Accession Number : AD0427004


Title :   PROGRAM DOCUMENTATION FOR MARK I STATISTICAL ASSOCIATION PROCEDURES FOR MESSAGE CONTENT ANALYSIS,


Corporate Author : MITRE CORP BEDFORD MASS


Personal Author(s) : Baker,J ; Vicksell,R ; Spiegel,J


Report Date : Dec 1963


Pagination or Media Count : 48


Abstract : A statistical method for automatic document retrieval and message content analysis is described. The method involves building a matrix for the corpus based of word co-occurrences within sentences; this matrix is then normalized in order to eliminate what are considered to be extraneous factors. The normalized matrix is used by the retrieval algorithm to expand a set of query terms to include terms associated with them, the new set, in turn, being used to select documents from the corpus. All of these operations are performed on an IBM 7090 computer. This report gives a detailed description of the computer programs involved. (Author)


Descriptors :   *COMPUTER PROGRAMMING , DATA PROCESSING , INFORMATION RETRIEVAL , DOCUMENTS , MATRICES(MATHEMATICS) , STATISTICAL ANALYSIS , DIGITAL COMPUTERS , MAGNETIC TAPE


Distribution Statement : APPROVED FOR PUBLIC RELEASE