–усска€ верси€ этой страницы
To the Start Page
The Chair of Mathematical Theory of Intelligent Systems and
Laboratory of Problems of Theoretical Cybernetics
of Moscow State University Official Website
To the Start Page News MaTIS Chair Staff Science Teaching Research Magazine Culture Full-text Search

Staff :: Strogalov Alexander Sergeevich :: Publications of A.S.Strogalov

Modeling educational process using expert system.

Kudrjavcev V. B., Waschik K., Strogalov A. S., Alisejtshik P. A., Peretruchin V. V.




This work deals with the problem of computer simulation of the process in not enough formalized subject's fields. Some requirements are settled to educational computer systems and there is given a description of the main units of the real acting educational system "IDEA for Windows" developed on the theoretic investigation results of authors' association


1. Introduction


Existing computer educational systems are basically a computer compilation of theoretic courses and exercises to them, having in the most advanced versions appropriate audio- and video effects for illustration. In that systems the main part is assigned to user. He is to learn subject step by step according to system's instructions. Thus the role of computer educational system can be characterized as a passive one. It is only the sum of knowledge in fact and this is a considerable work for user to learn it. Practically the modeling training process consists of three main parts: the teacher model, the learner model and the area of their interaction, for example, the textbook with exercises. In existing computer training systems the roles of teacher and learner are weak and the main role belongs to textbook with exercises and a set of instructions that indicate how to learn. On the first stage of the computer models and training means development these systems had played the posi! tive role and it was hardly to require from them to simulate widely the training process including also the parts of teacher and learner.

For "IDEA" project [1] the main task was to create such computer educational system that should simulate all three parts of educational process: teacher, learner and training material. It should include means for creation models of teacher and learner, introduce a text-book with exercises as a set of definitely organized academic material with multimedia elements. On this base it should be simulated the process of real education with regard for its character features as interdependence of teacher and learner, faculties of learner, optimum of the strategies of knowledge and exercises dosage by teacher, memorizing rate and forgetting rate of learner, duration and stability of his active state, etc.

We interpret teacher and learner as adaptive automata and the training process will consist of their iterative interaction. From automaton Ц Teacher's point of view it will be selected the optimum feeding of educational information to automaton- Learner on the base of how learner mastered such kind of information on the previous educational steps.

The educational system should be many-sided (if it is possible) enough for a given subject spheres and it should be opened and can be easily enlarged in all its main parts.

As a model class of such subject spheres we choose real languages. Thus, these educational systems after filling them with concrete content become computer educational systems for learning languages. According to all mentioned above, the problem of adaptive computer teacher synthesis includes the solution of the following main tasks:

  1. Development of automaton Ц Teacher.
  2. Development of automaton Ц Learner.
  3. Development of information system analogous to the text Ц book with exercises.
  4. Development of interface with wide services for user.

The solution of these problems is accompanied by consideration of complicated technical questions. Following problems can be related to them:

a) Development of dynamic data-bases and knowledge- bases contained large masses of syntactical-information with complex semantics and fuzzy logical connections. These bases should be compact and at the same time allow to obtain rapidly enough the necessary information.

b) Development of indication space of automaton Ц Teacher and Ц Learner states description with pointing out functional metrical relations between them that allow to set functions of these automata.

c) Development of optimum strategies of interaction automaton- Teacher with automaton- Learner both with the help of automata theory and fuzzy logic and with the help of such procedures as pattern recognition, etc.

Authors mentioned in the title of article involved have been developing these problems. Results of the work are shown in a number of articles (see, for example [1, 2]). A theoretic basis is automata model [3] of hybrid kind and also some models and methods developed in the limits of intellectual system theory. Investigations resulted in development of successfully working computers model. The structure and principles of the main units acting of computer educational system "IDEA for Windows" using expert system as "computer teacher" is described below.


2. General structure of system "IDEA for WINDOWS"


System "IDEA for Windows" (or "IDEA") consists of the following main modules:


Academic course

Academic course is a data set, necessary for training, It contains:

  1. Academic material Ц the description of a subject sphere in the form of texts, images, diagrams, schemes, video, sound recordings etc.
  2. Training exercises for confirming knowledge and skills obtained. Receiving of a new information can be organized of course as exercises. There isn't precise definition between training and text exercises because the expert system follows the educational process and therefore for making decisions the system can use data, obtained during execution of training exercises. "IDEA" has a permanently enlarged set of exercise types (at present Ц 20), among them is, for example, pronunciation training, that can be used while learning foreign languages.
  3. Description of the academic course structure as a tree (or a number of trees) of learning goals. The tree contains a global learning goal as a root , for example, learning academic course adoption. The knots of following level are subgoals of this goal and in the same way up to the knots of the lowest level Ц elementary learning goals. The example of elementary learning goal can be a knowledge of the forms of one irregular verb. As a rule there are a few global learning goals and, therefore a few trees corresponding to them. In the process of languages study these goals are grammar, communication skills, vocabulary, knowledge of standard language constructions etc. For every learning goal it can (and for goals of the lowest level Ц must be) indicated a collection of training and text exercises used by system for conforming or checking up knowledge, necessary for reaching the goal.
  4. Guide information Ц tables, dictionary, etc.
  5. Collection of learning strategies. An learning strategy is a planed sequence of exercises or learning goals.

