LBS连续查询的匿名序列规则挖掘方法研究

Investigation on Anonymous Sequential Rules Mining Method with LBS Continuous Query

随着LBS的深入发展与广泛应用,隐私保护成为LBS深入发展中亟待解决的关键技术问题。时空K-匿名是LBS隐私保护的主要类型,当前研究尚未涉及匿名集数据的可用性和隐私保护的安全性。针对上述问题,基于匿名集数据具有时空序列的特性,提出了一种基于双向不可逆扩展的匿名集序列规则挖掘算法。该算法在扫描序列数据库的过程中,对相应的项集进行位置标记,从而保证了对序列数据库一次扫描即能挖掘出用户移动的序列规则。通过对频繁模式进行扩展并发现用户的移动规律、行为模式,对所提出的算法进行了验证实验及其结果分析。实验结果表明,所提出算法的挖掘结果会涉及到敏感区域,如军事领域等,因此对于实现LBS位置隐私保护具有重要的实践意义,对于丰富隐私保护数据挖掘领域的研究具有一定的理论价值。

With the deep development and wide use of Location-Based Services （LBS） ,privacy protection has become the key technolo- gy to be solved urgently in LBS. The temporal and spatial K -anonymity is the main type of LBS privacy protection,in which the availa- bility of anonymous datasets and the security of privacy protection have not been involved so far. Aimed at this problem and found on a characteristic of spatial-temporal sequences in anonymous dataset,a mining algorithm of anonymity dataset sequence rules has been pres- ented with bidirectional irreversible expansion, which has marked the position of item sets in the process of scanning the sequence database to guarantee mining mobile sequential rules at just one scan of sequence database. Experiments of data mining and result validation have been conducted. Result mined by the algorithm covers sensitive regions like military possessions, which has been proved to have important practical value for realizing LBS privacy protection and certain theoretical value for enriching the study in privacy protection data mining area.