An Intelligent Tutoring System for Construction and ... - Springer Link

An Intelligent Tutoring System for Construction and Real Estate Management Master Degree Studies Arturas Kaklauskas, Edmundas Zavadskas, and Ruslanas Ditkevicius Vilnius Gediminas Technical University Sauletekio Ave. 11, LT-10223 Vilnius, Lithuania [email protected]

Abstract. Three e-learning Master degree studies were introduced at Vilnius Gediminas Technical University since 1999. In order to increase the efficiency and quality of e-learning studies, an Intelligent Tutoring System for Construction and Real Estate Management Master Degree Studies (ITS-CREM) was developed. ITS-CREM consists of six subsystems: Domain Model, Student Model, Tutor and Testing Model, Database of Computer Learning Systems, Decision Support Subsystem and Graphic Interface. ITS-CREM is briefly analyzed in the paper.

1 Introduction The e-learning Master degree studies "Real Estate Management" was introduced at VGTU in 1999, Master degree studies "Construction Economics" from 2000, and Master degree studies "Internet Technologies and Real Estate Business" from 2003 (see http://odl.vtu.lt/). There are currently 220 master students from all over Lithuania studying in these three e-learning master programs. Different multimedia and communication means are used during these studies, namely: electronic format of textbooks, video and audio, as well as computersoftware, computer learning systems, intelligent testing systems, intelligent tutoring system, computer conferencing, computer networks, a discussion forum and ‘face-toface’ contact. There are currently 220 master students from all over Lithuania studying in the above e-learning master programs. In order to increase the efficiency and quality of e-learning studies, an Intelligent Tutoring System for Construction and Real Estate Management Master Degree Studies (ITS-CREM) was developed. ITS-CREM is used for the whole master degree study programme, while traditional ITSs are used only for a single subject or module. In this specific case, ITS-CREM was adjusted to two master degree study programmes: Real Estate Management and Construction Economics. This paper is structured as follows. Following this introduction, Section 2 describes an Intelligent Tutoring System for Construction and Real Estate Management master degree studies. In Section 3 we have provided a description of the System Architecture. Finally, some concluding remarks are provided in Section 4. Y. Luo (Ed.): CDVE 2006, LNCS 4101, pp. 174 – 181, 2006. © Springer-Verlag Berlin Heidelberg 2006

An ITS-CREM Master Degree Studies

175

2 An Intelligent Tutoring System for Construction and Real Estate Management Master Degree Studies The Intelligent Tutoring System for Construction and Real Estate Management Master Degree Studies (ITS-CREM) consists of six subsystems: Domain Model, Student Model, Tutor and Testing Model, Database of Computer Learning Systems, Decision Support Subsystem and Graphic Interface. The subsystems are briefly analysed below. Domain Model contains information and knowledge that the tutor is teaching. During three semesters in the master student’s courses, students complete seven core modules and five optional modules. An elective choice is made from 21 modules within the “Real Estate Management” Master’s degree Program and 17 modules within the “Construction Economics” Master’s degree Program and students should optionally pass 5 examinations. During the fourth semester master students write a final thesis. After registration, students mark sections of optional modules they want to study in the electronic questionnaires. The system also offers study materials to students according to the repetitive key words in different optional modules (see Fig. 1). A mixed approach is also possible and available. The received information is used for action plans, i.e. "mini curricula" that are used to lead the learner/student.

Fig. 1. Search of information for optional module according to repetitive key words (left) and selected material for optional module with defined schedule (right)

Much data had to be processed and evaluated in carrying out the multivariant development and multiple criteria analysis of an optional module. Numbers of feasible alternatives can be as large as 100,000. Each of the alternatives may be described from various statistical (the repetitive key words) and qualitative perspectives. The problem arises, how to perform a computer-aided development of the alternative optional module variants, based on this enormous amount of information. To solve this problem, new methods of multiple criteria multivariant development and multiple criteria analysis of an optional module was developed by authors (see http://odl.vtu.lt/proj1/md_of_ opt.htm). The development and multiple criteria analysis of an optional module is carried out in 5 stages: forming the table of optional module alternative paragraphs codes; rejection of inefficient versions; computer-aided

