高级检索
    郭黎敏, 高需, 武斌, 郭皓明, 徐怀野, 魏闫艳, 王之欣, 焉丽, 田霂. 基于停留时间的语义行为模式挖掘[J]. 计算机研究与发展, 2017, 54(1): 111-122. DOI: 10.7544/issn1000-1239.2017.20150784
    引用本文: 郭黎敏, 高需, 武斌, 郭皓明, 徐怀野, 魏闫艳, 王之欣, 焉丽, 田霂. 基于停留时间的语义行为模式挖掘[J]. 计算机研究与发展, 2017, 54(1): 111-122. DOI: 10.7544/issn1000-1239.2017.20150784
    Guo Limin, Gao Xu, Wu Bin, Guo Haoming, Xu Huaiye, Wei Yanyan, Wang Zhixin, Yan Li, Tian Mu. Discovering Common Behavior Using Staying Duration on Semantic Trajectory[J]. Journal of Computer Research and Development, 2017, 54(1): 111-122. DOI: 10.7544/issn1000-1239.2017.20150784
    Citation: Guo Limin, Gao Xu, Wu Bin, Guo Haoming, Xu Huaiye, Wei Yanyan, Wang Zhixin, Yan Li, Tian Mu. Discovering Common Behavior Using Staying Duration on Semantic Trajectory[J]. Journal of Computer Research and Development, 2017, 54(1): 111-122. DOI: 10.7544/issn1000-1239.2017.20150784

    基于停留时间的语义行为模式挖掘

    Discovering Common Behavior Using Staying Duration on Semantic Trajectory

    • 摘要: 移动对象的语义行为模式挖掘是当前移动对象研究中关注的热点,有益于诸多应用场景,如朋友推荐系统、轨迹破案领域和个性化服务等.目前语义行为模式挖掘方法没有考虑移动对象在停留点的停留时间,不能准确地分辨出移动对象之间的不同行为模式.为了解决上述问题,提出了一种基于停留时间的语义行为模式挖掘(discovering common behavior using staying duration on semantic trajectory, DSTra)方法,首先挖掘每个移动对象的频繁语义行为模式,然后定义语义行为模式之间的相似性度量方法,最后采用层次聚类的方法对移动对象进行聚类,找出具有相似行为模式的移动对象群体.实验结果表明:该方法不仅具有合理性和有效性,同时还具有较高的准确率和较好的效率.

       

      Abstract: With the advancement of mobile computing technology and the widespread use of GPS-enabled mobile devices, research on semantic trajectories has attracted a lot of attentions in recent years, and the semantic trajectory pattern mining is one of the most important issues. Most existing methods discover the similar behavior of moving objects through the analysis of sequences of stops. However, these methods have not considered the duration of staying on a stop which affects the accuracy to distinguish different behavior patterns. In order to solve the problem, this paper proposes a novel approach for discovering common behavior using staying duration on semantic trajectory (DSTra) which can easily differentiate trajectory patterns. DSTra can be used to detect the group that has similar lifestyle, habit or behavior patterns. Semantic trajectory patterns of each moving object are mined firstly. Then, the time-weight based pattern similarity measurement is designed. After that, a hierarchical clustering method with pruning strategy is proposed, where each cluster represents the common behavior patterns from moving objects. Finally, experiments on both real-world dataset and synthetic dataset demonstrate the effectiveness, precision and efficiency of DSTra.

       

    /

    返回文章
    返回