Implementing SMART for minicomputers via relational processing With abstract data types

  • Authors:
  • Edward A. Fox

  • Affiliations:
  • -

  • Venue:
  • SIGSMALL '81 Proceedings of the 1981 ACM SIGSMALL symposium on Small systems and SIGMOD workshop on Small database systems
  • Year:
  • 1981

Quantified Score

Hi-index 0.00

Visualization

Abstract

Designed during the 1960's as a research tool for the field of information retrieval, the SMART system has been operating on an IBM 370 since 1974. SMART is now being enhanced, redesigned, and programmed under the UNIX operating system [28] on a DEC VAX 11/780. The techniques used should allow real-time operation on smaller minicomputers in the PDP 11 family. The implementation provides for a combination of database and information retrieval operations which make it applicable to office automation, personal information system management, and research studies. The SMART vector space model, which treats information requests and stored information records as vectors in an n-space (of terms), is integrated into the relational database model using the concepts of abstract data types (ADTs). Domains of relations are allowed to be any ADT; an extended relational algebra is described with operators that manipulate many complex data structures. After illustrating the application of these concepts to typical SMART tasks, a prototype implementation is outlined. Also included is a discussion of techniques to be employed in a more efficient version.