WeKnowIt D2.2 Contextual media analysis and fusion techniques
Author | Neil Ireson, Gregoire Burel, Annalisa Gentile, Phivos Mylonas, Marios Phinikettos, Yannis Kalantidis, Giorgos Tolias, Yannis Avrithis, Pavel Smrz |
Domain | |
Task | |
Publisher | |
Event | |
Project | WeKnowIt |
Dataset Used | |
Published | 05/11/2010 |
Copyright | The research leading to these results has received funding from the European Community’s Seventh Framework Programme (FP7) under grant agreement n° 215453 |
DOI |
Abstract
This deliverable examines the impact of contextual information on the media analysis techniques employed in WeKnowIt. In particular it examines the context provided by other information that can occur
1. within the analysed document itself (which may be derived from different media types in the case of multimedia documents),
2. the information provided by the user generating/modifying/enhancing the content,
3. that provided by the users’ social network contacts and
4. the general context reflecting the background knowledge in which the analysed docu- ments are created.
In addition the fusion of information is also examined, discussing how to identify where a number of facts concerning a given object (document) are related, i.e. (semantically) similar. Where those facts are complementary, i.e. consistent with each other, or conflicting, i.e. inconsistent, the deliverable discusses how such facts should be fused to reinforce confirming facts and resolve inconsistencies. The fusion of information is examined both within the separate media analysis techniques and also within the cross-media fusion of information. The deliverable provides experimental evidence showing that the consideration of context and the fusion information can significantly improve the performance of the media analysis.
Authors
Main Author(s) : Neil Ireson, Gregoire Burel, Annalisa Gentile (USFD), Phivos Mylonas, Marios Phinikettos, Yannis Kalantidis, Giorgos Tolias, Yannis Avrithis (CERTH-ITI), Pavel Smrz (BUT)
Citations
Links
Link to Deliverable : http://www.weknowit.eu/sites/default/files/D2.2.pdf Project Website : http://www.weknowit.eu/