SANDMAN: Synthesizing Activities and Narrative Descriptions from MOVINT through Active iNference

Advances in location acquisition and remote sensing have resulted in explosive growth of spatial trajectory data, which capture the mobility of diverse entities. Intelligence analysts need technologies that can provide automated assistance regarding the detection, understanding, and sharing of salient, culturally-informed activities from this mover intelligence (MOVINT) data. Aptima and PatchPlus Consulting propose to develop a system for Synthesizing Activities and Narrative Descriptions from MOVINT through Analysis of Networks (SANDMAN). SANDMAN will generate narrative descriptions and structured summaries of significant events, activities, and anomalies associated with locations and data from MOVINT, GIS, and local-area place details to facilitate and add meaning to the intelligence-analysis process. SANDMAN maps input data of multiple types (text-, geospatial-, and movement-based) into a common latent knowledge representation. With variational autoencoder (VAE) modeling techniques, this unified framework supports recognizing activities from MOVINT data, augmenting them with local contextual information, and synthesizing activity summary content. SANDMAN will be packaged into a containerized prototype with a user interface that enables users to enter in query constraints (e.g., place and time limits for small- and large-scale investigations), run the analysis, and view, manipulate, and share the human-readable results.