Finding Suitable Course Material through a Semantic Search Agent for Learning Management Systems of Distance Education

2011 35th IEEE Annual Computer Software and Applications Conference Workshops Finding Suitable Course Material through a Semantic Search Agent for Learning Management Systems of Distance Education Duygu ÇELİK, Computer Eng., İstanbul Aydin Univ., İstanbul/TURKEY Atilla ELÇİ, Software Engineering Prog., Toros Univ., Mersin/TURKEY 3 Eray ELVERİCİ, Educational Tech., İstanbul Aydin Univ., İstanbul/TURKEY 1 2 1 duygucelik@aydin.edu.tr, 2atilla.elci@gmail.com, 3erayelverici@gmail.com Recently, PowerPoint slides are recorded together with sound and the finished product is converted to a flash file using Adobe Presenter. Thus, asynchronous training media content is prepared and presented. However, the components of such systems can be strengthened in order to infuse more intelligence into the system. For example, the system can detect the levels and critical points of individual students and provide feedback through an intelligent agent. We believe that with such an approach it is possible to develop contributions of the Semantic Web. In this article, a system using Semantic Web technologies is proposed in order to support learning management system, to make such systems more efficient, and to get meaningful results. In order to demonstrate the proposed system work, necessary parts such as course contents, course materials, and students’ profiles in IAUDEC database are gathered to create ontology. For example, Materials, Student Profile, Course and Exam ontology knowledge bases; these ontologies are used in testing the success of the system. Ontologies were created using Protégé 4.1 tool [3]. In addition, the proposed system and its core module, the Semantic-based Course Material Browsing Agent (CMBA), were develop on JAVA platform by using JBuilder 2006 software [4]. Section 2 introduces theoretical backgrounds of this paper. The system architecture, its management module, the created ontology knowledge bases, and inferencing mechanism are given in the third section. The third section also gives example use cases and the results obtained. Section 4 concludes this paper. Abstract— Developing technology and increasing usage of computers around the world make Internet a convenient tool for Distance Education (DE) systems. Learning Management Systems-(LMSs) of DE around the world serve content-based courses to students/users via static pages on the Web. Additionally, querying for specific course content of an entire course in LMS is only possible by syntactic-based search agents which limit the search. Therefore, it is thought that querying of contents through semantic-based approaches would make LMSs more helpful for a student/user through the Semantic Web (SW) approach. Queries can be meaning based and this can enable the system to make meaningful inference in human-computer interaction. In this article, we propose a system architecture that performs semantic–based searches through an agent in order to plan student-based schedule of course documents automatically according to student’s performance in the course of a semester. OWL language of the SW technology is used during the development of the ontology of the LMS system that is used as a knowledgebase and queried to obtain required data. In this article, we concentrate on developing the system as a module for a LMS currently in use. Keywords; E-learning, Ontology, Semantic Web, Semantic Resource Browsing Agent. I. INTRODUCTION Distance education in the 1900s contributed to the evolution of audio-visual devices. Today, distance education practices have contributed to the development of media tools [1]. The success of distance education depends on a multitude of factors, a few of which are as follows: (1) technology used, (2) quality components, (3) suitability of LMS, (4) quality of multi-media based course content, (5) corporate/institutional culture, and, (6) an effective feedback mechanism. Distance education has long been used in Turkey; one example of this is the practice of the Distance Education Institute, Istanbul Aydin University (IAUDEC). According to the research on DE systems, in order to improve effectiveness of education, there should be more than one DE component and components must support one another [2]. These components may be synchronous training, asynchronous training, exams, LMS-web pages, and distance learning live support desk. High-tech education materials, especially Adobe products are used for synchronous and asynchronous training. Today, course contents of distance education which are presented as animations, graphics, videos and other multimedia materials are over PowerPoint presentations. 