176

A. Kaklauskas, E. Zavadskas, and R. Ditkevicius

development of optional module variants based on codes of paragraph alternatives; development of optional module versions; development of a summarised decision making table of all optional module versions obtained; multiple criteria analysis of the developed optional module versions, determination of priority and utility degree of alternatives. The Student Model stores data that is specific to each individual student. The Student Model is used to accumulate information about the education of a student, his/her study needs, training schedule, results of previous tests (if he/she has studied earlier in the above-listed e-learning MSc programmes or qualification improvement courses) and study results. Therefore, the Student Model accumulates information about the whole learning history of a student. The Student Model starts by assessing the student’s knowledge of the subject or what the student already knows. Student Model uses that data to create a representation of the student's knowledge and his/her learning process and represents the student's knowledge in terms of deviations from an expert's knowledge. On the basis of these deviations the system decide what curriculum module, or chapter (subchapter) of a module should be incorporated next, and how it should be presented (text, multimedia, computer learning system, etc.). Since the purpose of the Student Model is to provide data for the Tutor Module, all the gathered data should be able to be used by the Tutor module. Decision Support Sub-system is used in mostly all components of the ITS-CREM (Domain Model, Student Model, Tutor and Testing Model, and Database of Computer Learning Systems) by giving different levels of intelligence for these components. Decision Support Sub-system was developed by applying multiple criteria decision making methods developed by authors [1, 3], i.e. a method of complex determination of the weight of the criteria taking into account their quantitative and qualitative characteristics; a method of multiple criteria complex proportional evaluation of the alternatives; a method of defining the utility and market value of an alternatives; and a method of multiple criteria multivariant design of an alternatives. Decision Support Sub-system aids and strengthens some kinds of decision processes. The Decision Support Sub-system is a computer-based system that brings together information from a variety of sources, assists in the organization and analysis of information and facilitates the evaluation of assumptions underlying the use of specific models. The Decision Support Sub-system comprises of the following four constituent parts: data (database and its management system), models (model base and its management system), user interface and message management system. The Database of Computer Learning Systems enables the use of the following Web- based computer learning systems [1-4]: construction, real estate, facilities management, international trade, ethics, innovation, sustainable development and building refurbishment, etc. The above systems have been developed by using multiple criteria methods [1, 3] as were developed by the authors. The created Database of Computer Learning Systems (CLS) is innovative since it contains a CLS with three integrated subsystems: a Decision Support Subsystem, Knowledge Subsystem and Devices Subsystem. For example, a Real Estate’s Knowledge and Device-based Decision Support System (KDDSS-RE, see http://dss.vtu.lt/ realestate/) consists of a Decision Support Subsystem, Knowledge Subsystem and Devices Subsystem. Major KDDSS-RE functions include creating and maintaining customer’s personalized real estate objectives, preferences, and evaluation criteria;


177

participation of various stakeholders (buyers, sellers, brokers, etc.) in joint determination of criteria (criteria system, values and weights) defining real estate; market signalling, providing device-based data about the indoor microclimate and allergens causing allergy in the building; searching for real estate alternatives, finding alternatives and making an initial negotiation table, completing a multiple criteria analysis of alternatives, making electronic negotiations based on real calculations, determining the most rational real estate purchase variant, and completing an analysis of the loan alternatives offered by certain banks. The use of multiple criteria computer learning systems in solving various problems encountered in the course projects and thesis was also aimed at determining: student’s knowledge that is acquired at the university, student’s general level of education, student’s keenness of mind, student’s ability to quickly and adequately respond to changing situation. The Domain Model presents frames to the student. The Tutor and Testing Model provide a model of the teaching process and supports transition to a new knowledge state. For example, information about when to test, when to present a new topic, and which topic to present is controlled by this model. The Tutor and Testing Model reflect teaching experiences of associate professors or professors. The Student Model is used as input to this component, so the Tutor and Testing Model’s decisions reflect the differing needs of each student in optional modules. The Tutor and Testing Model formulates questions of various difficulties, specifies sources for additional studies and helps to select literature and multimedia for further studies and a computer learning system to be use during studies. A student can select the level of difficulty at which the teaching takes place. For example, the chapters of modules with mathematical orientation (i.e. mathematical methods used for the estimation of market or investment values) are quite difficult for some students. Traditional testing systems evaluate a learner's state by giving them a mark and do not provide a possibility to learn about one’s own knowledge gaps or to improve knowledge in any other way. The Tutor and Testing Model compares the knowledge possessed by a student (test before studies) and knowledge obtained by a student during studies (test after studies) and then it performs a diagnosis based on the differences. By collecting information on a history of a student's responses, the Tutor and Testing Model provides feedback and helps to determine strengths and weaknesses of a student’s knowledge, and his/her new knowledge obtained during studies is summarized and then various recommendations and offers are provided. After giving feedback, the system reassesses and updates the student’s skill’s model and the entire cycle is repeated. As the system is assessing what the student knows, it is also considering what the student needs to know and which part of the curriculum is to be taught next. Also, there are options for selection of the following question in a test, which depends on the correctness of answers to the previous questions. Correct answers lead to more difficult tasks, incorrect – to easier ones. The obtained knowledge is the difference between the possessed knowledge (test before studies) and the final knowledge (test after studies). The Tutor and Testing Model also explains why one or another answer is correct/incorrect and offers certain additional literature and multimedia related to the incorrectly answered questions.

