移动社交网络中时空数据分析技术研究

发布时间：2018-08-14 14:31

【摘要】：无线通信和移动计算技术的飞速发展,GPS和北斗等全球定位导航系统的应用普及,便捷的位置采集方式催生了大量描述移动物体(如人,车辆和动物)的时空轨迹数据。随着社交网络服务市场的快速增长和激烈的竞争,目前社交网络服务趋向于移动化形成移动社交网络。在移动社交网络中,具有共同兴趣的人会使用手机或平板电脑等移动设备进行交流,用户可以随时跟踪和分享位置相关的信息,移动社交网络中海量时空数据为研究移动社交网络中对象移动行为带来新的机遇。本论文面向移动社交网络中海量时空数据分析的需求,提出一种新颖的移动社交网络中用户移动模型构建和相似性度量的方法。另外,本文针对实时安全监控领域的需求,提出一种基于Hausdorff距离的网格序列聚类方法对大规模轨迹数据进行异常轨迹检测的算法。本文使用微软亚洲研究院发布的真实数据集GeoLife评估本文中所提出算法的有效性和实时性。实验结果表明,本文所提出的算法可以有效地构建移动用户的移动模型并且在相似性度量准确性上优于传统相似性的度量方法。此外,本文提出的异常轨迹检测算法在有效检测到异常行为的同时大大降低了传统异常检测算法的时间消耗,为实现在线实时异常检测做出一定的贡献。
[Abstract]:With the rapid development of wireless communication and mobile computing technology, GPS and Beidou global positioning navigation systems are widely used. The convenient location acquisition method has given birth to a large number of space-time trajectory data describing mobile objects (such as people, vehicles and animals). With the rapid growth of the social network service market and fierce competition, the current social network services tend to be mobile to form mobile social network. In mobile social networks, people with common interests use mobile devices such as mobile phones or tablets to communicate, allowing users to track and share location-related information at any time. Massive spatiotemporal data in mobile social networks bring new opportunities to study the mobile behavior of objects in mobile social networks. In this paper, a novel method of user mobility model construction and similarity measurement in mobile social networks is proposed to meet the needs of massive spatio-temporal data analysis in mobile social networks. In addition, in order to meet the requirements of real-time security monitoring, this paper proposes an algorithm based on Hausdorff distance based grid sequence clustering to detect the abnormal trajectory of large-scale trajectory data. In this paper, the real data set GeoLife published by Microsoft Asia Research Institute is used to evaluate the effectiveness and real-time performance of the proposed algorithm. The experimental results show that the proposed algorithm can effectively construct the mobile model of mobile users and is superior to the traditional similarity measurement method in the accuracy of similarity measurement. In addition, the anomaly trajectory detection algorithm proposed in this paper not only effectively detects abnormal behavior, but also greatly reduces the time consumption of the traditional anomaly detection algorithm, and makes a certain contribution to the on-line real-time anomaly detection.
【学位授予单位】：南京大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP393.09;TP301.6

【相似文献】