A Logic Based Approach for the Multimedia Data Representation and Retrieval

Overview & Synopsis

Nowadays, the amount of multimedia data is increasing rapidly, and hence, there is an increasing need for efficient methods to manage the multimedia content. This paper proposes a framework for the description and retrieval of multimedia data. The data are represented at both the syntactic (structure, metadata and low level features) and semantic (the meaning of the data) levels. We use the MPEG-7 standard, which provides a set of tools to describe multimedia content from different viewpoints, to represent the syntactic level. However, due to its XML schema based representation, MPEG-7 is not suitable to represent the semantic aspect of the data in a formal and concise way. Moreover, inferential mechanisms are not provided. To alleviate these limitations, we propose to extend MPEG-7 with a domain ontology, formalized using a logical formalism. Then, the semantic aspect of the data is described using the ontology's vocabulary, as a set of logical expressions. We enhance the ontology by a rules layer, to describe more complex constraints between domain concepts and relations. User's queries may concern the syntactic and/or semantic features. The syntactic constraints are expressed using XQuery language and evaluated using an XML query engine; whereas the semantic query constraints are expressed using a rules language and evaluated using a specific resolution mechanism.