KNOWLEDGE BASED TOOLS FOR FASTER THAN REAL-TIME EPISODIC CHAMPAIGN PLANNING FOR COMPREHENSIVELY ...

THIS PROJECT IS AN EXPLORATION OF THE APPLICATION OF KNOWLEDGE BASED TOOLS TO SCALABLE MILITARY CAMPAIGN SIMULATIONS. THE APPROACH TO BE TAKEN IS TO DEVELOP A GENERALIZED FRAME BASED KNOWLEDGE REPRESENTATION SYSTEM WITH A COMBINATION OF ACTIVE AND STATIC SLOT VALUES TO REPRESENT THE CHARACTERISTICS OF EACH LEVEL OF THE FORCE HIERARCHY. ACTIVE VALUES IN THE SYSTEM MAY CONSIST OF DATA BASE QUERIES, PROCEDURAL (DETERMINISTIC) PROGRAM CALLS, OR NON-PROCEDURAL (INFERENCE ENGINE) CALLS. AGGREGATION IS ACCOMODATED THROUGH ALGORITHMS CALLED BY THE ACTIVE VALUE SLOTS OF THE FRAME. OPPOSING FORCE ACTIVITIES ARE SIMILARLY SIMULATED FOR PLAY AGAINST THE COMPUTER. PROGRAMMING A SIMULATION IS TABLE DRIVEN AND WILL BE ACCOMPLISHED BY CHANGING THE SLOT VALUES IN THE CONTROLLING FRAME(S) OF THE INSTANTIATION. SYSTEM PERFORMANCE IS BROUGHT TO FASTER THAN REAL TIME BY USING A MODIFICATION OF A PROPRIETARY, SCALABLE, DISTRIBUTED, PARALLEL-PROCESSING SYSTEM RUNNING ON A NETWORK OF PERSONAL COMPUTERS. ANTICIPATED BENEFITS/POTENTIAL COMMERCIAL APPLICATIONS - THE KNOWLEDGE REPRESENTATION SCHEME TO BE DEVELOPED SHOULD BE OF VALUE FOR ANY HIERARCHICALLY ORIENTED SIMULATION (ESPECIALLY FOR MILITARY, BUSINESS, OR GOVERNMENTAL UNIT INTERACTIONS). THE MODIFIED PARALLEL PROCESSING SYSTEM WILL MAKE POSSIBLE MUCH CHEAPER AND MORE ROBUST SCALABLE, FAULT-TOLERANT COMPUTING SYSEMS. KEY WORDS - KNOWLEDGE REPRESENTATION, KNOWLEDGE BASED TOOLS, PARALLEL PROCESSING SYSTEM, FORCE HIERARCHY