Preamble


Analysis states of explored question


Definition of expert systems


Declaration of knowledges in SSAD


Notion about efficiency and development of expert systems


Generalization of scientific search results and analysis


Conclusion

Support of acceptance of decisions on the issue of specialists in DNTU

Supervisor: assistant professor, G.V.Мокry

Preamble

     Computerizing of society is one of basic directions of scientific and technical progress - caused the substantial changes in technology developments and uses of programmatic facilities.
     These changes were prepared by all development of theory and practice of artificial intelligence (AI), the most substantial result of which there was transition to the so-called new information technology and creation of consulting models (ES).
     First ES are medical mycin and dendral for support of acceptance of decisions of appendixes on chemistry appeared in the middle of 70th within the framework of the research programs on artificial intelligence. Already first ES appeared useful. The medical system of mycin successfully blended in clinical practice, helping in the choice of medications to the patients with meningitis, cystitis.
     Essence of happening technological changes consists in appearance of new class of the AI tools, which became basis of creation of the finished programmatic goods on the basis of on principle other technology, with new high-quality possibilities of the created products. These changes substantially promote the intellect of the programs, new facilities replaced a whole technological chainlet, in which between an end user and COMPUTER there were a few mediators.
     Historically greater part of investments in corporate computing facilities was related to the systems which generate or fix information, for example, in a record-keeping, treatment of orders, production and systems of information about clients. Now organizations inlay all more of resources in the appendixes and technologies which allow to extract an additional benefit from the such collected information. For this purpose more frequent than all come running to creation of depositories of information (data warehousing) operative analytical treatment (OLAP - Online analytical processing). These terms engulf the process of collection, cleanings and sifting of information from different working systems, and also grant of wide audience of business-users of access to obtained information for the analysis and preparation of reports. The permanent depositories of the cleared and systematized information, accessible for viewing by users, are named the depositories of information or windows of information (data marts).
     OLAP gives to organizations the most flexible and productive facilities of access, viewing and data related to business analysis. Foremost, OLAP represents information to the users by a natural, intuitional data model. Due to lightness of moving from data business-users can more effectively look over and analyse information from the depositories of information, that allows to organizations better to realize the value of these information. Secondly, OLAP accelerates delivery of information to the users looking over such multidimensional structures. To that end preparation of some calculated values in a data array is carried out beforehand, instead of during implementation. Combination of lightness of moving and high performance helps users to look over and analyse information much quick and more effectively, than it would be possible only on the basis of technology of relation databases. As a result they dedicate more of time to the data analysis and less - analysis of databases.
     So by appearance, appearance of consulting models and OLAP-appendixes on the whole gave a comfortable and very effective mechanism for acceptance of decisions, analysis and prognostication.



1. Analysis states of explored question

     Estimating the state of the examined questions as it applies to Donetsk National Technical University, which development is conducted in the conditions of, it is necessary to abolish, that to this time on the base of different programmatic and vehicle facilities the informative system of the on-line data processing is developed, supporting all types of basic and auxiliary document circulation related to work of Institute of higher. Also this system allows substantially to simplify all processes, related to the conduct of the state of employees, contingent of students, university entrants, material values. However much this system gives no analytical information, or its receipt is related to the heavy tolls of time and labour intensive of realization.
     However during creation and perfection of this computer system the large enough volumes of information, allowing to get not only on-line data for the prosecution of concrete daily needs of Institute of higher, were accumulated but also by auxiliary facilities to produce the analysis and prognosis with the purpose of acceptance of decisions in relation to one or another questions touching the increase of efficiency of organization of educational process. In theory as such auxiliary facilities usually come forward Microsoft Excel, Microsoft Access, Crystal Re-ports. These appendixes have powerful enough instruments allowing to process and represent operative information in a necessary format, and in addition, an experimental specialist can also conduct the accumulated data analysis. However in enough plenty of cases the decision the problems set by guidance and analysts by the transferred programmatic facilities results in the difficulties linked, from one side, with complication, and sometimes and impossibility of selection of necessary information, and from other side, with the necessity of presentation of them in a kind comfortable for the analysis, that is not always possible.
     Such complications related to analytical treatment of large volumes of information arise up, foremost, from heterogeneity of present information - because of stage-by-stage development of the informative system of establishment there are subsystems, information in which is kept in different formats, such as Microsoft SQL Server, Microsoft Access, DBF. Except for it, information is represented in one format, for example, databases of SQL Server, located on different servers frequently territorial separate. Also for the adequate analysis external information is frequently required (I.e. being not contained in internal DB of Institute of higher, which can have an unforeseeable format.)
     Thus, the situation folded with analytical treatment of information does not respond to request, by the pulled out specialists-analysts and leaders of establishment now can not provide the decision the set problems on the data analysis.
     Hereupon, there is the necessity of creation of the system of the analytical present data processing which in the end must result in development of System of Support of Acceptance of Decisions (SSAD), as one of the most powerful instruments of analysis of information. Planning and creation of depository of information must be the first step of realization of such system, able to unite and prepare the separately kept information to further treatment.



