Similarity search for spatial trajectories using online lower bounding DTW and presorting strategies

Marie Kiermeier, Martin Werner

Publikation: Beitrag in Buch/Bericht/KonferenzbandKonferenzbeitragBegutachtung

Abstract

Similarity search with respect to time series has received much attention from research and industry in the last decade. Dynamic time warping is one of the most widely used distance measures in this context. This is due to the simplicity of its definition and the surprising quality of dynamic time warping for time series classification. However, dynamic time warping is not well-behaving with respect to many dimensionality reduction techniques as it does not fulfill the triangle inequality. Additionally, most research on dynamic time warping has been performed with one-dimensional time series or in multivariate cases of varying dimensions. With this paper, we propose three extensions to LBRotation for two-dimensional time series (trajectories). We simplify LBRotation and adapt it to the online and data streaming case and show how to tune the pruning ratio in similarity search by using presorting strategies based on simple summaries of trajectories. Finally, we provide a thorough evaluation of these aspects on a large variety of datasets of spatial trajectories.

OriginalspracheEnglisch
Titel24th International Symposium on Temporal Representation and Reasoning, TIME 2017
Redakteure/-innenSven Schewe, Jef Wijsen, Thomas Schneider
Herausgeber (Verlag)Schloss Dagstuhl- Leibniz-Zentrum fur Informatik GmbH, Dagstuhl Publishing
ISBN (elektronisch)9783959770521
DOIs
PublikationsstatusVeröffentlicht - 1 Okt. 2017
Extern publiziertJa
Veranstaltung24th International Symposium on Temporal Representation and Reasoning, TIME 2017 - Mons, Belgien
Dauer: 16 Okt. 201718 Okt. 2017

Publikationsreihe

NameLeibniz International Proceedings in Informatics, LIPIcs
Band90
ISSN (Print)1868-8969

Konferenz

Konferenz24th International Symposium on Temporal Representation and Reasoning, TIME 2017
Land/GebietBelgien
OrtMons
Zeitraum16/10/1718/10/17

Fingerprint

Untersuchen Sie die Forschungsthemen von „Similarity search for spatial trajectories using online lower bounding DTW and presorting strategies“. Zusammen bilden sie einen einzigartigen Fingerprint.

Dieses zitieren