推荐系统中恶意攻击检测方法的实现
首发时间:2013-12-23
摘要:基于协同过滤的推荐系统根据用户的行为信息,过滤出用户可能会感兴趣的信息,并推荐给用户。但是由于推荐系统自身的开放新以及对用户概貌信息的敏感性,推荐系统很容易被恶意攻击者攻击。恶意攻击者通过向系统里面注入攻击用户概貌信息就可以影响推荐系统的正常推荐工作和推荐质量。因此,为了保证推荐系统的安全,需要找出合适有效的方法,检测出推荐系统中的概貌攻击信息。本文对推荐系统相关知识以及攻击模型进行学习,并完成了基于项目识别的用户概貌攻击检测算法的研究与实现。最终的实验结果表明,基于目标项目识别的用户概貌攻击检测算法对于推攻击的检测效果很好,对于核攻击的检测效果不是很理想。
关键词: 计算机应用技术 推荐系统 协同过滤 用户概貌 恶意攻击 目标项目识别
For information in English, please click here
The Implementation of Malicious Attack Detection Method in Recommendation System
Abstract:Based on user profile information, collaborative filtering recommendation system could filter the information that user may be interested in, and recommend to them for their reference. However, due to the openness of the system and the sensitivity of user profile information, recommendation system can easily be attacked by malicious attackers. Malicious attackers can affect the normal recommendation operating and quality by putting phony user profile information into recommendation system. So in order to ensure the safety of the recommendation system and obtain the user's trust, effective malicious attack detection method needed to be provided to detect the attack profile information. In this paper, we studied the knowledge of recommender system and attack model, completed the implementation of the attack detection algorithm based on the target project identified. The result shows, the user profile attack detection algorithm based on the target project identified can help detect the push attack, but to the nuclear attack, it may not work well.?????
Keywords: computer application recommendation system;collaborative filtering;user profile information;malicious attack;target project identified
基金:
论文图表:
引用
No.****
同行评议
共计0人参与
勘误表
推荐系统中恶意攻击检测方法的实现
评论
全部评论0/1000