基于移动数据的异常区域时序分析 - 计算机应用研究.pdf

优先出版 计 算 机 应 用 研 究 第34 卷 基于移动数据的异常区域时序分析* 班雷雨，霍 欢，徐 彪 (上海理工大学 光电信息与计算机工程学院，上海市 200093) 摘 要：移动数据描述了大量的关于移动对象活动位置和时间变化的序列，反应出城市动态规划的语义知识。发现移动 对象活动的异常区域，是发现移动对象时序变化的关键分析前提。因此，针对移动对象的活动轨迹分别从时间和空间的 角度进行了研究，首先，从空间区域语义知识的角度分析，利用网格对移动对象的活动区域进行划分,并结合核函数和 Top-k 排序方法发现异常区域；接着，从时间角度分析，采用基于二进制序列的方法，发现移动对象活动周期;最后，在 真实数据集上，验证了该方法的可行性和有效性。 关键词：移动数据；异常区域；网格索引；核函数；傅里叶变换；活动周期 中图分类号：TP391 Temporal analysis of anomalous region based on mobile data Ban Leiyu, Huo Huan, Xu Biao (School of Optical-Electrical Computer Engineering, University of Shanghai for Science technology, Shanghai 200093, China) Abstract: Mobile data shows a large number of the changing sequence of location and time about the moving objects, reflecting semantic knowledge about the city dynamic planning. In this paper, the discovery of anomalous regions visited by moving objects is a critical premise for the discovery of temporal changes about moving objects . Thus, this paper analyzed respectively the movement trajectories of moving objects from the temporal and the spatial . Firstly, from the perspective of spatial regional semantic knowledge, the grid divided the moving objects ’ entire activity space and then to find the anomalous regions by using the kernel function and Top-k sorting method . Secondly, from the perspective of the temporal, the paper proposed a method based on binary sequence to find activity periods of moving objects. Finally, experimental results validate accurately the feasibility and effectiveness of the above methods on the practical data sets. Key Words: mobile data; anomalous region; grid indexing; kernel function ;