中山大学学报自然科学版 ›› 2019, Vol. 58 ›› Issue (5): 17-25.doi: 10.13471/j.cnki.acta.snus.2019.05.003

• 论文 • 上一篇    下一篇

基于时空切分和词向量相似性的轨迹伴随模式挖掘

李欣   

  1. 中原经济区“三化”协调发展河南省协同创新中心∥河南财经政法大学资源与环境学院,河南 郑州450046
  • 收稿日期:2018-10-22 出版日期:2019-09-25 发布日期:2019-09-25

Trajectory accompanying patterns mining method based on spatialtime segmentation and word vector similarity

LI Xin   

  1. Collaborative Innovation Center of Three-aspect Coordination of Central Plain Economic Region∥College of Resource and Environment, Henan University of Economics and Law, Zhengzhou 450046, China
  • Received:2018-10-22 Online:2019-09-25 Published:2019-09-25

摘要:

设计了一种基于时空Hausdorff距离切分、词向量相似性的轨迹大数据挖掘方法,以准确高效地分析数据中的伴随规律,真实反映人群和车辆的流动行为。基于时序特征的一对三Hausdorff距离算法可以排除反向轨迹、挖掘伴随关系;利用时间滑动窗口切分得到的轨迹段集合可为相似性度量建立数据基础。而基于词向量的轨迹相似性度量方法,建立了轨迹和词句的类比关系,体现了轨迹的空间、时间和方向异质性,能较为准确地度量伴随轨迹在结构上的相似程度,为发现同类目标或检测频繁路径等提供参考依据。

关键词: 轨迹数据, 伴随模式, Hausdorff距离, 词向量, 轨迹相似性

Abstract:

A trajectory big data mining method based on spatial-time Hausdorff distance segmentation and word vector similarity is designed in this paper. It can analyze the accompanying rules accurately and efficiently, and truly reflect the flow behavior of people and vehicles. The one-to-three Hausdorff distance algorithm based on time series characteristics can exclude the reverse trajectory and mine the accompanying relations. The set of trajectory segments separated by the time sliding window can establish the basis for the similarity measurement. The method of trajectory similarity measurement based on word vector establishes the analogical relationship between trajectory and sentences, reflects the spatial, temporal and directional heterogeneity of the trajectory, and accurately measures the structural similarity of the accompanying trajectories. It provides a reference for exploring similar objectives, detecting frequent paths as well as other related applications.

Key words: trajectory data, accompanying pattern, Hausdorff distance, word vector, trajectory similarity

中图分类号: