RELATIONAL DATA FILE II: IMPLEMENTATION.
Abstract
The basic features of the Relational Data File are: Input is based in a system of forms on which data extracted from the sources is entered. Extraction and entry is performed by aides. Keypunched forms are collected on a magnetic tape that is then operated on by programs written in a special form retrieval and manipulation language. FOREMAN. Storage is of binary relational sentences derived from the input form tape by FOREMAN programs. Each sentence is named, and by entering the name of one sentence in another sentence, even non-binary relationships may be expressed. A computer code dictionary encodes each binary sentence as four computer words and full or partial copies of each sentence are entered into four separate files on disk. Each file is divided into subsections, called 'buckets,' and ordered according to one of the four sentence constituents - name, domain element, relation, range element. Only buckets, not individual sentences are transferred between disk and core. Output processes are specified by the user in a procedural language whose elementary commands correspond to the basic operations of arithmetic, set theory, and logic. By combining those commands, the user may implement simple retrievals; compute counts, correlations, and trends; or execute complex inference schemes. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 01, 1966
- Accession Number
- AD0636311
Entities
People
- Roger E. Levien
Organizations
- RAND Corporation