ABEM

Need

The Agent-Based Evidence Marshalling (ABEM) system was designed to help analysts identify clusters in large data sets. However, the original interface provided no insight into the clustering process. It also did not allow analysts to see the data set in the context of the thresholds used to define the clusters.

Idea

Resilient Cognitive Solutions’ re-design of the ABEM interface provides Observability and Directability of the underlying clustering algorithm by showing the direct relationship between algorithm parameters and the impacts on the resulting data clustering. In addition, we provided a simple but powerful mechanism by which analysts could make their perception of clusters match the algorithm defined clusters.

Benefit

The re-designed ABEM interface allows Analyst users to effectively team with the sophisticated algorithms underlying the data clustering by showing the progress of the clustering process in addition to the effect of any changes to key parameters. Furthermore, perceptually determined cluster thresholds ensure that the algorithmically defined clusters are consistent with what the Analyst sees in the data.