Services for an Individualized Dynamic Environment using Knowledge, Information, and Context for Workspace Management (SIDEKICK)

Intelligence analysts are flooded with complex data, creating failures and delays in exploiting critical information. Attempts at automated decision-aids have failed because they ineffectively strive to supplant the analyst, rather than enable collaboration. Instead, analysts manually sift through data, leading to human errors from skill and knowledge gaps and human biases (e.g., confirmation bias). Analysts need automated assistants that provide a collaborative human-automation partnership, forming a common ground with the analyst that drives workspace management for information observability. To address these needs, we propose to design and demonstrate Services for an Individualized Dynamic Environment using Knowledge, Information, and Context for Workspace Management (SIDEKICK). SIDEKICK builds on our team"s previous research designing a framework for intelligence analysis, expanding on iterative broadening and narrowing with a Joint Activity Theory-based approach for automation-analyst coordination. SIDEKICK includes: (1) an Analysis Modeling Service that uses Probabilistic Relational Models to provide a robust modeling framework that can produce a common ground representation of the analysis based on observations of analyst behaviors; and (2) a Workspace Adaptation Service that uses case-based reasoning to analyze the common ground representation to recommend selections and adaptations of workspace tools and visualizations.