ASK is transportable in half a dozen ways

Reviewer: Martin E. Modell

According to the authors, “ASK, A Simple Knowledge System, is a total system for structuring, manipulating, and communicating information. The primary ASK user interface is a simple dialect of English.” The rather odd title of this paper is somewhat of a misnomer in that the most commonly used definition of transportability is from machine domain to machine domain. In this instance the authors treat the verb “to `transport' as a subcase of to `extend.'” That is, they present six ways in which the ASK system can be extended to include new capabilities. The six ways in which ASK is extendible and thus transportable are: :9Bto a new domain, to a new object type, to access data from a foreign database, to a new natural language, to a new programming language, and to a new computer family. Each of the six proposed areas of extension is discussed in some detail, including the steps and methodology employed. The authors' actual experiences are described, both pro and con, along with helpful hints. Comparisons are made with similar implementations using other systems. The authors employ extensive query/response examples to illustrate the proposed methods of extension. Their style is easy to read, and the material is presented in a highly undertandable manner. Some of the authors' statements are of an editorial nature; however, this does not detract in any way from their presentation. On the contrary, their comments seem right on the mark. For example: :9BIt would be nice if there were a “universal” database structure, but there is not. The “epistemological” levels . . . of current database theories leave little room for linguistic subtleties. Ninety-nine percent of ASK is written in standard PASCAL. The machine-dependent parts of ASK are few in number, well isolated and well defined. It would seem that ASK should be easily transportable. Sadly, writing in standard PASCAL does not make that true. . . . Truly transportable PASCAL is not by any means documented, and it is doubtful that it exists. One does not need a preexistent knowledge of ASK, or any natural language or expert system to appreciate the content of this paper. The liberal use of illustrations from the authors' own work indicates extensive work with, and deep understanding of both the specific subject matter of ASK itself, and the wider topic of designing and implementing user-friendly, and thus user-effective, systems. Much of the reference literature originates at the California Institute of Technology, Pasadena, CA, and includes other papers from the same authors. If this paper is any indication, the other papers should be well worth reading.