A Real-Time Information Fusion and Decision Aid System

Phase I demonstrated the feasibility of using the integrated Dempster-Shafer (D-S) method and the Quasi-Axiomatic Theory (QAT) for battlespace management decision aids. Phase 1/will build upon Phase I to develop a general purpose, reusable commercialsoftware product referred to as A Real-Time Information Fusion and Decision Aid System. Specifically, the D-S method will be used for information fusion, rule induction and knowledge extraction from distributed databases. A natural language processor viaQAT will be developed. A real-time knowledge processing module, which includes a mechanism for rule base compression and an inference scheme based on matrix operators, is embedded in the QAT for performing real-time adaptive decision aiding. It permitsthe processing of different types of data (e.g., symbolic and numeric data, graphs, etc) by the same reasoning tools and the combination of different reasoning techniques and computational methods at the same time, in the uniform QAT decision supportenvironment. The distinguishing features of the product are increased efficiency of information use, knowledge discovery, and inference mechanisms in the presence of positive and negative examples, and machine learning. An application concerning theTheater Missile Defense (TMD) decision aid is conducted to show the usefulness of the proposed approaches.The end product of this project is a commercial software package, A Real-Time Information Fusion and Decision Aid System, which is a significanttechnological breakthrough with enormous commercial value. The Phase II product will be utilized to provide QAT based decision aids in a variety of areas such as the economy, policy making, defense, air traffic management, intelligent control, machineunderstanding of natural language, etc.