Accession Number : ADA319714
Title : Records Analysis and Classification System: A Proof of Concept System for the Automated Classification of United States Air Force Records.
Descriptive Note : Master's thesis,
Corporate Author : AIR FORCE INST OF TECH WRIGHT-PATTERSON AFB OH
Personal Author(s) : Snoddy, David W.
PDF Url : ADA319714
Report Date : DEC 1996
Pagination or Media Count : 201
Abstract : The records management process utilized within the Department of Defense (DoD) is currently labor intensive. Work is being done to automate portions of this process, but classifying documents and assigning disposition instructions remains a labor intensive, manual operation. Although the requirement for this capability was identified by a DoD sponsored study, an automated computer-based system which can classify and apply disposition instructions has yet to be developed for use within the DoD. This thesis study presents a proof of concept computer program called the Records Analysis and Classification System (RACS) which was developed to demonstrate computer-based techniques for the automated classification of official records. To demonstrate the operation of RACS, a sample of 113 records was collected from the files of an organization at Wright-Patterson AFB. An analysis of the results of the tests conducted with the RACS system indicated that it was capable of accurately classifying 72 out of the 113 records on average. Additionally, the RACS program was designed as a learning system and the test results indicated that it was, in fact, capable of improving its classification accuracy over time.
Descriptors : *CLASSIFICATION, *RECORDS MANAGEMENT, COMPUTER PROGRAMS, AIR FORCE, AUTOMATION, LITERATURE SURVEYS, ACCURACY, THESES, DOCUMENTS, COMPUTER APPLICATIONS, ARTIFICIAL INTELLIGENCE, KNOWLEDGE BASED SYSTEMS, MANUAL OPERATION, RECORDS, LEARNING, NATURAL LANGUAGE.
Subject Categories : Information Science
Distribution Statement : APPROVED FOR PUBLIC RELEASE