Abstract
Content
Введение
The educational process in a technical University is provided by its divisions with a rigid hierarchical structure (rectorate, faculties, departments). On the lower Cathedral level training to professionally focused, distributed in space, time-based, dynamic content and therefore, as a process of training, is difficult to create a modern computer-assisted distributed learning.
Established by the present rules of education do not allow to react to changing market requirements, to fully take into account the individual capabilities and desires of students to learn discipline, accelerated time (externally) provide mandatory personal presence of the teacher at all stages of transmission and control learning, hard binding students to schedule classes. The student always has the opportunity to learn individually, as without communication with a teacher, despite a full training and methodological support, it is difficult to obtain a good level of knowledge. The modern trend in engineering education is characterized by individualized educational schemes, fully meet the rapid changes in market conditions. Therefore, the classical scheme of centralized management of education from the rigid structure should allow it to be transformed into more flexible schemes [1-4].
1. Actuality of theme
Master's work is devoted to the actual task of developing an application that will allow to respond to changing market demands, take into account the individual capabilities of students and adapt to students, but the most important avtomatizacija of the learning process that allows knowledge to be acquired most efficiently and with minimal downtime.
2. The purpose and objectives of the study, expected results
The aim of the study is the design and development of the training system, semiadalia students with faculty and was assigned the following tasks:
- 1. Perform analysis of the Department and of the interaction of the student with teachers
- 2. To design agent-oriented model of the Department
- 3. To implement the agent-orientirovannaya system using JADE(Java Agent Development Framework)
Scientific novelty - the resulting system will minimally depend on the user, each user will have his own agent, who will analyze and make decisions(selection of lectures, check students ' works...).
3. Overview of research and development
It should be noted that there are different tools for distance learning, but they are not absolutely automatic. Often these tools involve regular communication between student and teacher, which in turn actually reduces the load on users (particularly teachers) since the test results of the student and individual approach to the student still requires user intervention(teacher) in the learning process.
However, this system will control the process automatically, where instructors will retrieve a set of lectures and tests, and the student will be sufficient to choose the course.
3.1 Review of international sources
University without walls - correspondence courses from the Moscow state University named after M. V. Lomonosov. These courses are only supplementary to the education that one gets at University or school...the User can choose the learning direction, and can select only separate independently from other courses. This system is good for users that were(are) trained in the direction to which belongs the course. However, if the user has just started training in this direction, to Orient the user is difficult because of dependencies with other courses in the area.
3.2 Review of national sources
The centre for distance education DonNTU this system in fact is not an automated training system. When logging in, the user is shown a list of courses related to their course of study and direction of education, where can read received lectures from the teachers, and then at the appointed time to pass the control test. However, for validation testing sometimes requires the intervention of the teacher, because the verification system does not univesally. Also, the system has absolutely no updates at the end of the course, polowater not goes not next level course by passing the previous(requires a request from a student or invitation from instructor).
Organization of learning process based on the agent-oriented approach
In the educational process of the Department involves the following entities: the faculty of the Department (head of Department M, the lecturers L1, L2,..., Ln, assistants, P1, P2,..., RK), technicians (Y1, Y2,...Ym), students (X1, x2,...,Xn). At each point in time (semester) students learn the curriculum of the specialty disciplines (D1, D2,... Dm). The study of each discipline includes listening to the lectures (Lect1, Lect2, ...Lectk), laboratory works, it is possible to perform the coursework and the exam (test). To be successful, students need to attend lectures and communicate with teachers on practical and laboratory classes, i.e. there are established interactions and relationships between subjects of educational process. The learning process can be described by the following components:
УП = (S, K, R, Аct, I, T, U)
,where S ={М, (Х1, Х2,…,Хn), ((L1,L2,…Lm),(Р1, Р2,…, Рk)), (Y1, Y2,…Yl)} – many of the subjects of educational process; K – environment (Department), which operates the process; R – relationships established for the subjects of the educational process (horizontal: lecturer-assistant lecturer-Manager; vertical: student-lecturer, student-assistant student-Manager); Act is the set of actions performed by actors; I – the set of installed communications and interactions between entities; T is the schedule of training sessions; U is the fulfillment of a student educational plan (magazines, academic performance, academic records, teaching cards disciplines). The educational process at the Department (K) is performed through actions (Act) and interaction (I) between specific subjects (Xi, Li, Pi, Yi) determined by the Department with respect to R (for example, lecturer-discipline-student, lecturer-assistant Professor) taking place on a strict schedule (T). The Department, as an environment for carrying out educational process, may impose various restrictions, for example, in the form of shortage of classrooms. Thus, the educational process as an object of modelling is distributed and dynamic. The aim is to create a computer learning environment (a virtual It Department?), which stores all the essential to study the relationship (R) and eliminated the rigid space-time limitations in the schedule of classes (T) (Img.1):
УП = {S, K?, R, Act, I, U}.
This environment can be successfully constructed on the basis of the agent-oriented approach, using the principles of distributed artificial intelligence[5].
Image1. The scheme of organization of individual student learning based on agent-oriented approach
Multi-agent system (MAS) is constructed as the Union of the individual subsystems (agents), knowledge-based, and is formally defined as follows[1]:
МАC = (А, Е, R, ORG, ACT, COM, EV),
where a is the set of agents able to function in some environments E that are in certain relations R and interacting with each other, forming some organization ORG, having a set of individual and joint actions ACT (strategies and actions), including possible communicative action SOM, and is characterized by the features of the evolution of the EV. The main property of agents is connected with its autonomy, i.e. the ability to function independently. Additionally, agent-oriented decomposition allows to reduce the complexity of creating software systems to ensure their reliability and simplify their maintenance.
