 |
| |
 |
Current Status |
|
|
|
|
The current status of the project includes actions perfomed and achievements reached so far, concerning the main research areas, but also all horizontal and vertical activities of CASAM.
Overall Project Status
The CASAM project has completed its final review (since April 2011) and all activities and research aspects have been completed. Dissemination of the research and evaluation results of the final year continue.
Dissemination and Exploitation of the CASAM project results
- Continuous communication with the EC-FP7 Activities ensures proper Dissemination in the EU context. To this direction, the Project Factsheet has been created and sent to the EC for increasing the project awareness.
- Partners involved also in other similar projects enhance the collaboration and spread of knowledge.
- The Project Website has been updated to include the CASAM Presentation and updated dissemination event lists.
- The project partners have disseminated most of the work perfomed in CASAM in Journals, Conferences and Events.
- The final version of the project Dissemination Plan has been created.
- The final version of the project Dissemination Material has been created. Various promotional material ahve been produced, such as pens, note-books, USB flash drives, T-shirts and the CASAM poster.
- Specific Exploitation plans and markets have been considered and the CASAM Family Products have been identified for the final version of the Business and Exploitation Plan.
- A succesful response for the ICT 2010 conference in Brussels has been achieved.
- Participation in Events and Conferences has increased project awareness in the ICT, EC and scientific community.
Research Areas
Advances in each of the three research areas (multimedia analysis, multimedia interpretation, human-computer interaction) have been achieved throughout the span of the project in order to follow this novel approach to multimedia content annotation.
The main achieved outcomes of the project include:
- A unifying representation for related knowledge. The representation links domain-specific ontologies, in which concepts are represented by domain-specific terms, with multimedia interpretation and context ontologies that represent relations and constraints between concepts. The representation also includes rules that facilitate the maximization of information gain from the user’s input.
- A methodology for knowledge-driven multimedia analysis. The methodology provides ways to enhance multimedia analysis when knowledge is available about the context or the probability of presence of certain entities in the content. The proposed methodology also specifies how the knowledge represented in the multimedia semantic model is used to achieve information extraction from various media (text, image, video and audio). Ontology enrichment methods have been examined for the discovery of new concepts and rules, based on the interpreted content.
- A methodology of reasoning for multimedia interpretation. Development of intelligent methods that seek to acquire missing pieces of knowledge and disambiguate uncertain knowledge about multimedia documents. The algorithms are able to reason about what they know, what they don’t know, what required knowledge has the highest priority and how they acquire it. They are able to merge machine-derived and human-entered knowledge about the content and refine it through directed user feedback. The developed architecture communicates with both the knowledge-driven multimedia analysis module and the human-computer interface.
- Intelligent human-computer interface methodology for maximizing information gain in multimedia annotation. Development of human-computer interface protocols and methods that query the user to acquire specific information. The HCI methods are able to transform information requirements into user effort-optimal interaction scenarios. The HCI method also maintains a model of the user and uses it to customize the queries issued in order to maximize the information value of user feedback.
- A toolkit for computer-aided semantic annotation of multimedia content. The toolkit encapsulates all the methodologies developed and also supports ontology management, ontology enrichment and multimedia content management. At design time (configuration level) the expert constructs the initial domain-specific ontology. Further enrichment of the ontology with new concepts is possible and can be carried out also by the expert user. At runtime (operation level), end users are able to semantically annotate multimedia content.
|
|
|
 |
|
|
 |
| Events |
 |
|
 |
| |
|
Follow our discussions and news at:
|
|
 |
|