978-0-7695-4459-5/11 $26.00 © 2011 IEEE DOI 10.1109/COMPSACW.2011.71 II. SEMANTIC WEB Semantic Web, which was developed as an extension of today’s ordinary Web, was introduced by Tim Berners-Lee and colleagues. Semantic Web can allow computers and people work together by using well-defined meaning of information [5]. For healthy functioning of Semantic Web, computers should access to structurally organized data and sets of inference rules which are used for automatic reasoning. These sets are created as ontologies. Ontology refers to the common meaning of knowledge of the basic communication between people or systems which is recognized by the relevant systems [6] detailing the concepts in the field of information systems such as elearning [7 and 8]. In other words, ontology is the definition of concepts and relations that an agent or a group of agents may access [9]. Semantic Web requires a standard Web ontology language to interpret the data meaningfully on the 386 Web environment. To meet this need, W3C had proposed Web Ontology Language (OWL) in 2002, and, this language became one of the W3C recommendations (i.e., standards) in 2004 [10]. OWL 2 was evolved recently [11]. The development of ontology knowledgebase containing educational materials such as exams, slides, animations of DE applications, and processing of such ontology by the semantic-based Course Material Browsing Agent (CMBA) are considered next. The CMBA, which is at the centre of the system, is being introduced in detail in the next section. III. to interfere with the structure of the existing LMS system. The benefit of the module is that it is designed as an internal system which can be added in without breaking the interim order and the mechanism of used LMS systems. The CMBA System consists of the following components: User, Distance Education Database (DE DB), Management Module of Distance Education Database (MMDE-DB), Ontology Knowledge bases (OKB), Inference Engine (IE), and CMBA, the agent. The architectural structure of the CMBA System is displayed in Figure 1; parts, objectives, and working mechanism of the system will be discussed in detail in this section. ARCHITECTURAL STRUCTURE OF SEMANTIC-BASED COURSE MATERIAL BROWSING AGENT (CMBA) The architectural layout of the proposed system was considered to be developed as an add-on module in order not DE DB 5 4 DE DB: Distance Education Database MMDE-DB: The Management Module of Distance Education Database SPT: Student Performance Table CMBA: Semantic-based Course Material Browsing Agent IE: Infrence Engine OKB:Ontology Knowledgebase Student ID 2 Student/ Directors Section ORC Performance ... MMDE DB ORC: Other Relational Courses 4 5 SPT 1 6 Taken Courses Selected Course ID Student Course ID 6 3 CMBA 7 IE 7 OKB Figure 1. Architectural structure of Semantic Source Browsing Agent intended for students will not be mentioned due to the fact that it is out of topic.) The aim here is to provide the system with the ability of determining the standing of students in terms of three levels as good/average/poor. The information gathered will be shown as a table in the Management Module of Distance Education Database (MMDE-DB) and this Student Performance Table (SPT) will include the student ID number, the selected course, units of the course, achievement or performance score for the course, and the other related courses, etc. (see step 4-5, Figure 1). A. The Management Module of DE DB CMBA System is thought to be used as an interface by students and system directors alike. Namely, the CMBA interacts with LMS through a user interface. Logging in with a student ID card to the interface of the CMBA system, all of the courses (in the LMS) that the student has been registered for are seen. In the first interface of the system, units of the student’s lessons, exams and their results, the information about the student’s term performance can be reached via the online education database (see steps 1-3 above in Figure 1). The information about the student and the results of level evaluation are compared against a threshold value. The performance score showing the student’s semester-based performance is comprised of many parameters such as the grades of his/her exams, the times of registration to a course, attendance-absenteeism, the presented projects, assignments and their results, the instructor’s marks, etc. At that point, the system will take the necessary parameters required for statistical calculation of an average achievement score. (In this paper, statistical calculation of the achievement score B. Ontology Knowledgebase (OKB) Protégé 4.1 software was used to create the ontology knowledgebase of the system [3]. The most focused ontology is the Material.owl. Material ontology starts with 'owl: Thing' class. Under this, the ‘Material’ class is defined as a subclass. Four different subclasses are further defined under the Material class. These are; book, presentation, test and video (Table-1). As most of today's distance education course materials are MS PowerPoint, Adobe PDF or MS Word documents, and generally they are presentations, we will deal with presentation section. However, in the created 387 Material ontology we can define all types of lesson materials semantically. and ObjectPropertyAssertion for some defined individuals is given in Table 2. TABLE I. LABELS OF CLASS, DATATYPE, AND OBJECTTYPE PROPERTY IN MATERIAL ONTOLOGY TABLE II. