电子学报 ›› 2013, Vol. 41 ›› Issue (2): 382-387.DOI: 10.3969/j.issn.0372-2112.2013.02.027

• 科研通信 • 上一篇    下一篇

自动建立信任的防攻击推荐算法研究

黄世平1, 黄晋2,3, 陈健4, 汤庸3   

  1. 1. 中山大学信息科学与技术学院,广东广州 510006;
    2. 深圳移动互联网应用中间件技术工程实验室,广东深圳 518060;
    3. 华南师范大学计算机学院,广东广州 510631;
    4. 华南理工大学软件学院,广东广州 510006
  • 收稿日期:2012-07-16 修回日期:2012-09-30 出版日期:2013-02-25
    • 通讯作者:
    • 陈健
    • 作者简介:
    • 黄世平 男,1987年生于湖南邵东,中山大学信息科学与技术学院博士研究生.研究领域为协同软件技术、数据库、推荐系统. E-mail:hship@mail2.sysu.edu.cn 黄 晋 男,1976年生于海南琼中,博士,华南师范大学计算机学院讲师.研究领域为数据库理论、信息检索技术及推荐系统应用. E-mail:dr.huangjin@gmail.com
    • 基金资助:
    • 国家自然科学基金重点项目 (No.60736020); 国家自然科学基金 (No.60970044,No.61272067,No.61272065); 广东省自然科学基金 (No.S2012010009311); 广东省科技项目 (No.2011A091000036,No.2011168005,No.2011B080100031); 华南理工大学中央高校基本科研重点项目 (No.2012ZZ0088)

Anti-Attack Recommender Algorithm Based on Automatic Trust Establishment

HUANG Shi-ping1, HUANG Jin2,3, CHEN Jian4, TANG Yong3   

  1. 1. School of Information Science and Technology, Sun Yat-sen University, Guangzhou, Guangdong 510006, China;
    2. Shenzhen Engineering Laboratory for Mobile Internet Application Middleware Technology, Shenzhen, Guangdong 518060, China;
    3. School of Computer, South China Normal University, Guangzhou, Guangdong 510631, China;
    4. School of Software Engineering, South China University of Technology, Guangzhou, Guangdong 510006, China
  • Received:2012-07-16 Revised:2012-09-30 Online:2013-02-25 Published:2013-02-25

摘要: 随着互联网中信息资源的日益增多,个性化推荐技术作为缓解"信息过载"的有效手段,得到了越来越多的研究者的关注.由于互联网天然的开放性,在商业利益的驱动下,部分恶意用户通过伪造虚假数据来影响系统的推荐结果,从而达到盈利的目的.本文提出一个自动建立信任的防攻击推荐算法,在考虑了用户评分相似性的基础上,引入适当的信任机制,通过为目标用户动态建立和维护有限数量的信任对象来获得可靠的推荐.大量基于真实数据集的实验表明,提出的算法能大大提高推荐系统的鲁棒性和可靠性,并在一定程度上提高了推荐的精准度.

关键词: 推荐系统, 用户信任, 恶意攻击

Abstract: As the information resources available on the Internet are booming nowadays,personalized recommendation technique,which is an effective approach to ameliorate information overloading,has increasingly received attentions from researchers.Due to the native open nature of the Internet and driven by commercial motives,some malicious users attempt to influence the recommendation result via faking data,hoping to gain profits by manipulating recommendation.This paper proposes an anti-attack recommendation algorithm based on automatic trust establishment.Considering the similarities between user ratings,the proposed algorithm introduces a trust mechanism to obtain reliable recommendations through dynamically constructing and maintaining trusted references for users.Enormous experimental results obtained from real datasets reveal that the proposed algorithm could significantly improve both robustness and reliability of recommendation system,and meanwhile enhance the accuracy of recommendation to some extent.

Key words: recommender system, user trust, malicious attack

中图分类号: