The key concept for M-Brain's approach is an automatically generated semantic description of media content. This opens up possibilites far beyond purely keyword-matching-based methods by both removing the sample bias and by providing a common, powerful presentation for media content.

From raw data to information

The information extraction process consists of two steps:

• Optimisation of coverage by selecting the smallest possible subset of documents containing all of the relevant documents.
• Optimisation of precision by interactive tools operating in a semantic representation of media content.

As automatic methods never achieve perfect coverage and precision, M-Brain's media analysts then proceed to mold the raw information into a more refined form.