Research Reports

Collective Intelligence Decision Making via Dynamic Knowledge Graph with Trusted Cross-Organizational and Privacy Preserving Integration

PUBLIC RELEASE
October 2024

COMPLETED
July 2024

AUTHORS:  Dr. Feng Liu1, Dr. Xinyue Ye2
STEVENS INSTITUTE OF TECHNOLOGY 1, TEXAS A&M UNIVERSITY 2

The research behind this report aims to develop a comprehensive decision-making framework that harnesses collective intelligence to effectively determine the optimal integration and collaboration mechanism for responsible agencies within the Department of Defense (DoD) when tackling complex tasks that entail cross-domain and cross-organizational operations.

The DoD currently employs the Correspondence and Task Management System (CATMS) that allows users to create, delegate, assign, respond, and search the knowledge repository for tasks and correspondence. The CATMS system primarily handles intra-organization task management, without an integration of the recent development of knowledge graph (KG) theory and privacy preserving mechanism. In this complex and volatile era, the acquisition, deployment, and long-term support of the DoD capabilities typically occur within the purview of the five military services and multiple agencies. The current approach often prioritizes intra-service missions and requirements, potentially overshadowing the critical aspects of inter-service integration and interoperability. The process of establishing, approving, and managing these tasks within the DoD is not only highly time-consuming, but it also tends to be fragmented into service-specific requirements. In addition, the exchange of information between different agencies can be challenging due to data sensitivity, format compatibility issues, and varied protocols. Designing a privacy-preserving federated learning-based knowledge graph (Fed-KG) system, leveraging recent advances from academia and industry that enables users to perform semantic searches at a global level, can facilitate enterprise-level decision-making and foster inter-organizational cooperation.

The developed DoD specific Fed-KG with information embeddings extracted from the publicly available news release, websites, white papers, and other documents. The Fed-KG can predict the joint ownership of new tasks and provide collaboration suggestions between individual agencies and departments, so that enterprises-wise interoperation plans and subtasks ownership plans with underlying budgeting formation can be made. The Fed-KG system can provide suggestions and training to cross-functional teams that address the critical objectives and outputs outlined in the organizational strategy. This research project addresses the following research areas: 1) Acquiring and Integrating Interoperable Capabilities Across Organizational Boundaries and 2) Requirements Setting and Managing Requirements that Cut Across Organizations.