A TOOL FOR MANAGEMENT AND MAINTENANCE OF LARGE MULTI-USER KNOWLEDGE BASES

LARGE MULTI-USER DATABASES ARE ALREADY ONE OF THE MOST SIGNIFICANT APPLICATONS OF COMPUTERS, AND THEIR IMPORTANCE WILL ONLY GROW AS WE ENHANCE OUR ABILITIES TO AUTOMATICALLY GENERATE AND CAPTURE HUGE AMOUNTS OF DATA. USERS INCREASINGLY WILL DEMAND ACCESS TO VERY LARGE DISTRIBUTED KNOWLEDGE BASES THAT CAN ASSIMILATE VAST QUANTITIES OF INFORMATION FROM MULTIPLE DIVERSE SOURCES WHILE STILL PRESENTING A CONSISTENT CONCEPTUAL PICTURE TO A LARGE, DISTRIBUTED GROUP OF USERS. THIS DEMAND WILL BE MET ONLY BY EXPERT DATABASE SYSTEMS CAPABLE OF CLASSIFYING, EXTENDING, AND CROSS REFERENCING THEIR CONTENTS AND PROVIDING VERY FLEXIBLE AND GENERAL USER INTERFACES. THE FGP (FOR FETCH, GENERALIZE, PROJECT) EXPERT DATABASE MODEL IS ONE THAT POSSESSES THE REQUIRED CAPABILITIES. WE PROPOSE TO INVESTIGATE A FULL IMPLEMENTATION THAT WILL SUPPORT A NUMBER OF KEY FEATURES INCLUDING: 1. HEURISTICALLY-BASED GENERALIZATION AND EXTENSION OF STORED KNOWLEDGE ("SPECULATION"); 2. AUTOMATIC REFLECTION OF THE EVOLUTION OF THE KNOWLEDGE BASE IN RESPONSES TO QUERIES, INCLUDING REASSESSMENT OF EARLIER RESPONSES IN LIGHT OF NEW INFORMATION ("SYMMETRIC RECALL"); 3. HEURISTIC MERGING OF RECORDS ("FORGETTING") BASED ON SIMILARITY MEASURES; 4. GRACEFUL DEGRADATION IN THE PRESENCE OF INCONSISTENT OR NOISY DATA; AND 5. GREAT GENERALITY AND FLEXIBILITY IN ITS USER INTERFACE. ANTICIPATED BENEFITS: THE PRODUCT THAT WILL RESULT FROM THIS WORK WILL BE A DOMAIN-INDEPENDENT EXPERT DATABASE SYSTEM THAT WOULD BE ADAPTABLE TO A WIDE VARIETY OF KNOWLEDGE BASES BOTH WITH THE DOD AND IN COMMERCIAL INDUSTRY. AS THE USE OF SUCH KNOWLEDGE BASES BECOMES COMMONPLACE DUE TO THE AVAILABILITY OF MASSIVELY PARALLEL MULTIPROCESSORS AND ENHANCED DESKTOP COMPUTING CAPABILITIES WITH HIGH-SPEED WIDE-AREA NETWORKS, OUR TECHNOLOGY WILL PLAY A CRUCIAL ROLE AS THE UNDERPINNING OF WIDELY USED COMMERCIAL PRODUCTS.