基于位置推荐中的隐私保护方法研究-《计算机技术与发展》

文章信息/Info

Title:: Research on Privacy Protection Method Based on Location Recommendation

作者:: 张海涛1 ; 汪佩佩2; 1. 南京邮电大学地理与生物信息学院,江苏南京 210023;
2. 南京邮电大学通信与信息工程学院,江苏南京 210003

Author(s):: ZHANG Hai-tao1 ; WANG Pei-pei 2; 1. School of Geographic and Biologic Information,Nanjing University of Posts and Telecommunications,Nanjing 210023,China;
2. School of Telecommunications &Information Engineering,Nanjing University of Posts and Telecommunications,Nanjing 210003,China)

Keywords:: location-based recommendation; sensitive location; independent location; privacy protection

摘要:: 基于位置的推荐系统能结合用户所在的地理位置和用户的兴趣爱好,对用户提出的服务请求进行分析,快速为用户提供有效的个性化服务。但是,在位置推荐过程中,用户得到的推荐列表中可能包含具有敏感特性或独立性特征的位置数据,这些数据对用户的隐私会造成一定的威胁。传统的基于位置推荐的隐私保护方法不能有效应对此类问题。为此,提出了一种基于敏感位置和位置独立性的推理攻击的隐私保护方法。首先,位置推荐服务器对用户发送的位置服务请求进行分析,生成推荐列表;然后,在隐私保护过程中设置一定的隐私保护等级,对获得的推荐列表进行敏感位置和位置独立的检测;最后,将检测到的敏感位置进行隐藏,将独立性位置数据进行删除。实验结果表明,该方法可以实现推荐结果中敏感位置的动态隐藏,有效消除基于敏感位置和独立性位置的推理攻击,并且具有代价小、速度快、可用性强的优点。

Abstract:: The location-based recommendation system can analyze the service request submitted by the user in combination with the geographical location of the user and the user’s interest,and provide the effective personalized service for the user quickly. However,in the location recommendation,the recommended list received by the user may contain location data with sensitive or independent features that pose a threat to the privacy of the user. Traditional location-based privacy protection methods cannot cope with such problems effective-
ly. For this,we present a privacy protection method of reasoning attacks based on sensitive location and independent location. First,the location recommendation server analyzes the location service request sent by the user to generate a recommendation list;then,we need to set a certain level of privacy protection during privacy protection,and detect the sensitive locations and independent locations in the recommended list;finally,the sensitive location will be hidden and the independent location data will be deleted. Experiments show that the proposed method can realize dynamic hiding of sensitive locations in the recommendation results,and effectively eliminate reasoning attacks based on sensitive location and independent location,with the advantages of low cost,high speed and strong usability.