Multimedia semantics integration using linguistic model

Abstract

The integration of multimedia semantics is challenging due to the feature-based representation of multimedia data and the heterogeneity among data sources. From human viewpoint, multimedia data objects are often considered as perceptions of the real world, and therefore can be represented at a semantic-entity level in the linguistic domain. This paper proposes a paradigm that facilitates the integration of multimedia semantics in heterogeneous distributed database environments with the help of linguistic analysis. Specifically, we derive a closed set of logic-based form expressions for the efficient computation of multimedia semantic contents, which include conceptual attributes and linguistic relationships into the consideration. In the expression set, the logic terms give a convenient way to describe semantic contents concisely and precisely, providing a representation of multimedia data that is closer to human perception. The space utilization is also improved through the collective representation of similar semantic contents and feature values. In addition, the optimization can be easily performed on logic expressions using mathematical analysis. By replacing long terms with equivalent terms of shorter lengths, the image representation can be automatically optimized. Using a heterogeneous database infrastructure, the proposed method has been simulated and analyzed. © Springer-Verlag Berlin Heidelberg 2006.