-5- 
(i.e., the same field type may occur more than once in a record). 
Data files are described to the system via Data Description 
Tables and hence the system is data independent. 
The system provides for maintenance and use of a thesaurus 
for query expansion and for data validation and also permits 
use of tables for data element conversion to code during file 
maintenance. 
The system was written under contract for NASA. It is 
operational but is being extended and improved. It operates 
on the IBM 360/50 under the MFT operating system. The batch 
program was written in assembly language. RECON was written 
in PL/I except for the mas ter I/O control program which was 
also written in assembly language. 
FILE ORGANIZATION — The primary file is made up of data 
records stored sequentially in accession number order. Each 
record is composed of a header containing fixed length fields 
of information which are standard for all records in the file. 
The header is followed by a series of tagged, variable-length 
fields with the possibility of repeating a particular tag 
within the record as many times as is necessary. There is, 
however, no hierarchy in the sense that a set of fields may be 
related to each other and repeated as a set relative to the 
header. The record size is unlimited. The file is stored on 
a data cell at NASA. It is stored using a specially codecal 
variable length ISAM access method. 
When a file is first defined to the system through a data 
description table, any field, whether in the header or in the 
tagged part of the record may be designated for creation of an 
inverted index. These inverted indexes are stored on disc. 
The system also maintains a thesaurus of legal descriptors 
used for validation during file maintenance. This file also 
contains "see" and "see also" references and narrower and broader 
terms. It can be used to expand a query and is available to the 
requestor at a terminal for browsing. 
FILE MAINTENANCE — In the file maintenance operation, fields 
in existing records can be altered, records can be deleted from 
the file and new records can be added to the data base. These 
are added at the end of the master file. Inverted indexes are 
maintained by the system as changes occur. Data validation 
through thesauri and encoding of data via table look-up is also 
supported. 
RETRIEVAL — The system operates in either of two modes for 
retrieval: batch or on-line. In the batch mode, records may be 
selected based upon a Boolean combination of any fields in the 
record regardless of whether they have been designated as inverted 
indexes. In this mode also, the query output can be sorted before 
being presented to the requestor. 
In the on-line mode only inverted index fields may be used as 
terms in the selection logic in order that only qualifying records 
be accessed from the data cell. No sorting of data is permitted. 
