Expert systems

Expert systems represent a direct example of how the science can be useful to business.

Development of ES

In the early eighties in researches on artificial intelligence the independent direction which received the name "expert systems" (ES) was created. Development of programs which at solving of tasks, difficult for the expert-person, receive the results which are not of a lower quality and efficiency to the solutions received by the expert was the purpose of researches in this new direction. For designation of this discipline also often use the term "knowledge engineering" entered by E. Feygenbaum as "introduction of the principles and tools of researches from area of artificial intelligence into the solution of the difficult applied problems requiring knowledge of experts". For  1960 — 1985 progress in business of mastering of the artificial intelligence (AI) concerned generally research developments which showed suitability of AI for practical use. In  1988 — 1990 expert systems began to be applied actively in commercial applications. At the beginning of emergence the programming languages used for their creation, technologies of application development and used did integration of ES into traditional program systems quite difficult, and at times even by an impracticable task. However now development tools of ES are used in full accordance with current technology trends of traditional programming that solves the problems arising during creation of compound applications.

The place in an IT infrastrkture

Appointment

Scheme of static Expert system

The name "Expert systems" means a possibility of replacement of the expert-person with the software solution. It allows the enterprises to reduce labor costs of specialists, and to specialists to address at the solution of any questions within the activity directly the program. Such opportunities reduce time of solution and allow young specialists to study directly at the workplace. Virtual "assistants" in software packages of operating systems of computers can be an example of the simplest expert system. Such algorithms of the solution of standard questions save developers from excessive, unreasonable and unjustified loading on communication with the end user.

Expert systems and the systems of artificial intelligence have the main difference from data processing systems the fact that in them are generally used a character method of representation, a character output and heuristic search of the solution. Expert systems are intended for the solution only of difficult practical tasks. On quality and efficiency of the solution expert systems should not concede to solutions of the expert-person. Solutions of expert systems. can be explained to the user at the qualitative level, i.e. have transparency. The transparency of expert systems is provided with their capability to argue on results of the work and knowledge bases. Important property of expert systems is also that they are capable to study. ES solve problems:

• interpretations
• predictions
• diagnostics
• planning
• designing
• control
• debuggings
• instructing
• managements

Such tasks arise in the most different areas scientific, business and industrial areas. The software tools based on technology of expert systems gained considerable distribution in the world. Importance of expert systems is as follows:

• significantly expand a circle of almost significant tasks which solution brings considerable economic effect
• are the most important means of reduction of duration and, therefore, the high cost of development of difficult applications
• consolidation of ES technology with technology of traditional programming adds new qualities to software products due to ensuring dynamic modification of applications with the user, but not programmer, bigger "transparency" of the application, the best graphics, the interface and interaction.

Unformalized tasks

Special attention should be paid to unformalized tasks because for their solution and expert systems were created. Unformalized tasks usually have the following properties:

• inaccuracy, ambiguity, incompleteness and inconsistency of initial data
• inaccuracy, ambiguity, incompleteness and inconsistency of knowledge of a problem area and solvable task
• the big dimension of space of the solution, i.e. search by search of the solution can be very big
• dynamically changing data and knowledge

Unformalized tasks represent big and very important class of tasks. Tasks of such plan are the most mass class of the tasks solved by a computer.

Architecture client-server

There are work benches of artificial intelligence supporting distributed computing on architecture the client-server. It provides the following advantages:

• reduction in cost of the equipment used in applications
• an opportunity to decentralize applications
• reliability augmentation and overall performance
• reduction of amount of information, sent between the equipment

Advantages

There are several advantages of expert systems both before the person operator, and before normal algorithmic databases:

• integrability. There are work benches which are easily a part of other information technologies and means
• openness and portability: they have no prejudices and they are steady against different noises;
• lack of hasty outputs;
• issue of an optimal solution
• unlimited sizes of the knowledge base.
• permanent data storage: the expert can  never forget something, the machine —.

Development perspectives

According to leading specialists in the field of programming, in near perspective of ES will play an important role in such spheres as:

• all design phases, developments, productions, distributions, sales, supports and rendering services
• application integration from the ready intelligent interacting modules in the main large software solutions
• solution of unformalized tasks

Connected subjects

• Expert systems (Architecture)
• Development of expert systems
• Expert systems (representation of knowledge)

Links

• Consulting and trainings of Moscow
• Online Magazine
• Martelk