Яндекс.Метрика Рейтинг@Mail.ru
KARKAS: a toolkit for creating a knowledge base
KARKAS: the shell for creating a knowledge base

Registration »         Forgot password »

04 Август 2022

07 Февраль 2019
MODEL OF HIERARCHICAL FUNCTIONAL SYSTEM FOR CLUSTER ANALYSIS Collection of Scientific Papers of KhNUPS 2(56). Kharkiv, 2018, p. 82 - 88. The model of the hierarchical functional system of the subject area...

24 Апрель 2015
Laboratory workshop on the system "KARKAS" / Computer Based Training Contains laboratory work, the purpose of which is the practical development of skills in the construction of knowledge bases by users using a tool...

Artificial intelligence

Paradigm - a set of concepts, starting points accepted and disseminated by the scientific community

The concept of intelligence from the standpoint of computer science can be characterized by the following properties as the ability to solve complex problems, as the ability to learn, generalize and analogy, as the ability to interact with the outside world through communication, perception and awareness of what is perceived. The material carrier of intelligence is the human brain.
With the concept of artificial intelligence (AI), hopes are associated with the creation of a thinking computer that can compete with the human brain and possibly surpass it. Since the number of neurons in the human brain can reach up to one trillion, simply copying a biological neural network does not make sense.
It is undeniable, but the fact is that intelligent systems - knowledge-based systems (KPS) are already being implemented and widely used in the practice of human activity. These are fairly well-known expert systems, machine translation systems, neurocomputers, robots, games that have every right to be called intellectual. Although the level of development of these systems does not allow them to pass Alan Turing's test formulated in 1950: a computer can be considered intelligent if it is able to make us believe that we are not dealing with a computer, but with a person.
Well-known futurologists based on emerging modern industries such as nanotechnology, biotechnology, global information networks (Symantec Web) predict that by 2020-2030 artificial intelligence will appear and computers will surpass their creators.
Currently, there are two main directions in the development of AI systems. The first model simulates the work of the human brain and is implemented in the creation of neural networks, the so-called strong AI. The second is related to achieving a good match between the results of the work of natural and artificial intelligent systems, and it does not matter how this is achieved by the so-called weak AI. This direction is connected with the implementation of reasoning presented in an explicit symbolic form.
An expert system (ES) is a computer system that allows, on the basis of a knowledge base compiled by experts from a specific subject area, to solve a problem with the help of a logical conclusion.
An expert learning system (ETS) is a computer system built on the basis of the knowledge of subject matter experts (qualified teachers, methodologists, psychologists) that implements and controls the learning process. The purpose of such a system is that, on the one hand, it helps the teacher to teach and control the student, and on the other hand, the student learns independently.
One of the disadvantages of ES and EOS is that the knowledge base (KB) is replenished by an expert or knowledge engineer (cognitologist).
Self-learning intelligent systems are based on the fact that the knowledge base is replenished from the accumulated experience of the system. Such systems are based on the methods of clustering situations from real practice, on the methods of inductive learning (learning by examples), on finding solutions by analogy from the database (decision-making based on precedents).
The paradigm of intelligent multi-agent systems is based on the ability of such systems to develop and communicate in accordance with objective changes in the subject area.
The need for intelligent multi-agent systems arises when the subject areas they support are constantly evolving. They must meet a number of specific requirements:
adequately reflect the knowledge of the subject area at any given time;
be suitable for easy and fast reconstruction when the subject area changes.
The concept of agents, developed within the framework of multi-agent technologies and multi-agent systems (MAS), assumes the presence of activity, that is, the ability of the program to independently respond to external events and choose appropriate actions.
Today, agent technologies offer various types of agents, their behavior models and properties, a family of architectures and libraries of components that are oriented to modern requirements, such as, for example, distribution, autonomy.
On the one hand, we are talking about open, active, developing systems, in which the main attention is paid to the process of interaction between agents as the reason for the emergence of a system with new qualities.
On the other hand, multi-agent systems can be built as a combination of dynamic ES, which can operate both collectively and separately.
Agents can work as non-interactive individuals or as a collective. In the first case, the system is very simple: agents do what they are asked to do (passive agents). In the second case, agents need their interaction (active agents).
To build a MAC, you need a toolkit consisting of two components: development tools; runtime environment.
The rules of behavior of agents are described by a production, to which is added another component of the time of its application (antecedent - consequent - time).
One of the factors of interest in MAC was the development of the Internet. To function successfully in such an environment, agents must

2012 г. it-karkas.com.ua
Vladimir Burdaev
Info for users
Site map


Copyright 2012 - it-karkas.com.ua
Website development - alphastudio.com.ua