ROVER(ID:4394/rov001)

Heuristic querying language 

Heuristic data processing and querying language (and system)

Aiko Hormann  System Development Corp 1960

Trained by played go-moku with it!
References:
  • Hormann, A. M. "Introduction to ROVER, an information processor" System Development Corp., Santa Monica, Calif., Report No. FN3487, 1960 view details
  • Hormann, Aiko "Computer languages for heuristic processes" view details Abstract: The applicability of computers to many problem-solving situations—chess playing, mathematical proofs, music composition, etc.—has been demonstrated. We wish to explore further heuristic processes suggested by human problem-solving activity. One step toward this goal has been the design of a system named ROVER. ROVER refers both to the simulated computer and the information processing language it uses. Extract: Introduction
    The applicability of computers to many problem-solving situations--chess playing, mathematical proofs, music composition, etc.--has been demonstrated. We wish to explore further heuristic processes suggested by human problem-solving activity. One step toward this goal has been the design of a system named ROVER. ROVER refers both to the simulated computer and the information processing language it uses.
    ROVER'S memory may be said to have a generalized list structure ; consider a pair of items A and B which need not be physically adjacent to each other in the memory. An artificial "adjacency" is introduced to them by attaching directional "links" called "up link" and "down link" to each item, thus determining either that A precedes B or A follows B (and no other item comes in between). An extension of this with three or more items to form a list and a degenerate case of a list with a single item should be clear.
    By forming a list of links when necessary, ROVER is able to have a multiple-linked item appearing simultaneously in any number of distinct lists. (IPL's also can do this but only indirectly.) This feature allows programmers to use a convenient "table look-up" technique which is lost in other list type memories. This, together with ROVER'S flexible language, allows the entire pattern of data, both structure and content, to evolve during the course of processing. The ability of data to carry its own structure and undergo changes during processing is believed to be one of the desirable features in synthesizing complex processes.
    Manual simulation of machine learning using the game of "gomoku" is being performed. Its objectives are to test how desirable or undesirable the features of ROVER are and to understand the basic difficulties involved in synthesis of complex processing. Eventually, it may be possible to build a more practical and elaborate system for such purposes.
          in [ACM] CACM 4(03) (March 1961) view details