- Formalize media interpretation using logic-based representation formalisms
- Bridge the semantic gap between low-level media analysis and high-level interpretation of media content
- Provide a foundation for future information retrieval and knowledge management systems
In this tutorial we would like to show how formal knowledge representation and reasoning techniques can be used for the retrieval and interpretation of multimedia data. The tutorial explains what we mean by an "interpretation" using examples of audio and video interpretation. Intuitively, interpretations are descriptions of media data at a high abstraction level, exposing interrelations and coherencies. We introduce description logics (DLs) as the formal basis for ontology languages of the OWL (Web Ontology Language) family and for the interpretation framework described in the tutorial. As a concrete example, we consider the interpretation of images describing sports events. It is shown that interpretations can be obtained by abductive reasoning, and a general interpretation framework is presented. We also demonstrate how fusion of the information coming from multiple modalities can be formally modeled.
In order to give a comprehensive overview of the latest research in the complementary fields of multimedia analysis and the Semantic Web we will also cover approaches for dealing with uncertain knowledge and sketch how this can be used in an abductive interpretation framework. Based on a formalization of context knowledge using probabilistic description logics, first results on prioritizing results of logic-based interpretation processes will be presented.
- Introduction: the need to automatically derive high-level content descriptions
- Logical foundations: description logics, rules
- Media interpretation using abudction (high-level)
- Feedback between interpretation (high-level) and analysis (low-level)
- Fusion (high-level) as an abduction process
- Uncertainty for ranking alternatives depending on context information
- Event recognition based on description logics and rules
- Information Retrieval (querying, pull technology)
- Information Distribution (publish-subscribe, push technology)