HPS: A novel heuristic hierarchical pruning strategy for dynamic top-k trajectory similarity query
Top-k Trajectory Similarity Query (k-TSQ) is a fundamental operation in trajectory analysis, aiming to identify the k most similar trajectories to the queried trajectory. However, with the increasing demand for real-time data processing, static k-TSQ fails to capture the fresh value of trajectory an...
Saved in:
Published in | Information processing & management Vol. 61; no. 6; p. 103828 |
---|---|
Main Authors | , , , , , , |
Format | Journal Article |
Language | English |
Published |
Elsevier Ltd
01.11.2024
|
Subjects | |
Online Access | Get full text |
Cover
Loading…
Summary: | Top-k Trajectory Similarity Query (k-TSQ) is a fundamental operation in trajectory analysis, aiming to identify the k most similar trajectories to the queried trajectory. However, with the increasing demand for real-time data processing, static k-TSQ fails to capture the fresh value of trajectory analysis. Specifically, directly applying existing static methods for dynamic k-TSQ scenarios will inevitably expose the system to two primary challenges: (1) low response speed and (2) inefficient resource utilization. In order to address the aforementioned issues, we propose a novel heuristic Hierarchical Pruning Strategy (HPS). Firstly, HPS employs a three-layer pruning paradigm, facilitating precise filtering of large datasets and efficient resource utilization. Secondly, to address the issue of resource wastage, we develop a “respond-first, update-later” strategy. Besides, HPS is an orthogonal work with high extensibility, which could be extended or plugged into any state-of-the-art k-TSQ. Empirical experiments on both real-world (T-Drive, 1 GB & UrbanCPS, 7 GB) and synthetic datasets (32 GB) indicate that HPS outperforms existing methods in terms of response time and resource overhead. Specifically, HPS filters out over 99% of trajectory data and improves utilization by 39% for dynamic k-TSQ.
•We propose a novel three-layer pruning paradigm HPS that incorporates well-recognized constraints and the novel LB_TY.•The “respond-first, update-later” strategy of HPS maximizes resource utilization while mitigating load pressure.•Powerful orthogonal characteristics allow HPS to combine with any state-of-the-art methods, jointly achieving better efficiency. |
---|---|
ISSN: | 0306-4573 |
DOI: | 10.1016/j.ipm.2024.103828 |