178


By applying ITS-CREM, a tutor does not renew tests for every learner. Questions are saved in a question database and hundreds of test alternatives are developed casually. The question base of the Tutor and Testing Model accumulates the following information: questions according to modules, possible answers to a question, evaluation of the correctness of possible answer versions. An incorrect answer is evaluated by zero and a correct is evaluated by one (see Fig. 2 (right)); intermediate answers score from 0 to 1, the difficulty of a question is determined on the basis of the results of previous tests taken by other students (see Fig. 2 (left)), links to the study material that is related to the question and time allocated for testing. The developed ITS-CREM helping us to collect, organize numerical information in the form of tables, graphs, and charts, analyze and interpret of numerical data and make informed decisions. Applying the System is possible: to find interesting patterns that identify the student’s behaviour on the system; to store students interactions and feedback with the ITS, in order to maintain past memory experiences and then derive new paths of teaching.

Fig. 2. Questions sorted according to difficulty (left) and information on the correctness of an answer (right)

The system provides information on testing process in a matrix and graphical form: information on correct and incorrect answer, time distribution to every question, number of times a student has changed an answer to each question of a test, etc. Also the complex parameters are presented, where not only the correctness of the answer is evaluated, but also the time required for student to answer as well as the doubts of selection. Evaluating the answer by a complex parameter, the knowledge assessment may even change. By having such a question data base, it is also possible to create tests in a nonrandom way, but to individualize tests for each student according to the number of questions, their difficulty and proportion of questions from different topics. The received test results are saved in the results data base. By using statistics provided by the Tutor and Testing Model, students can see the question’s difficulty, average the evaluation of the whole group and learn about their position in the group before and after studies. Saving the data on a question’s difficulty provides the opportunity of giving easier questions first of all and later moving


179

on to more complicated ones. Similarly the topics can be selected – from the simpler to the more difficult by repeating the most complicated topics. Therefore, on the basis of the compiled question’s base, questions for tests are formulated not randomly; they are individually adjusted to each student according to the number of questions, their complexity and the proportion of questions from different modules. It is also possible to give easier questions at the beginning and then proceed to more complex questions. Similarly, it is possible to select the taught subject from easier to more complex and to repeat subjects that are not yet mastered. A modern intelligent tutoring system can implement its functions effectively only when users have active dialogue with a computer by using means for dialogue organisation, which is determined by how the information is provided and how the information and commands are interchanged. Therefore, the system-user dialogue is important, as well as the interface (dialogue system) because they help one to have comfortable and effective dialogues. This system has a graphic interface: icons in windows opened in the computer screen show data, models and other objects available in the system. By applying the graphic interface a user can control data, knowledge, models and subsystems and review results on the computer screen or have them printed. In order to clarify a number of issues related to the e-learning Master degree studies process (first and foremost the student motivation, efficiency of advertising, issues related to the quality of study materials, reaction of social environment, etc.) a survey and statistical analysis was conducted [2]. The survey and statistical research analyse in more deatail the composition of the student body by emphasis the problems of the learning process, the social, economic, moral issues related to the labour market integration of trained professionals, etc. An analysis of the ITS-CREM was performed by a designed questionnaire. The letter was attached to the questionnaire was as follows “We would like you to draw on your experience and expertise to help us to test whether the ITS-CREM can also meet your needs as a student. Please read through the following questions circling your response”. A more complete study is underway to study the satisfaction of students. Since it is impossible to cover the body of students from last six years, we selected 125 respondents from those who have already completed their e-learning studies (16 respondents) and those who are still studying (109 respondents). The e-learning Master degree studies are very popular – over the period of 5 years the number of students has risen 6,7 times [2]. Students examination time by using ITS-CREM has decreased 3 times. According to performed a survey and statistical analysis these are the main advantages of ITS-CREM: convenient form of studies, an opportunity to get acquainted with new information technologies, saving of time, increasing quality of studies, fast communication, flexible choice of optional module, good professional training, etc.

