基于时间序列的轨迹数据相似性度量方法研究及应用综述

Research and application survey of similarity measurement methods on trajectory data based on time series

移动通信和传感设备等位置感知技术的发展,将人和事物的地理位置数据化。由此形成的轨迹数据正以极快的速度产生并呈指数级增长。轨迹数据中蕴含着丰富的信息,对其分析和挖掘具有重要的社会和应用价值。轨迹相似性度量研究是轨迹数据管理和分析的基础,在轨迹计算中起决定性作用。轨迹相似性度量通常以时间、空间或若干关键词作为参数,度量轨迹之间的相似程度。本文总结归纳了现有的轨迹相似性度量经典研究工作。首先,总结归纳了轨迹大数据的高维异构、多粒度、不确定、高冗余的特点,对轨迹的相似性度量问题进行了形式化的描述。其次,依据不同的数据类型,就经典的空间相似性、文本相似性和时间相似性的评价方法分别进行了说明和总结;依据轨迹形式与度量范围,将现有工作分为了基于离散点的轨迹全局和局部相似性,基于线段的轨迹全局和局部相似性的计算方法,并评价了各种方法的优缺点。再次,分析了轨迹相似性度量在交通管理、城市规划、智能推荐、智慧出行等领域的具体应用。最后,总结展望了轨迹相似性度量在未来研究与应用方面的发展方向。

The development of location-aware technologies such as mobile communications and sensing devices digitizes the geographic location of people and objects.Trajectory data is generating at an extremely fast rate and growing exponentially.Trajectory data contains a wealth of information.It shows great value of the applications based on analyzing and mining trajectory data.Trajectory similarity measurement research is the basis of trajectory data management and analysis,it plays a decisive role in trajectory calculation.To the best of our knowledge,trajectory similarity measure usually takes the temporal,spatial,or textual information as the parameters.The existing classical research works on trajectory similarity are inducted.Firstly,four features of big trajectory data,such as high dimensional heterogeneity,multi-granularity,uncertainty and high redundancy with noise,are summarized and then the trajectory similarity problem is formed.Secondly,the well-known measurements on spatial similarity,text similarity and time similarity are summarized according to different data type.Based on the trajectory form and the measurement range,the existing similarity work can be divided into four categories,i.e.The global and local trajectory similarity on discrete points,global and local similarity of trajectory on segment.Thirdly,various applied scenarios are discussed,including traffic management,urban planning,intelligent recommendation,smart travel,and so on.Finally,the development direction of trajectory similarity measurement on research problem and application scenarios are predicted.