APDS: A framework for discovering movement pattern from trajectory database

Guan Yuan,Li Yuan,Jian Zhang,Fukai Zhang,Zhixiao Wang,Zhongqiu Wang

doi:10.1177/1550147719888164

Abstract

Currently, the boosting of location acquisition devices makes it possible to track all kinds of moving objects, and collect and store their trajectories in database. Therefore, how to find knowledge from huge amount of trajectory data has become an attractive topic. Movement pattern is an efficient way to understand moving objects’ behavior and analyze their habits. To promote the application of spatiotemporal data mining, a moving object activity pattern discovery system is designed and implemented in this article. First of all, raw trajectory data are preprocessed using methods like data clean, data interpolation, and compression. Second, a simplified density-based trajectory clustering algorithm is implemented to find and group similar movement patterns. Third, in order to discover the trends and periodicity of movement pattern, a trajectory periodic pattern mining algorithm is developed. Finally, comprehensive experiments with different parameters are conducted to validate the pattern discovery system. The experimental results show that the system is robust and efficient to analyze moving object trajectory data and discover useful patterns.

Highlights

With the rapid growth and widely used of GPS devices, RFID sensors, wireless communication, and satellites technologies during recent years, various moving objects can be traced all over the world
Most research works focus on designing academic algorithms, and few of them concentrate on developing moving object data mining systems from systematic view
We aim to designing and implementing an effective and efficient moving object data mining system, which can be used to group moving object similar movement pattern and find their periodic pattern

Summary

Introduction

With the rapid growth and widely used of GPS devices, RFID sensors, wireless communication, and satellites technologies during recent years, various moving objects can be traced all over the world. More and more trajectories are collected and stored in databases. These data often contain a great deal of knowledge that requires urgent analysis. Most research works focus on designing academic algorithms, and few of them concentrate on developing moving object data mining systems from systematic view. A powerful moving object data mining system can greatly promote the development and applications of moving object data mining and related technologies.

Objectives

Methods

Conclusion