A theoretical basis for the system with which non-programmer user can communicate in man-machine environment as if it were the human able assistant is presented. Through the consideration on the communication processes, a set of requirements for such the system is listed up and is discussed. Then the outline of the system is explained of which representation of knowledge, the deductive inference algorithm and the file processing algorithm are described somewhat in detail.
The purpose of this paper is to present a theoretical basis for and an architecture of a system with which non-programmer user can communicate in man-machine environment as if it were the human able assistant. In this paper, the author first lists up a set of requirements for such the intelligent system, and then presents the outline of the system that satisfies most of the requirements.
In the man-machine environment, man expects the machine either to respond him with right information to his question or to perform some actions responding t? his request. But we should note here that, in order to obtain good results, man must describe his desires very precisely. Very often the amount of information necessary for it is far much larger than that of the answer given by the system. Because of this nature of communication, the conventional man-machine systems required tedious processes for interaction. Note, however, that the situation is the same even in the case of communication between men. But in this case, both speaker and listener usually have common knowledge on the subjects of conversation and even if the speaker's information is not enough to describe the matters completely, the listener makes up for the loss of information with his own knowledge. Thus, the speaker can be freed from tedious descriptions on the details of things other than the essentials of the subjects. This fact tells us that the smooth conversation is feasible only among those who have common knowledge on the subjects of conversation under way. The case of man and machine is not the exception. The computer, if it is to be useful in the man-machine environment, should keep knowledge on the subjects. Representation, storage, retrieval, update and utilization of knowledge should be considered as the basic techniques in the man-machine systems.
How to represent knowledge is, in general, different between men and, as the matter of course, between man and machine, In order to use such knowledge, each intelligent entity (man and machine) must have an ability of converting a representation of knowledge to another keeping its meaning. This is illustrated in Fig.1.
Much of this paper describes the deductive inference algorithms used in the system. The detailed algebraic formula given in the paper make it almost impossible to transliterate without error.
Several published papers in journals give a flavour of the paper provided for the Seillac II Workshop:
S. Ohsuga, "A Knowledge Structure and a Deductive Inference Rule Based on Knowledge Structure", Information Processing Society of Japan Vol 16, 1976.
S. Ohsuga, "Semantic Information Processing in Man-Machine Systems", Proc. 1977 IEEE Conference on Decision & Control, pp 1351-1358, Dec 1977.
S. Ohsuga, "Theoretical basis for a knowledge representation system", IJCAI'79,
Proceedings of the 6th international joint conference on Artificial intelligence - Volume 2
Pages 676-683.
Theoretical basis for a system named the SBDS (Structure Based Deduction System) including a formalism for
representation of knowledge and an inferential algorithm is presented.
This system is designed as a part of an intelligent interactive system to support man carrying out such
the intelligent activities as engineering design, decision making, scientific research and so forth.
The major emphases on the system are : (1) high processing efficiency, (2) high expressive power so that user
can represent any knowledge he wishes and (3) easy access to databases.