Your high-level information is my low-level data: A new look at terminology for multi-level fusion

Once the realm of the military, fusion is increasingly spreading into civilian applications such as crisis management, homeland security, and law enforcement. While some of these applications make use of sensor technologies, in the form of cameras, videos, weather systems, etc., there is an increasing reliance on text-based information, which comes with a plethora of challenges on its own. It is increasingly clear that the ability to automatically and seamlessly combine both hard and soft data into a single model. In part, this is due to the fact many civilian applications generally depend less sensor-based data and more use of human information such as open source (generally web-based) information, emergency services call center logs, increasingly, social media. In some instances, this is not purely a matter of single model for fusion, but a matter of looking at the interplay between multiple models, in particular, between models that process hard data and those that process soft. This paper examines changing needs and new challenges for the fusion community as soft data plays an ever-increasing role. Finally, this paper also revisits the question of how relevant the JDL Data Model is for more complex models of information fusion