In the same course one or another learning strategies can be selected depending on its tasks (it can be not obliged to study all learning material, but only some part of it). Selection of strategy is influenced on educational conditions (frequency and duration of lessons), start knowledge of learner, a type of learner, determined on analysis of educational process, specific events during educational process (for example, unexpected difficulties when learning certain material) and, perhaps, the age of learner. Thus, academic course, containing enough amount of strategies, allow expert system to use individual approach to learner. This is the main advantage of "IDEA" system over other computer educational systems. Of course, during the educational process some local deviations from strategy can occur.That connected with decisions of expert system or learner himself , but the global task of expert system is to guide learner along selected strategy and therefore to solve an educa! tional problem according to this strategy.


Interface of learner

This module solves following tasks:

  1. Display for learner the content of academic course, all its parts mentioned above.
  2. Allow to execute exercises and during execution give an access (allowed by course author) to reference information. For this access learner's interface uses its navigational means or the system of hypertext references.
  3. Display results of exercises and recommendations of expert system.
  4. Give learner (or teacher examining the education) an opportunity to interfere in the educational process, local deviated from the sequence of exercises planed by system. Thus deviations can be:
    • Temporary alternation of educational conditions. There are three conditions depending on how the next exercise is selected: by expert system, by learner himself after overlooking the course content or independently of previous results next exercise is selected from current strategy.
    • Alteration of current strategy.
    • Pause Ц both between exercises and during its execution.
  5. Display educational statistics Ц executed exercises, their results, other events, occurred during execution (excess of planed time, use of reference information, etc.), recommendations of expert system, deviations from learning strategy made on learner' s initiative (see "Data base of educational history).
  6. Give learner an idea about current state of his knowledge Ц the level of reaching all learning goals described in the goals tree and also goals corresponding to learning strategies.


Author system

Author system is an instrument for creation and editing of all academic course parts.


Module of source editing and its access in learner's condition

This module provide the access of academic course data to module of learner's interface and to author system. The access level ( possibility of overlooking and selecting data) is determined by interface modules. Particularly, not only learner but also author (depending on his qualification) can neither see nor have an opportunity to edit some course parts.


Data base of educational history

This data base keeps and renews information about all events, occurred during the educational process. This information is a basis for making decisions by expert system and also is accessible in part to learner through his interface.

Next events are fixed:

  1. All executed or started exercises and interrupted exercises are fixed separately.
  2. Exercises executed successfully and separate parts of exercises.
  3. Errors made during execution of exercises with qualification of error class. "IDEA" uses 4 classes of errors:
    • casual errors (misprints, casual pressure of "mouse" button not in necessary place, etc.)
    • errors connected with knowledge (skills) lack of the subject learned at present
    • errors connected with knowledge (skills) lack of other subjects;
    • unrecognized errors (for example, meaningless sequences of letters in the place where it was necessary to write a word)
  4. The use of guide or another additional information.
  5. Considerable excess of time planed by author or, on the contrary, execution of exercise earlier than planed before.
  6. Conducted on learner's initiative deviations from learning strategy and other interferences in training process.

All mentioned above information is accessible both for separate exercises, elementary learning goals and for upper levels of learning goals trees as an accumulated information about subgoals.


Expert system

Expert system consists of core and knowledge base of production type. The task of expert system is to determine the tactics of conducting learner along learning strategy and therefore to solve an educational problem according to this strategy. Besides this, expert system corrects the behavior of learner during educational process, giving him advices. The basis for making decisions is the following information:

  1. Current information (the educational task).
  2. Data -base of educational history.
  3. Hypothesis about learner's type that is constantly corrected during training.
  4. Local events occurred in the last exercise, in current learning goal, occurred during the last lesson or for the last hour, etc.
  5. Learning goals tree Ц is used if needed additional training, return to not enough learned subject and testing knowledge after its repetition.


3. Training using expert system


Beginning the work with "IDEA" for the first time, learner must answer a number of questions planed by the author of academic course. As a rule, these are questions about name, age, start knowledge level of learned subject, about knowledge in adjacent fields (for example, when learning foreign language it is the knowledge of another one.)

It can be possible questions about goals of learner while learning the course concerned, about intended frequency and duration of lessons etc. On the basis of start inquest expert system recommends learner one of the learning strategies. Each strategy contains summary that allows learner to decide himself either follow expert system's recommendation or try to begin his training according to another strategy. The second task of start inquest is to give expert system information for development of preliminary hypothesis about learner's type. During educational process this hypothesis will be corrected. According to current hypothesis expert system will apply one or another tactics of training. There is also another significant aspect that allows to individualize the educational system, that is "IDEA" calls learner by name and addresses him "you" and etc.

