A new metric for multimedia retrieval in structured documents

Sana Fakhfakh, Mohamed Tmar, Walid Mahdi

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

2 Scopus citations

Abstract

Most documents available in Textual Database or in Internet are strongly structured. This is the case for example for scientific papers or written documents using markup languages (HTML, XML). This information provided by the structure can be exploited by systems of information retrieval to define the granularity of elements to return in response to a request made by a user or to improve the relevance of these results. In this article, We are interested in recovering multimedia elements. Like this, we propose a new metric for multimedia retrieval in XML documents which is based on computing a geometric distance between XML nodes while taking into account kinship ties and proximities between them. This measure will introduce a new source of evidence for multimedia retrieval in structural documents which aims at finding relevant multimedia element that focus on the user information need. Experiments have been undertaken to show the effectiveness of our method.

Original languageEnglish
Title of host publicationICEIS 2013 - Proceedings of the 15th International Conference on Enterprise Information Systems
Pages240-247
Number of pages8
StatePublished - 2013
Externally publishedYes
Event15th International Conference on Enterprise Information Systems, ICEIS 2013 - Angers, France
Duration: 4 Jul 20137 Jul 2013

Publication series

NameICEIS 2013 - Proceedings of the 15th International Conference on Enterprise Information Systems
Volume2

Conference

Conference15th International Conference on Enterprise Information Systems, ICEIS 2013
Country/TerritoryFrance
CityAngers
Period4/07/137/07/13

Keywords

  • Geometric distance
  • Image
  • Mutlimedia retrieval
  • Structure

Fingerprint

Dive into the research topics of 'A new metric for multimedia retrieval in structured documents'. Together they form a unique fingerprint.

Cite this