An artificial intelligence system is a system that operates with knowledge about the problem area. Without a knowledge base, artificial intelligence systems do not exist. For the formalization and representation of knowledge, special models of knowledge representation and languages for describing knowledge are developed, various types of knowledge are distinguished.
Models of knowledge representation refer to the pragmatic direction (based on the assumption that human mental activity is a “black box”) in research on artificial intelligence. At the same time, in knowledge-based systems, information structures are presented in the form of declarative (descriptive) knowledge, and algorithms and heuristics - in the form of procedural knowledge.
It is recognized that it is advisable to develop knowledge-based systems with a selection of a number of subsystems. The main ones are:
- declarative knowledge presentation subsystem;
- procedural knowledge presentation subsystem;
- knowledge manipulation subsystem (logical inference mechanism).
The purpose of this separation is the possibility of using the system in various subject areas. To do this, it is enough to replace the declarative knowledge presentation subsystem without changing the other parts. In fact, having a universal logical inference mechanism, it suffices to fill the system shell with specific rules and facts from the required subject area in order to get the system ready for use.
The history of the development of knowledge engineering methods indicates that this idea has been a fundamental area of research for a long time. For example, A. Newell and G. Simon from Carnegie Mellon University (USA) have developed a GPS program - a universal task solver. In formal logic, a resolution method was developed. In the 70s. XX century., However, most researchers have concluded that there is no universal system for the presentation of procedural knowledge, suitable for all areas of activity. This is confirmed by the high value of expert experts in narrow subject areas. The problem of knowledge representation was formulated.
The problem of knowledge representation lies in the inconsistency between the information about the dependencies of this subject area, the specialist’s methods, the methods used to solve problems, and the possibilities of formal (uniquely limited) presentation of such information in a computer. Often the problem is complicated by the difficulties for the expert in formulating the knowledge that he has in explicit form.
The general problem of knowledge representation includes a number of particular problems:
- presentation of declarative knowledge as data endowed with semantics (facts);
- presentation of procedural knowledge as a relationship between model elements, including in the form of procedures and functions;
- representation of meta-knowledge — rules for processing facts, ways of organizing inference, methods for solving problems by the user, generating new knowledge, etc.
Comments
To leave a comment
Knowledge Representation Models
Terms: Knowledge Representation Models