Designing agent-based models of the Department
To develop multi-agent applications that automate the learning process at the Metropolitan level, was conducted by agent-oriented (AO) domain analysis of the educational process on the methodology Gaia [6]. Using this methodology the developed model, necessary to describe the virtual Department and subsequent software implementation. The learning process, in accordance with the methodology Gaia, described by the following models: model roles model interactions model agent model services model ties. Image2 shows the relationship and the content of the models in agent-based design.
Playback functions of the University Department is a complex task, which, in accordance with the Gaia methodology, it is natural to consider how the organization of many really existing and interacting roles: lecturer, assistant, technician, student[7].
Developed agent-oriented models has allowed the system to move from stage to stage of a software implementation of a computer environment with elements of quasi-real communication between subjects of educational process of studying of disciplines of the Department.
Since software agents delegated the execution of office subjects of the educational process, they have to simulate interaction, which to some extent correspond to their professional activities. To simulate professional activities of each software agent should have knowledge about assigned job duties, knowledge about the agents with which communication is possible, and also the rules that govern its behavior in terms of the performance of their duties.
Image2 – Relationship models in agent-oriented design of the learning process
The nature of the educational process, it follows that except for the reactivity, autonomy, activity and sociability, the architecture of a software agent must have internal mechanisms, which are mental properties such as beliefs, commitments, capabilities and rules of conduct. From an existing classification to create a more suitable system architecture based on classical principles of artificial intelligence, i.e. intelligent agent architecture based on production rules.
A Jade environment for developing agent-orientirovannoi system
Agent platforms provide interim Executive level between software agents and the operating system. A multi-agent system runs "on top" of the agent platform and uses its services.
To develop the agent-based model Department of the University was chosen as the platform of Jade (Java Agent Development Framework). Is a software development environment multi-agent systems and applications that support FIPA (Foundation for Intelligent Physical Agents) standards for intelligent agents, which includes:
1) the runtime agents the agents are registered and work under the control of the environment;
2) a library of classes that are used to develop agent-based systems;
3) a set of graphical tools for administration and monitoring vital functions active agents.
In addition to the agent abstraction, Jade provides a simple but powerful model execution and structuring of tasks, as well as a disposable agent communication based on the paradigm of asynchronous message passing[9].
The basis for the functioning of the agents is a mechanism of semantic interpretation (Interpreter Behaviour), which is based on two key concepts – the semantic reprezentatsii (SR – Semantic Representation) and the rules of semantic interpretation (SIP Semantic Interpretation Principle).
The main activity of the agent consists in the deduction of perceived events and of modifications to existing beliefs and patterns of behavior of the agent based on the processing of semantic representation by applying the rules of semantic interpretation.
The main components of the framework (Img. 3) are:
1) base beliefs – beliefs the agent;
2) table of action descriptions (pre - and Postconditions) and the IDs of all actions that may be performed by the agent;
3) model add – in add-in tools;
4) table semantic rules of interpretation – the basic rules of semantic interpretation.
Image3 – Scheme of message processing
The framework takes on the function of semantic analysis of a message, removing it from the functionality of agents and routes messages between agents after the beginning of their interaction, which limits the activity of the agent. Thus, the ontology agent can cover only a narrow area of expertise, having no idea about the whole system[8].
Insights
Implemented agent-oriented analysis of the learning process of students at Cathedral level on the basis of which received the new model Department of the University. Department of the University presented in the form of a distributed system whose elements are the teachers and the students. Their behavior and the authority delegated to artificial software agent using the model roles. Communication between agents is described by models of communication and interactions.
As a physical platform on which are projected the resulting model agent, was selected instrumental environment is Jade, which is a simple but powerful tool for software implementation of multi-agent systems. In the future we plan programming virtual agents of the Department with the workbench Jade.
References
1. Тарасов В.Б. От многоагентных систем к интеллектуальным организациям: философия, психология, информатика.- М.: “Едиториал УРСС”, 2002. -352 с.
2. Гаврилова Т. А., Яшин А. М., Фертман В. П. Взаимодействие интеллектуальных агентов для поддержки сервера дистанционного обучения// Материалы междунар. конф. ”Интеллектуальные системы и информационные технологии в управлении IS&ITC”, Псков, 2000, с.224-227.
3. Курейчик В.М. Эволюционная адаптация интерактивных средств открытого образования /В.М. Курейчик, Л.А. Зинченко //Открытое образование. - 2001. - N1. - С.43-50.
4. Глибовец Н.Н. Использование JADE (Java Agent Development Environment) для разработки компьютерных систем поддержки дистанционного обучения агентного типа// Educational Technology&Society 8(3) 2005 ISSN 1436-4522 pp. 325-345.
5. Федяев О.И., Жабская Т.Е., Лямин Р.В. Система индивидуального обучения студентов агентно-ориентированного типа // Сб. тр. VIII международной конференции «Интеллектуальный анализ информации, ИАИ-2008», Киев, Просвiта, 2008. – С. 502–511.
6. F. Zambonelli, N. R. Jennings, and M. Wooldridge. Developing Multiagent Systems: The Gaia Methodology. In ACM Transactions on Software Engineering Methodology, 12(3), 2003.
7. Zhabska Tetiana , Fedyaev Oleg . The development of agent-based intellectual e-learning environment // Proceedings of the IADIS International conference Intelligent systems and agents, 2011, Rome, Italy, July 24 - 26, 2011. - Pages 143-147.
8. Филимонов А.Б., Кромин О.А. Особенности разработки .мультиагентных систем на платформе Jade semantic agent framework // Сборник трудов Московского государственнего университета приборостроения и информатики, 2008. – С. 397-403.
9. http://jade.tilab.com/documentation/tutorials-guides/ - Access mode: http://jade.tilab.com
RU
UA
ENG