Dissertation/Thesis Abstract

Convoy Detection Using Sequence Alignment
by Li, Kai, M.S., Southern Illinois University at Edwardsville, 2019, 55; 27665392
Abstract (Summary)

This paper aims to detect convoys in trajectory data with locations sampled at irregular time intervals. In such cases, convoys that exist may not be detected in some algorithms. To avoid this loss of convoy detection due to sampling irregularities, we propose a new method to extract useful position information and, therefore, to find the missing hotspots. We explore three methods, the add-all-points method, the relaxing time method, and the sequence alignment method. The add-all-points method involves adding interpolated points to all trajectories at all times, the relaxing time method introduces flexibility to the temporal dimension which allows to let a trajectory to be measured against the line segments of the second trajectory, and the sequence alignment method uses sequence alignment to reduce the impact of sampling irregularities. The methods are evaluated against a real-world data set. We find that add-all-points method has good performance, but increasing the size of dataset and performing expensive clustering. Relaxing time method and sequence alignment method do not increase the size of dataset. Sequence alignment method detects more convoys than relaxing time method. Therefore, sequence alignment is a promising method to discover convoys without increasing the dataset size.

Indexing (document details)
Advisor: McKenney, Mark
Commitee: Crk, Igor, Matta, John
School: Southern Illinois University at Edwardsville
Department: Computer Science
School Location: United States -- Illinois
Source: MAI 81/7(E), Masters Abstracts International
Subjects: Computer science
Keywords: Clustering, Convoy, Sequence alignment, Spatial temporal data
Publication Number: 27665392
ISBN: 9781392626474
Copyright © 2021 ProQuest LLC. All rights reserved. Terms and Conditions Privacy Policy Cookie Policy