Pharos

The background:

Content is increasingly coming from multiple sources in many different formats and with varying degrees of structure, and there is the need for a coherent approach to its representation format and to its retrieval. A growing variety of audiovisual formats, standards and tools exist, each of them solving only parts of the problem but none addressing it as a whole, with the result that enterprise and businesses remain without a unique and stable reference solution.

The challenges:

Multimedia information retrieval is the challenge of the future.

A first major challenge is to find uniform representation formats that allow heterogeneous data to be merged and indexed and to provide query functionalities that let users retrieve appropriate content from these collections. Furthermore, despite the emergence of sophisticated content-based search methods, the use of the content itself to search for an image or an audio file has not been widely adopted. A further limitation is that web search engines only let users search for and browse entire documents and global metadata.

A further challenge concerns automatic annotation. Think of the possibilities if vast video archives could be searched for specific content with them having to be previously manually annotated! Automated annotation will be provided in PHAROS through a library of annotators applied to audiovisual archives. Querying search and browse capabilities will also be extended to portions of documents with the content itself being used as example for further searches without burdening the user with complex interfaces.