Endre søk
RefereraExporteraLink to record
Permanent link

Direct link
Referera
Referensformat
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Annet format
Fler format
Språk
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Annet språk
Fler språk
Utmatningsformat
  • html
  • text
  • asciidoc
  • rtf
On Searching and Indexing Sequences of Temporal Intervals
Stockholms universitet, Samhällsvetenskapliga fakulteten, Institutionen för data- och systemvetenskap.
2017 (engelsk)Inngår i: Data mining and knowledge discovery, ISSN 1384-5810, E-ISSN 1573-756X, Vol. 31, nr 3, 809-850 s.Artikkel i tidsskrift (Fagfellevurdert) Published
Abstract [en]

In several application domains, including sign language, sensor networks, and medicine, events are not necessarily instantaneous but they may have a time duration. Such events build sequences of temporal intervals, which may convey useful domain knowledge; thus, searching and indexing these sequences is crucial. We formulate the problem of comparing sequences of labeled temporal intervals and present a distance measure that can be computed in polynomial time. We prove that the distance measure is metric and satisfies the triangle inequality. For speeding up search in large databases of sequences of temporal intervals, we propose an approximate indexing method that is based on embeddings. The proposed indexing framework is shown to be contractive and can guarantee no false dismissal. The distance measure is tested and benchmarked through rigorous experimentation on real data taken from several application domains, including: American Sign Language annotated video recordings, robot sensor data, and Hepatitis patient data. In addition, the indexing scheme is tested on a large synthetic dataset. Our experiments show that speedups of over an order of magnitude can be achieved while maintaining high levels of accuracy. As a result of our work, it becomes possible to implement recommender systems, search engines and assistive applications for the fields that employ sequences of temporal intervals.

sted, utgiver, år, opplag, sider
2017. Vol. 31, nr 3, 809-850 s.
Emneord [en]
Temporal intervals, Event-interval sequences, Indexing temporal interval sequences, Embeddings
HSV kategori
Forskningsprogram
data- och systemvetenskap
Identifikatorer
URN: urn:nbn:se:su:diva-136632DOI: 10.1007/s10618-016-0489-3ISI: 000400196100007OAI: oai:DiVA.org:su-136632DiVA: diva2:1055514
Tilgjengelig fra: 2016-12-12 Laget: 2016-12-12 Sist oppdatert: 2017-05-29bibliografisk kontrollert

Open Access i DiVA

Fulltekst mangler

Andre lenker

Forlagets fulltekst

Søk i DiVA

Av forfatter/redaktør
Papapetrou, Panagiotis
Av organisasjonen
I samme tidsskrift
Data mining and knowledge discovery

Søk utenfor DiVA

GoogleGoogle Scholar

Altmetric

Totalt: 21 treff
RefereraExporteraLink to record
Permanent link

Direct link
Referera
Referensformat
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Annet format
Fler format
Språk
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Annet språk
Fler språk
Utmatningsformat
  • html
  • text
  • asciidoc
  • rtf