Vienna. 25 March 2009: For over 80 years the NAB show is the world’s largest digital media show for content. The ‘Military and Government Summit’ will take place the very first time and Mark Pfeiffer, CVO of SAIL LABS has been chosen as one of the speakers for this event: ‘I am really proud of having the great opportunity to speak at the session ‘Advances in Technology’. During the three-day summit ways will be identified to utilize cutting-edge commercial video applications and technologies by government and military officials. Mark’s speech 'CAVA (Continuous Automatic Vocabulary Adaptation)' is scheduled for April 21, 8:30 am.

CAVA from SAIL LABS is a framework that automatically adapts and changes automatic speech recognition language models over time to current affairs.  Although the language model of SAIL LABS’ products has a huge vocabulary there are changes which require an adaption nevertheless; e.g. names, people, places are mentioned will be added but perhaps can also be removed again after a special timeframe. The key ideas were to adapt language models to the interests of the users without them needing to interfere into the language model building. The technology allows deployed systems to quickly adapt to changing environments and also identifies the point in time when the vocabulary of a language model should be changed.  The user determines areas of interest, that are to be harvested for domain, or topic specific data, which subsequently form the basis of automatic language model creation.

‘We develop CAVA with the vision of creating a technology to allow for long-term, unsupervised adaptation of a system deployed in the real-world. Users have the opportunity to gather current Open Source Information (OSINF) faster and with an even higher quality so they can focus on analyzing the information and not wasting time with finding it,’ says DI Gerhard Backfried, Head of Research.

Data sources are combined according to their relevance to yield a series of domain, time and topic specific language models. This mix of language model data and sources then add and subtract words and terms depending on their relevance in the current deployment environment. The processed data is then optionally quality checked by human interaction or runs unattended.

‘As the importance and amount of OSINF increases constantly every year it becomes more and more important to detect the information efficiently. Gaining results from different sources allows analysts to compare these in order to improve the reliability and reduce ambiguity of the information,’ states Christoph Prinz, CEO.

Not only more data is available and can be captured through the internet, broadcasts, radio etc. but also the people in need of information have become more and more skilled to search through this openly available content. Our mission at SAIL LABS is to support governments and private entities to achieve an advantage through better qualification of information and intelligence. Other tools recently developed by SAIL include a visualization suite for the OSINT product range. ‘I hope to share the insight and vision SAIL LABS has gained in the last decade by serving the intelligence communities of this world,’ teases Mark.

Please visit Mark Pfeiffer’s speech during the ‘Military and Government Summit’ on April 21, 8:30 am.