AUTOMATED LIBRARIAN

HNC HAS DEVELOPED AN AUTOMATED MECHANISM FOR INDEXING AND RETRIEVING FREE TEXT DOCUMENTS CALLED DOCUMENT RETNEVAL USING MEANING (DKUM). DRUM IS BASED ON CONTEXT VECTORS THAT ENCODE A REPRESENTATION OF THE MEANING OF DOCUMENTS. THIS HNC PROPRIETARY APPROACH WAS CHOSEN FOR FUNDING BY DARPA AS PM OF THE TIPSTEK TEXT DETECTION AND EXTRACTION PROGRAM. DRUM IS ABLE TO: * PERFORM FREE TEXT DATABASE DOCUMENT RETRIEVAL AND ROUTING. * ADAPTIVELY REFINE CONTEXT SEARCH PARAMETERS USING RELEVANCE FEEDBACK. . PROVIDE AUTOMATIC SYSTEM GENERATION CAPABILITIES AND AN EASY-TO-USE QUERY INTERFACE. * ALLOW EASY EXPANSION TO DOMAIN SPECIFIC TOPIC AREAS. THE DARPA PROTOTYPE SYSTEM ADDRESSES THE SOLICITATION REQUIREMENTS OF FREE TEXT RETRIEVAL AND DYNAMIC ADAPTATION, AND HAS BEEN DEMONSTRATED ON TEXT CORPORA OF OVER 10,000 DOCUMENTS. HOWEVER, THIS SYSTEM NEEDS TO BE ENHANCED TO ADDRESS THE REQUIREMENTS OF AUTOMATIC INDEXING, CATALOGING, AND S U MM ARI Z A TI O N . HNC PROPOSES TO ADDRESS THESE ENHANCEMENTS BY EXTENDING THE EXISTING DKUM SYSTEM TO INCLUDE AUTOMATIC INDEX TERM GENERATION AND AUTOMATIC SUMMARIZATION OF DOCUMENTS. BOTH CAPABILITIES CAN BE EASILY INTEGRATED INTO THE EXISTING DARPA SYSTEM. THE PROPOSED APPROACH WILL PROVIDE THE AIR FORCE WITH A HIGH PERFORMANCE SYSTEM AT LOW TECHNICAL RISK.