Enhancing Multimedia News Exploration and Retrieval using Multimodality Ontology

Abstract

A huge amount of non-textual information is available nowadays in electronic form (e.g. images, videos). The uses of media tools, such as YouTube and Facebook additionally make communication more ef- fective. This requires dealing with multimedia data as a major source of content. That is why; intelligent systems are gaining high attrac- tion for different purposes. Different systems such as those for digital archiving, multimedia analysis and content summarization are arising. In general, multimedia intelligent systems should count on effec- tive indexing and automatic retrieval for multimedia data. Indexing can be done in two ways; either by using keyword annotations or by processing of raw content of media. The latter extracts some low-level features such as colour, texture, and shape describing the multimedia object. Extra semantic concept can be added to enhance the retrieval process. This thesis presents a framework based on modeling sport images in sport news reports through object recognition and semantics,