Research On Privacy Preserving Vehicle Trajectory Data Mining Technology Based On Semantic

With the rapid development of vehicle-mounted intelligent terminal equipment,various positioning technologies(GPS,GSM,RFID,etc.)and storage technologies,massive amounts of vehicle trajectory data are collected and stored,and applied to multiple areas including user behavior analysis,traffic flow prediction,abnormal vehicle detection,etc.However,a large amount of redundant information and sensitive information contained in the vehicle trajectory data restrict the development of vehicle trajectory data mining.Therefore,based on the offline vehicle trajectory dataset,this dissertation uses the spatio-temporal characteristics of vehicle trajectories and related semantic knowledge to conduct research.The specific research work is as follows:(1)Aiming at the low applicability of the stay point extraction method based on trajectory spatio-temporal attributes,This paper analyzes the characteristics of the vehicle trajectory from the perspective of trajectory semantics,and proposes an extraction algorithm for vehicle trajectory stay points that does not depend on the real road network environment.The experimental results using the real taxi trajectory dataset show that the proposed algorithm has better practicability than the existing stay point extraction algorithm.(2)Aiming at the low practicability of the region of interest mining method based on trajectory space attributes,This paper proposes a density-based spatiotemporal clustering algorithm for fine-grained discovery of changes in the region of interest from the perspective of trajectory space-time attributes and user travel habits.At the same time,it uses geographic anti-encoding technology and ArcGIS software to semantically annotate interest regions and visualize them on the map.Simulation experimental results show that the proposed algorithm has better applicability than the existing region of interest mining algorithms,and can intuitively discover the distribution of interest regions at different time intervals.(3)Aiming at the problems that vehicle location privacy be revealed when mining for vehicle trajectory movement patterns,This article protects sensitive locations from the perspective of vehicle trajectory semantics,an anonymous mobile model mining algorithm based on semantic space is proposed.This algorithm makes use of the geographical spatial distribution features of the points of interest to spatially anonymize the trajectory stay points to satisfy the(k,l)privacy preserving model.At the same time,it implements frequent mobile pattern mining.Theoretical analysis of the algorithm's security,simulation experiments verify that the algorithm not only reduces the loss of trajectory information,but also interprets the spatial semantics of the frequent movement patterns found.