2. Definition of expert systems

    
     A consulting model (ES) is the intelligence system, using the accumulated knowledges for providing of high-effective decision of tasks in a narrow professional region. Consulting models behave to the systems of support of acceptance of decisions (SPPR), based on knowledges. Traditional SSAD universal and is used for the decision of unique problems in different subject domains, and ES is given by answers for questions in a narrow subject domain and build the conclusions which would be done by the man-professional of high qualification. Integration of traditional SSAD with ES forms more difficult kind - so-called expert system of support of acceptance of decisions (ESSAD). Such system, coming the common requirements produced to ES from, must explain the advices to the end user, and, in addition, give it universal facilities of free design. Thus, the just the same variant of organization of the system will be examined in master's degree work, because it gives the most rational approach to the receipt of effective analysis.
     In general case information in the system of support of acceptance of decisions is the isolated base, information of which is got either from an expert - man or it is simple as the statistical information collected by set rules in the flow of some interval of time. A programming language and supporting software package serve as the tools of the ES construction, used for the ES creation.
     Are ES the real practical appendix of artificial intelligence, that is underlined by another determination: ES is the informative system, based on knowledges of certain complex subject domain, executing the role of expert-consultant for end users. Vehicle, programmatic and still human resources are the resource components ES.
     Vehicle resources consist of the off-line microcomputer systems, and also microcomputer work stations and terminals connected to minicomputers or large computers through a telecommunication network. Difficult ES is sometimes developed on the powerful computers of the special setting, consulting models or languages directly projected for the front-rank programmatic packages of development or programming languages LISP or PROLOG.
     Programmatic resources are the mechanism of input-conclusion and presentation of information, and also mean for work with knowledges and for communication with end users. The programs of receipt of knowledges are not part of consulting model, and are programmatic facilities only for development of knowledge base. A knowledge base in the developed system is based on conception of depository of information. By this conception it is possible simply enough to agregate information from different sources, different format, size and method of presentation. On the basis of depository of information the programs-shells of consulting models without its kernel (basic maintenance) are comfortable facilities of the ES development, I.e. without its knowledge and facts bases. Confessedly also, that to the end users and experts large practical possibilities of the ES creation give facilities of DB (Microsoft SQL, Oracle etc). Especially useful at the SSAD development, based on the depositories of information, relyatsyonnye SUBD supporting multidimensional presentation of information. Facilities of DB are named the also limited generators of support of acceptance of decisions, as they give a few basic analytical instruments to the user (queries of "if-then", correlation-regressive and some other types of statistical analysis, optimization, construction and analysis of trends). For the decision of intricate problems of analysis and prognosis of these facilities not enough, in this connection there is the task of development or selection of needed methods.
     Still human resources. When is created large ES, a knowledge base and process of examination is usually designed by an engineer on knowledges from facts and rules given by an expert. ES gives recommendations to the end user. Experts and end users can be and itself by engineers on knowledges, if is able to use the programmatic shells ES or intellectual possibilities of spreadsheets.
     On the spheres of the ES use of them it is possible to divide into productions and administrative. Production give ES expert conclusion on the production process control, administrative - help managers to make decisions. The variant of the combined system will be examined in this work, because a managing personnel needs both the decision of tasks on organization of educational process and possibility of receipt of analytical information on certain questions.



