack of collaborative recommender based on interest combination

Abstract

The invention provides a system and method for detecting the attack of a collaborative recommender based on interest combination. The system comprises a user information record module, a user information storage module, a user interest combination mining module and an attack user judging module which are connected sequentially. The user interest combination mining module comprises a clustering module, a filtering module and an interest combination confirming module which are connected sequentially. The clustering module is connected with the user information storage module, and the interest combination confirming module is connected with the attack user judging module. The method comprises the following steps of: recording the existing grading information of a user; establishing a user-program grading matrix; carrying out cluster analysis, filter analysis and interest combination confirmation by the user interest combination mining module to obtain a standard user interest combination; analyzing a target user and determining the attack user by the attack user judging module; and finishing the determination of the attack user.

Description

technical field [0001] The invention belongs to network application technology, and in particular relates to an interest combination-based collaborative recommendation attack detection system and method. Background technique [0002] The recommendation system based on Web has been widely used in the current e-commerce environment. It aims to analyze the information that users may be interested in by analyzing the record information left by users in the e-commerce system through data mining, pattern recognition and other fields. The content is recommended to users to achieve the purpose of improving user experience and winning good loyalty from users. Collaborative Filtering method is one of the most widely used methods in recommender systems at present. It attempts to find out the relationship between user behavior and items by capturing the relationship between user behavior patterns integrated in Web sites and Web objects. contacts in order to make recommendations for use...

AI Technical Summary

Problems solved by technology

Most of the existing attack detection models start from analyzing the behavior characteristics of the attacking users, and investigate their attack behavior by judging whether the potential attacking users conform to various known attacking user pattern characteristics, which is not in line with the reality. The detection system is unknown to the attacking user model. The actual situation

Moreover, due to the limitation of the known attack user model, this attack detection model is not ideal for detecting hybrid attacks

Images