 |
|
|
|
Intelligent Media Framework
|
The main focus of this work area will concentrate on the development of intelligent media objects that allow for the interaction of cognition based content knowledge with social based consumer knowledge (connect item based knowledge with user based knowledge).
To achieve the overall goal, that media objects can themselves find their ways to the interested user, an open framework for broadcasting environments has to be developed, which allows for the creation, destruction and interaction of media objects.
|
|
|
|
|
| |
Research Focus
|
|
Our idea of intelligent media objects is strongly connected with the scientific field so called recommendation systems. The basic functionality of recommendation systems is usually divided into item based filtering and user based filtering methodologies.
Item based filtering embraces all methods, which provide cognitive knowledge about the content of media objects (feature extraction, content based filtering, object recognition, automatic meta data generation etc.). User based filtering on the other hand seeks for knowledge about the behaviour and interests of users (collaborative filtering, social filtering, user modelling, real-time profiling etc.).
The intelligent media framework has to connect item based knowledge with user based knowledge in such a form that the media content itself is in the position to find its way to an interested audience.
Our approach of interaction of content based items with user based items is novel and bears a series of research challenges. Some of the basic scientific and software engineering questions are:
- Which kind of item based metadata and annotation features can be developed that are suitable to give answers to user based interests and questions? The typical cognitive oriented methods of video and speech analysis (semantic feature detection, moving object tracking, camera movement, face recognition, etc.) have to be extended to meet the challenges of user based social interactions.
- What knowledge about the user is essential and should be used as keywords for human live annotation? Can any kind of knowledge automatically be extracted from user interactions, which is appropriate to support content based analysis methodologies?
- How can the interaction between item based knowledge and user based knowledge be realised? What kind of metaphor is suitable to describe, investigate and implement the interaction of user items with content items? Is – for example – an agent view with respect to productivity better than a biological motivated view of cell interaction and keying?
- What software architecture is suitable for broadcasting environments? The architecture has to meet the demands and constraints of state of the art television platforms as well as the requirements modern software engineering methodologies.
|
|
|
|
|
|
Expected Results
|
|
|
- Development of methodologies combining item based knowledge with user based knowledge in such a form that the media content itself is in the position to find its way to an interested audience within the intelligent media framework
- General agreement on the methodology that will be used to combine item based approaches with user based approaches to realise intelligent media objects. This result feeds into the basic system architecture.
- Analysis of scientific and practical approaches for the development of media objects that allow for the interaction of item based and user based knowledge.
- First specification of the intelligent media framework.
- Final specification of the intelligent media framework.
- Report on implementation and the tests of the intelligent media framework.
|
This site conforms to the following standards:
|
