Shape Understanding System

This book presents the selected results of research in a newly established area of scientific research which we call machine understanding. Machine understanding is based on further development of the shape understanding system (SUS) that was described in our previous books titled “Shape Understanding System: the First Steps toward the Visual Thinking Machines” and “Shape Understanding System: Knowledge Implementation and Learning”. This is the third book that presents the results of research in the area of thinking and understanding, carried out by authors in the newly founded the Queen Jadwiga Research Institute of Understanding. Machine understanding is the term introduced by authors to denote understanding by a machine (SUS) and is referring to the new area of research the aim of which is investigating the possibility of building the machine with the ability to understand. SUS, as the machine that is designed to have the ability to think and understand, learns both knowledge and skills in the process of learning is called the knowledge implementation described in book [103]. A machine to be able to understand needs to some extent mimic human understanding and for this reason machine understanding is based on the assumption that the results of understanding by the machine (SUS) can be evaluated according to the rules applied for evaluation human understanding. The important part of machine understanding approach is investigation of the different forms of explanations how to solve a problem (text problem) or explanations of the causes and context of an object or phenomenon. Machine understanding is based on further development of the shape understanding system (SUS) that is the implementation of the shape understanding method described in our previous books.

Machine understanding is the term introduced by authors to denote understanding by a machine and is referring to the new area of research the aim of which is investigating the possibility of building the machine with the ability to understand. A machine to be able to understand needs to imitate the way in which humans understand and is based on the assumption that the results of understanding by the machine (SUS) can be evaluated according to the rules applied for evaluation of human understanding. Machine understanding will be defined in Chap. 4 in the context of both human understanding and existing systems that can be regarded as the simplest understanding systems. This chapter is not intended as a survey of literature on the vast topic concerning understanding, but rather as a presentation of the point of view of selected thinkers on this topic and a discussion of some aspects of understanding considered to have implication for material presented in other chapters of this book.

Machine understanding, the term introduced by authors, is referring to the new area of research the aim of which is investigation of the possibility of building of the machine with ability to understand the world or the language. In Chap. 2 the point of view of selected thinkers on the topic concerning understanding was described whereas in this Chapter a short survey of the existing systems that can be regarded as the simple understanding systems is presented.

Machine Understanding is the term introduced by authors to denote understanding by a machine and is referring to the new area of research the aim of which is to investigate the possibility of building the machine with the ability to understand. A new research area such as machine understanding needs the framework within which the problems will be formulated and solved and the machine, in order to be able to understand, needs to imitate the way in which humans understand the world and the language (text). Machine understanding that denotes understanding by the machine (shape understanding system—SUS) stresses the dependence of learning and understanding processes. In this chapter machine understanding is defined in the context of both human understanding and the existing systems that can be called understanding systems.

Machine understanding is referring to the new area of research the aim of which is to investigate the possibility of building a machine with the ability to understand. The term machine understanding, introduced by authors, denotes the process of understanding by the machine Shape Understanding System (SUS). Machine understanding stresses dependence of learning and understanding processes. A machine, in order to be able to understand, needs to imitate the way in which humans understand the world and language (text). SUS as the machine that is designed to have an ability to think and understand needs to learn both knowledge and skills. Learning knowledge and skills that supplies material for thought leading to understanding is called knowledge implementation [see book [103]) for description]. In the previous chapter machine understanding was defined in the context of both human understanding and the existing systems that can be called understanding systems. In this chapter a general outline of machine understanding that is based on the shape understanding method is presented.

Machine understanding refers to the categorical structure of the learned knowledge. In our previous books [102, 103] the visual object category, the sensory object category and the text category were described. Some of these categories that are relevant to the material presented in this book are briefly outlined in the following chapters. In this chapter the abstract categories and their relations to SUS ‘intuition’ are described. The abstract category is introduced based on the assumption that concepts formed during SUS understanding process are the result of perceiving the visual objects. In philosophy there is a view that all ideas formed in the mind are the result of the sensory impressions and that the basic ideas (concepts), the result of the faculty of mind called intuition, are formed based on the impression that comes from the abstraction of the sensory material. Following this philosophical finding, SUS “intuition” is related to the SUS perceptual visual field that is the fundamental basis of the basic abstract categories. The basic abstract categories are related to the abstract categories that are defined in the different areas of science such as mathematics, theoretical physics or chemistry and become the important part of understanding process.

Machine understanding is the term that refers to the new area of research the aim of which is to investigate the possibility to build an understanding machine. Nearly all activities connected with understanding can be regarded as problem solving and for this reason machine understanding can be regarded as a problem solving. Machine understanding is based on the assumption that the result of understanding by a machine can be evaluated and compared to the result of human understanding. If understanding is defined as the ability to solve problems, then assuming that problems (tasks) are well defined the understanding can be tested by testing whether these problems can be solved by the machine (SUS).

Visual understanding is the part of machine understanding that refers to understanding of objects from the visual object category. Visual understanding regarded as the problem solving involves naming and recognizing the visual objects, or solving visual problems. In this chapter naming objects that are members of the real world object category is presented.

Machine understanding can be regarded as problem solving in this context understanding means finding a solution to a problem. Machine understanding aimed at understanding objects from many different categories of objects described in [102, 103] involves understanding objects from the category of visual objects, the category of sensory objects or the category of text objects. Understanding a visual object means solving problem of naming/recognizing of the objects. In Chap. 8 naming objects that are members of the real world object category was presented. In this chapter naming objects that are members of the sign category will be described. Some issues connected with learning naming of objects from the sign category were presented in [102, 103]. In this chapter understanding signs is regarded as problem solving connected with finding meaning of these signs.

Understanding a text, in the context of machine understanding, refers to understanding objects from the text category. The text category T, members of which are composed from the basic linguistic elements—the letters—members of the letter category, was introduced in the previous our book [103]. The letter category, described in the previous chapter, is in some aspects similar to the visual symbol category (derived from the sign category). However, understanding objects from the letter category is different from understanding objects from the visual symbol category such as the electronic symbol category. For this reason instead of understanding the letter category machine understanding refers to understanding the text category, members of which are composed from the letters.

A machine in order to be able to understand needs to some extent mimic human understanding and for this reason machine understanding is based on the assumption that the results of understanding by the machine (SUS) can be evaluated according to the rules applied for evaluation of human understanding. The essential part of evaluation of the machine (SUS) ability to understand is to formulate problems and to use these problems to test if the machine (SUS) is able to solve these problems. However, while the ability to solve the problem by a machine can to some extent prove that the machine can understand, there is also a need proving this by testing the machine ability to explain how to solve the problem or to explain the causes, context, and consequences of given facts. Explanations can be often confused with arguments and when arguments try to demonstrate that something is, will be, or should be the case, explanations attempt to reveal why or how something is or will be. When arguments are referring to knowledge and its aim is to enrich the knowledge, explanations is referring to understanding and make contribution to understanding. Also explanation is often confused with justification and when justification is the reason why belief is properly holds, the explanation is the reason why the belief is true and statements which are justifications of some action can take the form of arguments.