Abstract: The problem of a semantic gap between multimedia content and its textual annotation defined by user is urgent. This paper presents the combined approach to presentation of multimedia semantics which is based on aggregation of several multimedia that are similar by content and textual annotations into the classes. Such classes will contain generalized descriptions of objects and links reflected in the multimedia, and keywords from some thesaurus. To create and update these classes we also describe the hierarchical clustering and machine learning operations. This approach should expand the multimedia search and navigating area due to including the media documents with similar content and text descriptions.

Keywords: multimedia data mining, content based multimedia retrieval, text based multimedia retrieval.

ACM Classification Keywords: H.3.1 Content Analysis and Indexing - Abstracting methods. H.5.1 Multimedia Information Systems - Evaluation/methodology.

Link:

THE COMBINED APPROACH TO PRESENTATION OF MULTIMEDIA CONTENT AND TEXTUAL ANNOTATION

Dmytro Nochevnov

http://foibg.com/ijitk/ijitk-vol03/IJITK03-1-p06.pdf