3. Declaration of knowledges in SSAD

     In a knowledge base formalized knowledges are kept in some coded kind. On the modern stage of the ES development a few forms of knowledge representation are used. We will select from them four basic:
     1."Three" object-attribute- value, for example: home- color- green; patient- temperature- high. This form of knowledge representation determines a "object", possessing some attributes (by properties) which can take on values from a known set.
     2. Rules of production in a kind: If a patient is patient with a flu And stage of disease initial, a temperature is high with probability = 0.95 And head pain is with probability = 0.8.
     The rule of products consists of two parts: parcels (IF) and conclusions (), each of which consists of kon'yunktsyy of claims of more low level of working out in detail.
     3. Frame. It is the named table with the several of slotы- cells having the names and recipient in the process of work of machine of conclusion some values. As the values constants, references to the frames of more high or more low level, and also some calculable procedures, can be present.
     4. Symantic network. It is the oriented count, the tops of which correspond to the objects (to the events), and arcs describe the relations between tops.
     First from the indicated forms of knowledge ("three") representation is the most early form, the sort of transitional, from presentation of information to knowledge representation. The best application "тройки"- domain diagnostic ES in a subject domain with plenty of simple-clusterizeided objects, in every class of which there is plenty of general attributes. Such region diagnostics of the difficult technical systems can exemplify.
     Knowledge representation by the rules of products is the most widespread form of the KB realization. By produсing rule it is possible to describe practically any system of knowledges. On them based all early ES, such as MYCIN, and also swingeing majority of modern ES in different subject domains. In accordance with this form of the KB realization most systems are done - "shells" and many instrumental environments. On the whole the products form of knowledge representation is natural and comfortable for formalization of the knowledges got at an expert.
     The Frame structures up to a point became development of method of spreadsheets in area of treatment of knowledges. This second on frequency of met in concrete realization the ES form of knowledge representation. It is most reasonable to apply the frame of - structure in a subject domain with the expressly expressed hierarchical structure.
     Semantic networks initially arose up as model of of long duration human memory in psychology. Large interest to work with semantic networks is professed by the developers of the systems of co-operation in natural languages.
     On the modern stage of works in area of ES the mixed form of knowledge representation is used quite often. Such ES are named hybrid. Makes sense to take advantage exactly of this conception, because plenty of the decided tasks requires a different knowledge representation.



4. Notion about efficiency and development of expert systems

     ES well work out narrow specific problems in the certain sphere of knowledge, but lose in the decision of tasks requiring a broad mind. ES effective at the decision of analytical tasks. For example, ES helps a financial consultant on investments, giving out recommendations for clients. However much ES can not estimate the nuances of current affairs, economy, social development or conduct of user. These important factors must be analysed by a consultant-man. At acquisition or the ES development it is necessary to be comprehend and compare advantages of consulting model and expense on her. Cases are possible, when a man-expert decides a task for a few minutes, and the ES creation requires creations of several hundred rules and a few months of planning.
     Usually large ES is developed by engineers on knowledges by the method of prototyping, I.e. by the gradual approaching from a sketch to the ultimate goal.
     An engineer on knowledges is a professional, working with experts in the search of knowledges (facts and heuristics) which they process. An engineer on knowledges builds a knowledge (and if necessary and all ES) base and must be able to work with experts in many subject domains. Abroad the ES shell relatively undear. They help an end user to develop own consulting models. Some shells use the format of spreadsheet, facilitating development of the IF - THEN rules.



5. Generalization of scientific search results and analysis

     Thus, as a result of search and analysis conducted at implementation of course project, I selected points, which further development, hardwares for realization programmatic and are chosen, and also conceptual ground, will be conducted on.
     A task can be decided with the use of different programmatic facilities and platforms. Different their representatives differ by the analytical resource given to the user kit, tool, format of data storage, possibilities of administration. The cost of the licensed version of product is the not last factor. Therefore in optimum correlation price-productivity it should be noted the Microsoft SQL Server 2000 platform. His built-in facilities of analysis and treatment are a good database for construction of the system. It should be noted that as compared to some other facilities of the OLAP(for example, Darvin, Pilot Dis-covery, Idis, ORACLE Express) design, Microsoft SQL Server 2000 possesses in a number of failings, such as insufficient methodology of booty of knowledges, bad support of rare table of fact, plenty of redundant information. However at all of it this platform is simple in the use, allows well to co-ordinate information from different outsourcings, possesses a high performance and reliability, and main, the cost of the licensed version of her is much below than that ORACLE Express that, undoubtedly, it is important in the conditions of the weak financing of programmatic separations of Institute of higher.
     In this connection the search will be conducted in the direction of methods of analysis, which do not give or give the standard facilities Microsoft SQL Server 2000 in an incomplete measure. The methods of search of knowledges and exposure of the hidden conformities to the law, specific for the explored question, will be developed similarly.



Conclusion

     As a result of implementation of course project were got me following results:
- the ground of necessity of the SPPR introduction to the structure is given to the ACE ВУЗа
- the conceptual providing is arranged for future work
- an elementary technical base for development
is chosen and grounded - basic conceptions of the analysis данных
are chosen and grounded - considered and methodology of given is analysed problems
- the further directions поиска
are certain - the specific of given is reflected задачи
On the whole implementation of course project laid the foundation for further continuation of research and search, defined basic priorities and tasks of future development.