Time：10:00-12:00 Wednesday 7/12/2017
Place：Room 119，School of Electronics and information，Chang'an Campus
Reporter：Paulo C. G. Costa, Associate Professor
Invited Guest：ZHANG Jindong
Topic：First Order Probabilistic Semantics in High-Level Information Fusion
Research on the subject of information fusion has focused primarily on lower-level data alignment (e.g. multi-sensor data fusion, syntactic protocols, distributed simulation, etc), on semantic mapping solutions (e.g. Semantic Web approaches, specialized semantic mapping solutions, etc), or other topics that do not fully address the fundamentals of high-level knowledge integration. As information flow in many real world applications grows larger and more complex, it becomes clear that advances in connectivity and computation alone are insufficient to address the problem of merging knowledge from heterogeneous sources. The sheer volume of data creates informational and cognitive bottlenecks. Incompatible formats and semantic mismatches necessitate tedious and time-consuming manual processing at various points in the decision cycle. As a result, massive amounts of potentially relevant data remain unexploited, narrow processing stovepipes continue to provide stop-gap solutions, and decision makers’ cognitive resources are too often focused on low-level manual data integration rather than high-level reasoning about the situations to be addressed.
This knowledge gap has been recognized and in spite of recent advances in HLIF research there is still a lack of a theoretical framework to enable HLIF applications. In this presentation, I introduce First-Order Probabilistic Semantics as a candidate for filling this gap, as it addresses the various challenges in merging complex data while properly accounting for the inherent uncertainty that comes from such data. I will present the key concepts of the framework and provide an update on the current status of its development, while showcasing a few examples of how the framework is being applied in diverse application areas.