3 System Architecture The architecture of ITS-CREM consists of six main components: Domain Model, Student Model, Tutor and Testing Model, Database of Computer Learning Systems,

180


Decision Support Subsystem and Graphic Interface. We have carefully studied the requirements of Lithuanian stakeholders in developing e-learning services. By reflecting on these requirements, our design of the ITS-CREM, aims at a system with the following features: accessibility, reliability, effectiveness, modularity and extendibility, easy to use, and multimedia. These features are described below. The services provided by the system are widely accessible, by using Internet to reach a wide range of life long learning students and are on-line updates of information and maintenance of the services. ITS-CREM can be unreliable either inherently (unreliable primary data) or become unreliable within the system. The level of reliability depends on the reliability of databases and subsystems as well as the ‘quality’ of the students’ learning process. In general, the system is considered reliable if it functions properly during the moments when needed by the student and if (when performing the set functions) it retains the necessary operational values for a required period. Also the system deals with multiple students at different locations. The system is installed in a server, which runs on Windows XP Server and Internet Information Services and is based on the client/server model. HTML, ASP and ActiveX internet technologies are applied in the system. The system provides a student-friendly interface to facilitate easy use of the teaching services. It provides an introduction to the system, a comprehensive windowbased menu of services, and other relevant information about the university and its facilities management services. The information in the database and knowledge base can be represented in different ways; i.e. it can be digital, textual, graphic, audio, video and combined. Since all the above-mentioned data and knowledge presentation forms can represent the same learning material in different ways then the most effective one has to be selected. Together, these features and properties allow an open and flexible framework for providing services that are suited to both life long learning students. To achieve the above goals, the structure of the system must be well designed. For this purpose, we have chosen a client/server model. This model is a concept that is used for describing the interaction between two entities in a computer system, i.e. the student and the server. The main function of the Web Page Interface is to provide the student with an easy way to access the available services. Modules within this component provide functions that implement the services. One of the system’s components provides a channel for the stakeholders to market the latest required information and allows the stakeholders to maintain and update the information on-line. Updating and maintaining data in the database can be a difficult and arduous task. By having easy use of this component, the maintenance cost can be reduced. The Teaching Editor’s Module provides the means for the teachers to edit date, which contains digital, textual, graphic, formula-type, audio, video, and combined data. To handle the complex interactions among students, teachers and the server, a synchronization mechanism are needed which coordinates accesses to the system data so as to maintain consistency. Otherwise, problems may occur because, for example, files can be corrupted, updated information may get lost and appointment books can be out of order.


181

4 Conclusions The traditional Intelligent Tutoring System (ITS) model contains four components: the domain model, the student model, the teaching model, and a learning environment or user interface. Subsystems (Domain Model, Student Model, Tutor and Testing Model, Database of Computer Learning Systems, Decision Support Subsystem and Graphic Interface) of the Intelligent Tutoring System for Construction and Real Estate Management Master Degree Studies (ITS-CREM) created by authors are similar to typical subsystems of Intelligent Tutoring Systems (ITS). We believe that ITS-CREM has several advantages compared to traditional ITS: •

•

•

ITS-CREM is used for the whole master degree study programme, while traditional ITSs are used only for a single subject or module. In this specific case, ITS-CREM was adjusted to two master degree study programmes: Real Estate Management and Construction Economics. The created Database of Computer Learning Systems (CLS) is innovative since it contains a CLS with three integrated subsystems: a Decision Support Subsystem, Knowledge Subsystem and Devices Subsystem (see http://dss.vtu.lt/ realestate/). The created Decision Support Subsystem is innovative since it contains a multivariant development and multiple criteria analysis module, which allows to make multi-variant development and multiple criteria analysis of various alternatives (e.g. of an optional module). This subsystem is briefly described in http://odl.vtu.lt/proj1/md_of_opt.htm.

References 1. Kaklauskas, A., Ditkevičius, R., Gargasaitė, L. Intelligent tutoring system for real estate management // International journal of strategic property management. Vilnius: Technika, 2006, Vol. 10, No. 2, p. 113-130. 2. Rimkuviene, S., Lepkova, N. Analysis of experience and efficiency of distance learning Master's degree programme in construction economics and property management. Journal of Civil Engineering and Management, Vol X, No 1. Vilnius, Technika, 2004, 51-60 p. 3. Zavadskas, E., Kaklauskas, A., Vainiūnas, P., Gikys, M.: Property management in postgraduate Internet studies in Vilnius Gediminas Technical University. 4th Baltic region Seminar on Engineering Education, Lyngby, Copenhagen, Denmark. 1-3 September 2000. – UICEE. Seminar proceedings, edited by Zenon J. Pudlowski and Hans Peter Jensen. pp. 83-86. 4. Zavadskas, E., Kaklauskas, A.: Efficiency increase in research and studies while applying up-to-date information technologies. Journal of civil engineering and management. – 2000, Vol. VI, No. 6, p. 397-414.