Searching and hyperlinking using word importance segment boundaries in MediaEval 2013

Kim Schouten, Robin Aly, Roeland Ordelman

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

1 Citation (Scopus)
62 Downloads (Pure)

Abstract

This paper reports a set of experiments performed in the context of the Searching and Hyperlinking task of the MediaEval Benchmark Initiative 2013. The Searching part challenges to return a ranked list of video segments that are relevant given some textual user query, while for the Hy-perlinking task the aim is to return a ranked list of video segments that are relevant given some video segment. The main focus is on finding a way to compute flexible segment boundaries. This is performed by extending the term frequency part of tf-idf to include the temporal dimension of videos. Although the contribution is theoretically sound its performance is relatively poor, which we attribute to the focus on speech data and the hyperlinking process. We plan to refine our method in the future overcome these limitations.

Original languageEnglish
Title of host publicationMediaEval 2013: Multimedia Benchmark Workshop
Subtitle of host publicationProceedings of the MediaEval 2013 Multimedia Benchmark Workshop, Barcelona, Spain, October 18-19, 2013
EditorsMartha Larson, Xavier Anguera, Timo Reuter, Gareth J.F. Jones, Bogdan Ionescu, Markus Schedl, Tomas Piatrik, Claudia Hauff , Mohammad Soleymani
PublisherCEUR
Number of pages2
Publication statusPublished - 2013
EventMultimedia Benchmark Workshop, MediaEval 2013 - Barcelona, Spain
Duration: 18 Oct 201319 Oct 2013

Publication series

NameCEUR workshop proceedings
PublisherRheinisch Westfälische Technische Hochschule
Volume1043
ISSN (Print)1613-0073

Conference

ConferenceMultimedia Benchmark Workshop, MediaEval 2013
Abbreviated titleMediaEval 2013
Country/TerritorySpain
CityBarcelona
Period18/10/1319/10/13

Fingerprint

Dive into the research topics of 'Searching and hyperlinking using word importance segment boundaries in MediaEval 2013'. Together they form a unique fingerprint.

Cite this