Why are we developing a knowledge-based system

Knowledge based systems


(1) Definition: Knowledge-based systems are used for tasks that usually require human intelligence. To do this, they use the techniques and methods of artificial intelligence, which map human cognitive processes through software. They are characterized by the independent, purpose-oriented processing of networked information by the computer to generate or apply knowledge. Knowledge-based systems are used in business administration for decision support, for classification and evaluation tasks as well as for forecasting and consulting activities.
(2) Types: Expert systems are knowledge-based systems for decision support that depict the advisory and problem-solving skills of human experts. The competence of at least one human expert in a delimited specialty area is modeled in a knowledge base; a problem-solving component applies the knowledge to the task. Case-based reasoning systems process tasks on the basis of known solutions for similar problems by comparing the “new” problem with cases that have already been saved. With newer knowledge-based techniques of Sofi Computing, methods are being developed that are particularly suitable for complex problems that cannot be solved or can only be solved to a limited extent with conventional mathematical and combinatorial methods due to uncertainty and only partially available information. Fuzzy logic systems extend the classic two-valued logic. In classical logic, a property either applies completely or not at all, a statement is either true or false. Fuzzy logic systems, on the other hand, allow values ​​between true and false in order to be able to include fuzziness and uncertainties in the modeling process. Artificial neural networks (ANN) are systems that work in parallel, the structure of which is based on the functionality of the human brain. An artificial neural network is not programmed, but parameterized with training data. It delivers results even with problems that are difficult to formulate algorithmically. Genetic algorithms map evolutionary mechanisms with the help of mathematical algorithms. They imitate fundamental principles of evolutionary theory, e.g. natural selection or mutation. Genetic algorithms are used, for example, in the banking sector to generate rules in stock market trading, to support hedging decisions, to evaluate portfolios and to evaluate options. See also knowledge management (with references).

Literature: Bodendorf, F .: Data and Knowledge Management, Berlin 2005.

(English knowledge based system) A knowledge-based system is understood to be a system in which knowledge is represented and processed in a suitable form. In connection with a problem-solving component, which is intended to provide a user with the problem-solving skills of a human expert, one speaks of an expert system. Knowledge is generally understood to be cognitively processed and “understood” information, especially in the sense of knowledge of (effect) relationships. Appropriate knowledge is used in the operational context to prepare and carry out decisions and actions. In this sense, special knowledge-based systems can be used in operational use, especially as components of decision support systems. Knowledge can be divided into general and specialist knowledge. General knowledge is not based on an immediate area of ​​responsibility. While the presentation of such knowledge (“common sense”) as well as uncertain knowledge has so far only partially succeeded, the formalization and processing of knowledge is possible within certain limits if it is restricted to specialist areas. The lofty approach in the 1960s and 1970s to develop systems that should have general, intelligent problem-solving abilities (“General Problem Solver”, in connection with the then dominant research approach of symbolic knowledge processing of “Artificial Intelligence”) can, however, initially have failed be considered. Previous implementations of knowledge-based systems are therefore mostly limited to limited areas. In this context it should be noted that the attribute “knowledge-based” is used inconsistently and in some cases in an inflationary manner. The justification for using the corresponding attribute is often doubtful (depending on the definition and delimitation of the terms data, information and knowledge). Knowledge can still be divided into declarative and procedural knowledge. While symbolic descriptions of facts are reproduced through declarative knowledge (“what”), procedural knowledge is more likely to depict procedures (“how”). A primary task in the design of knowledge-based systems is knowledge acquisition (knowledge acquisition); appropriate system interfaces are required for this. Closely related to this is the problem of a suitable representation of knowledge; various calculi are used here, in particular logic formalisms (e.g. predicate logic, rules such as “if A, then B”). In addition to such application-related knowledge (knowledge base), knowledge-based systems i. d. Usually an application-independent inference mechanism. Such an inference mechanism represents a problem-solving component by means of which knowledge is processed and, if necessary, conclusions can be derived. Areas of application of corresponding expert systems are, among others. in the field of diagnostics or the configuration of complex systems. For example, a simple knowledge-based system can be implemented using the Prolog programming language. Prolog allows the specification of rules and facts and has a simple inference mechanism. In this context, reference should also be made to the “constraint programming” approach: the specification of problems with declarative specification of rules, conditions and objectives, along with a generic solution method.

Previous technical term: knowledge barriers | Next technical term: knowledge base, organizational



Report this article to the editors as incorrect & mark it for editing