Design challenges in developing a soft data association process
Hannigan, Megan Patricia
MetadataShow full item record
The intent of this research is to address challenges and present possible methods for an initial system design approach for a data association process in the domain of counterinsurgency where multiple streaming soft (textual message) observation reports are a critical input to the process. An overview of the system includes processes from intelligent input control of soft data to the formation of associated, merged messages that are based on a methodology employing a graph-based approach. The design issues that are discussed to yield a workable and efficient association process include finding efficient ways to search between graphs and the selection of appropriate semantic similarity metrics to associate nodes and arcs. In addition to the baseline architecture, design tradeoff issues regarding the association process, to include the level of specificity with which the input is addressed, and optional techniques for associated-message merging, were explored. Applying data association to a counterinsurgency problem can potentially produce an improved comprehensive evidence base that will assist in reducing search time for subsequent discovery and inferencing operations; and provide more accurate results for analysts making real time decisions.