Course author has a possibility of using the value of that variables in exercises' texts.

Then expert system selects the first exercise. All events occurred during its execution are fixed in data- base of learner results. After execution of the exercise expert system gives learner advices, explains current training situation and proposes next exercise. In any moment learner can deviate from proposed exercise sequence. This fact and all events occurred during such deviations are fixed in data- base of learner results and will be used by expert system later on when making decisions.


4. Optimization of time of making decisions by expert system


The task of optimization of time of making decisions by educational expert system attains special importance in view of considerable volume of data and knowledge bases and of necessity of calling expert system very frequently and for immediately reaction on occurring events. While solving this problem the following feature of expert system "IDEA" was used. Knowledge base uses in the left part of productions almost all information kept in the data- base of educational history and therefore when calculating the truth of conditions we have to call data base every time. But the content of data -base is changing very slowly because its main part is accumulated information about learning goals. That's why data- base compiler creates special index structure setting up relationship between data -base records and knowledge- base productions. When renewing information in data Ц base of educational history, the list of knowledge- base productions and conditions necessary to calculate is selected immediately and this list reduced essentially the time of making decisions.


5. The language of educational history description


For expert system a special language was developed. It allows to use in knowledge- base productions accumulated information about events kept in data base of educational history.

The language is based on the following ideas:

  1. Learner type. "IDEA" doesn't use predetermined learner type and give to the knowledge- base author (specialist) a free hand in determination of his own types according to his methods.
  2. Event. Examples of events are mentioned above. The author can determine new events calculated on the base of previously determined events.
  3. Interval. This is either learning goal or learning strategy, exercise, or one of the temporal intervals, that is, the last lesson, last hour, etc.
  4. The type of curve. This is a description of event frequency dynamics on interval in one or two words: frequently; seldom; first seldom then more frequently, etc. "IDEA" uses a finite of curve types determined by author (in the working knowledge- base there are 14 types).
  5. The curve weight on the interval. This is the ratio of the number of occurred events to the number of possible events or, as in case of using guide information to the number of expected events (in that case it can be more than 1).

The experience of knowledge- base creations for system "IDEA" has shown that this language reflected well the structure of specialists' knowledge on training methods. They being acquainted with neither expert system theory nor programming languages are able to create a working knowledge -base. The present work is fulfilled in the limits of INTAS project "Educational computer systems in human spheres of knowledge", N 94-135.




[1]. Waschik K., Kudrjavcev V. B., Strogalov A. S. "Project "IDEA". Information into the new software generation of ICB type for making a knowledge and skills using expert system". Link & Link software GmbH., Dortmund, Germany, 1995, Ц 40 p.

[2] Kudrjavcev V. S., Waschik K., Strogalov A. S., Aliseytshik P. A., Peretruchin V. V. "Educational computer system of automatic machine type". In the book: Problems of theoretic cybernetics, M.: Publish. center RSHU, 1996, p. 111.

[3] Kudrjavcev V. B., Aleshin S. V., Podkolsin A. S. "Introduction into the automatic-machine theory". M.: Science, 1985. Ц 319 p.




Dr. Sci. Prof. Valery Borisovitsh Kudrjavcev
Head of chair MaTIS Dept. of math. and mech.,Moscow State University
Tel.: 007-095-4347014, Fax.: 007-095-4347014,
E-mail: kudryavtsev@intsys.msu.su


Dr. Waschik Klaus
Vice-director Institute for russian and sovijet culture Ruhr-Uni., Bochum
Universitatstr. 150
44801 Bochum Germany
Tel.: +49 234-3223368 , Fax.: 0234-7094243


Dr.Sci. Associate Prof.Alexandr Sergeevitsh Strogalov
Director Moscow Scientific Center for culture and information technology
Tel.: 007-095-973-4707, Fax.: 007-095-4708888,
E-mail: strog@rsuh.ru


Dr. Pavel Alexandrovitsh Alisejtshik
Senior.Researcher , chair MaTIS Dept. of math. and mech.,
Moscow State University
Tel.: 007-095-3386073, Fax.: 007-095-9734707


Dr. Vadim Valentinovitsh Peretruchin

Leader programmer ,chair MaTIS Dept. of math. and mech.,
Moscow State University
Tel.: 007-095-3430096, Fax.: 007-095-9734707

Proc. of 2nd Russian-German Symposium New Media for Educational and Training in Computer Science, Moscow, 1996

To the top

   © 2001-2013. The Chair of Mathematical Theory of Intelligent Systems, Laboratory of Problems of Theoretical Cybernetics
 Full-text search
Only exact word forms      Give up to search results per page
Rambler's Top100 –ейтинг@Mail.ru