Fusion of soft information using TBM

Intelligence operations are dependent on humans for data gathering and processing. Measurements made by human observation are subjective and incorporate biases. Systematic processing of such data is not as straightforward as with electronic sensor data. This paper presents work in progress towards developing a system to fuse inconsistent information and minimising observation errors. A case-study is presented, relating the techniques to the application area of civilian intelligence systems. Within our system, the Transferable Belief Model (TBM) was used to fuse soft information from observers, and combine in both the discrete and continuous spaces.