A Pragmatic Approach to Event Driven Semantic Machine Understanding

We propose an innovative approach to facilitating the information discovery and assimilation that is associated with intelligence analysis. More precisely we will develop a novel approach to achieving multilingual semantic machine understanding under realistic conditions where sufficient parallel corpora and ontologies are unlikely to be available initially. Our InfoStalker system will overcome this challenge by flexibly exploiting heterogeneous and fragmentary knowledge sources as they become available in order to improve multi-disciplinary and cross-lingual information discovery and processing capabilities. We will further extend the practicality of our approach by integrating it within collaborative intelligence analysis processes – allowing us to increase user efficiency in information scanning and assimilation while revealing a rich body of contextual information that will reduce the ambiguity of the users’ information needs and queries. This integration will also support a work-centered information presentation approach that will provide valuable cueing and facilitate data assessment. Phase I research and development of limited prototypes will provide a solid foundation for the complete implementation of InfoStalker in Phase II and its commercialization as a semantic information discovery tool suitable for cross culture and rapid response knowledge sharing and intelligence analysis.