Knowledge Engineering Technology

In the usual practice of psychodiagnostic research, the interpretation of results and the writing of a psychodiagnostic conclusion is a creative process carried out by a psychologist after the routine processing of the obtained data. Moreover, the more experience a psychologist has, the more accurate, concrete and adequate a “portrait” of the subject's personality is obtained. The development of an automated psychodiagnostic conclusion involves modeling the psychologist's reasoning in interpreting the test results and “transferring” the knowledge and experience of the psychologist to some structures perceived by a computer. The main difficulties in carrying out such works are in the issues of extracting the knowledge and experience of the psychologist, their structuring and presentation in some form understood by the computer.

The direction of research, which originated in the framework of artificial intelligence and was distinguished into an independent discipline related to the issues of extracting, structuring, forming, processing and acquiring knowledge, is called Knowledge Engineering . The subject specialist whose knowledge and experience is “extracted” is called an expert, and the specialist who extracts and structures knowledge is a knowledge engineer.

Knowledge extraction refers to the procedure of interaction of a knowledge engineer with a knowledge source (expert, literature, etc.), as a result of which the process of specialists' reasoning when making professional decisions and the structure of their ideas about the subject area become apparent.

The problem of extracting expertise is traditionally considered a “bottleneck” in the design of intelligent systems. In terms of the number of citations in the literature, this maxim has already become classic and rather accurately reflects the state of affairs over the past few years. In order to answer the questions, what are the difficulties of extracting expert knowledge and how to solve them, specialists in the field of knowledge engineering turn to psychological and psycholinguistic research related to such issues as expert examination by experienced specialists and understanding of the nature of expertise, the application of theoretical knowledge in practice and the specificity of practical thinking, the characteristics of a person’s information processing system and the behavior of a person solving professional tasks, as well as the interrelation of natural language and human thought process. The study of these problems, as well as practical work on the extraction of knowledge, allows to note a number of effects or phenomena observed when working with experts and having a significant impact on the process of knowledge extraction. These include:

• the existence of so-called silent or implied knowledge (tacit knowledge), due to three reasons: the unconscious nature of expert skills, the difficulty of the verbalization process and the underestimation by the expert of the importance of some knowledge used in solving professional problems;
• a special form of organization of knowledge of experts compared to the organization of knowledge of beginners;
• rather high speed of solving professional tasks by an expert;
• limited amount of short-term human memory;
• incorrectness of some methods of obtaining information (assignment of weights of signs, assignment of probabilities, etc.);
• the existence of a cognitive protection mechanism;
• the presence of psycholinguistic problems, in particular, the difference between the communicative language and the language of the thinking process leads to a significant loss of information in the process of communication of the knowledge engineer with the expert.

To overcome the problems described above in the framework of knowledge engineering (both in theoretical terms and as a result of practical development of specific systems), so-called knowledge extraction techniques or knowledge elicitation techniques are created, representing a certain procedure or form of interaction between a knowledge engineer and an expert. .

Under the conceptual analysis of knowledge (or structuring) is understood the process of analyzing information received from a source of knowledge, and its synthesis (or coding) into some structures that are independent of any software implementation.

The extraction and structuring of knowledge is a single process of interaction of a knowledge engineer with an expert, differing only in its aspects. If the problem of extracting knowledge is to choose the form of interaction of a knowledge engineer with an expert, then the problem of conceptual analysis of knowledge involves considering the actual subject for which such interaction is organized, namely, the information, the meaning of which in the process of transformation from expert to computer representation should stay the same.

The process of transforming knowledge from expert to computer representation can be viewed as a problem of transforming information through the transition from one material knowledge carrier to another. Depending on the type of material carrier of knowledge, the following levels of knowledge representations can be distinguished:

• representation of knowledge in the memory of a person (expert);
• conceptual or semi-formalized presentation of knowledge as a result of the interaction of an expert and knowledge engineer (possibly on paper);
• formalized presentation of knowledge in specialized languages ​​of artificial intelligence (on paper or in computer);
• representation of knowledge on machine-readable media (knowledge base).

In this regard, the conceptual analysis of knowledge is a transition from knowledge representation in the memory of an expert to some conceptual, semi-formalized representations or structures, reflecting the understanding of the expert of the subject area as a whole and reasoning strategies in solving professional problems.