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 46 THE CREATION AND DEFINITIONS OF THE MATERIAL OF COURSE AND EXISTING UNITS <?xml version="1.0"?> Ontology xmlns=http://www.w3.org/2002/07/owl# xml:base="http://aydin.edu.tr/Ontology/2011/Materİal.owl" <Declaration> <Class IRI="#Book"/> </Declaration> <Declaration> <Class IRI="#Material"/> </Declaration> <Declaration> <Class IRI="#Presentation"/> </Declaration> <Declaration> <Class IRI="#Test"/> </Declaration> <Declaration> <Class IRI="#Video"/> </Declaration> <Declaration> <ObjectProperty IRI="#Covered_by"/> </Declaration> <Declaration> <ObjectProperty IRI="#Covers"/> </Declaration> <Declaration> <ObjectProperty IRI="#PreviousMaterial"/> </Declaration> <Declaration> <ObjectProperty IRI="#NextMaterial"/> </Declaration> <Declaration> <DataProperty IRI="#Interactive_Content_Property"/> </Declaration> <Declaration> <DataProperty IRI="#Visual_Content_Property"/> </Declaration> <Declaration> <DataProperty IRI="#Material_Address"/> </Declaration> <Declaration> <DataProperty IRI="#Material_Language"/> </Declaration> <Declaration> <DataProperty IRI="#Material_Author"/> </Declaration> 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 <ClassAssertion> <Class IRI="#Presentation"/> <NamedIndividual IRI="#01_Math101_01_005"/> </ClassAssertion> <DataPropertyAssertion> <DataProperty IRI="#Material_Address"/> <NamedIndividual IRI="#01_Math101_01_005"/> <Literal datatypeIRI="&xsd;URI">../lms/courses/math101</Literal> </DataPropertyAssertion> <DataPropertyAssertion> <DataProperty IRI="#Material_Author"/> <NamedIndividual IRI="#01_Math101_01_005"/> <Literal datatypeIRI="&xsd;Name">Prof.Dr. Ali Gunes</Literal> </DataPropertyAssertion> <ObjectPropertyAssertion> <ObjectProperty IRI="#Covered_by"/> <NamedIndividual IRI="#01_Math101_01_005"/> <Namedİndividual IRI=="#01_Cal102_01_001"/> </ObjectPropertyAssertion> <ObjectPropertyAssertion> <ObjectProperty IRI="#NextMaterial"/> <NamedIndividual IRI="#01_Math101_01_005"/> <NamedIndividual IRI="#01_Math101_02_001"/> </ObjectPropertyAssertion> <DataPropertyAssertion> <DataProperty IRI="#Question_Number"/> <NamedIndividual IRI="#01_Math101_01_005"/> <Literal datatypeIRI="&xsd;integer">12</Literal> </DataPropertyAssertion> x The unit of 01_Math101_01_005 is highlighted at line 3 in Table 2. It means the material numbered as 005 of the freshman Mathematic course, Section/Unit 1 (i.e. Grade_CourseName_Unit_MaterialID). The semantic declaration (class assertion) of this material is defined as an owl:individual in this ontology (see Table 2 line 1-4). x We can see the semantic relations between the focused course material 01_Math101_01_005 and the other course materials. For instance, the material 01_Math101_01_005 is Covered_by the material 01_Cal102_01_001 (see Table 2 at lines 15-19). x In DataProperty Assertions part, we can see that the number of the questions of this material is 12 (see Table 2 at lines 25-29). x The author of it is Prof. Dr. Ali Gunes, and the level of it is the 1st year (see Table 2 at lines 10-14). x The address information is also present in order for this material to be searched and found by a smart agent developed semantically. The address of this material includes an URL information which shows that it is in the Math101 file of LMS courses (see Table 2 at lines 5-9). x The relation between the fifth material of the first grade Math 101 course (01_Math101_01_005) and the second unit material of the first grade Math 101 course (01_Math101_02_001) is the next material. Namely, we can define the materials required in sequence by using the “#NextMaterial: ObjectProperty” (depicted in Table 2 at lines 20-24). We define common features of each material in this ontology. We do this by using the feature definitions of “ObjectProperty” and “DataProperty” that are present in OWL language. The definitions of “ObjectProperty” that we use in the ontology of materials are “#Covered_by”, “#Covers”, “#PreviousMaterial and “#NextMaterial” (see Table 1 at lines 20-30). Additionally, definitions of “DataProperty” are: “#Material_Address”, “#Material_Language”, “#Material_Author”, “#Page_Number” and "#Course_Level" of the material, “#Question_Number” in the materials, “#AnsweredQuestions_Number” in the materials, and “#Multimedia_Content” (some of these are depicted in Table 1 at lines 31-46). Furthermore, the multimedia is comprised of five different subclasses: interactive content, text content, visual content, image content and audio content properties. A part of the material ontology that showing examples of DataPropertyAssertion 388 C. Inference Engine The SCA-Semantic Composition Agent-based System we proposed in our earlier studies was about combining the definitions of atomic work processes in order to constitute a new and advanced process compatible with the query of a user. In other words, it was aimed at a new complex work process, which was not present but needed by a user, through combining present process [13]. However, there is need for suitable and correct inference rules in order to define a semantic work flow. The ‘Armstrong’s Axioms’ (AAs) seem to serve the purpose. AAs form a set of axioms (inference rules) [12]. AAs are used in order to infer functional dependencies-FDs that are encountered in a relational database. There are seven different inference rules as seen in Table 3. Those are Reflexivity, Augmentation, Transitivity, Pseudo Transitivity, Additivity, Accumulation, and Projectivity. In the light of such semantic information, a smart search agent can perceive and interpret all features related to Math 101. In that way, it can decide on if a student needs this material in terms of the information about the student’s performance. Furthermore, the system can strengthen the study program with present materials of related courses and units. Thanks to that, the smart search agents developed semantically can detect the subsequent unit materials and have the ability to make propositions when they are needed. Similar to material ontology, course ontology and student ontology have also been prepared in Protégé. However, in this paper we concentrate on material ontology because we focus on the role of a semantic search agent on the search for necessary materials intended for students. TABLE III. ARMSTRONG'S AXIOMS AND REVISED ARMSTRONG’S AXIOMS. Armstrong’s Axioms 1. Reflexivity: A set of attributes X determines a subset Y of itself: XoY if YŽX. 2. Augmentation: It allows enlarging the left-side of a FD or both side conventionally with one or more attributes. Formally, if XoY then X||ZoY||Z for any Z. 3. Transitivity: If we have functions f : XoY and g : YoZ then we have a function g O f : XoZ, where (g O f )(x) = g(f (x)). Formally, If {(X o Y) and (YoZ)},then XoZ. 4. Pseudo transitivity: is a generalization of Transitivity. It is requires the entire right hand side of a FD appears as attribute(s) of the determinant of another FD. Formally, if {(XoY) and (Y||ZoW)} then (X||ZoW). 5. Additivity: If there are two FDs with the same determinant on the left, it is possible to form a new FD that preserves the determinant and has as its right-hand side the union of the right-hand sides of the two FDs. Formally, if {(XoY) and (XoZ)} then (XoY||Z). 6. Accumulation: If there are two FDs with the complementary determinant, it is possible to form a new FD that preserves the determinant X and forms its right-hand side as the union of the right-hand sides of the two FDs except the complementary determinant Z. Formally, if {(XoY||Z) and (ZoC||W)} then (XoY||C||W). 7. Projectivity: If X determines Y and Z, then X determines Y, therefore it is possible to break each functional dependency XoY down to XoAi for i = 1..n where Y = {A1, . .An}. Formally, if (XoY||Z) then XoY and XoZ. In our earlier work [13], we used the AAs to detect the relation between existing work processes in order to get new work process. We proposed to merge the two inference rules Reflexivity and Augmentation, so brought about Pseudo Factorization axiom. Our aim with this single axiom was to prevent derivation of excessive numbers and possibly incompatible processes. In addition to this, we renamed Projectivity axiom as Dissection. Thus obtained The Revised Armstrong’s Axioms (RAAs) by transformations equivalent to AA axioms, and during these transformations the correctness and integrity of AA axioms were not disrupted. In the first column of Table 4, we see AAs, and in the second column we see RAAs (see Table 3). Formal OWL meanings can provide various characteristics of property. Those are transitive, symmetric, functional, inverse of, inverse functional, etc. For instance, if a property is transitive, for any X, Y, and Z materials the property transitions of P(X, Y) and P(Y, Z) bring about the meaning of P (X, Z). Kang et. al [14] proposed to monitor and order the semantically defined work processes, namely, to infer new property definitions among those work Revised Armstrong’s Axioms 1. Pseudo factorization: If {(X||YoZ) ,(T||ZoW)} and if (ZT||Z and T≡X or T≡Y) then X||YoW. 2. Transitivity: If {(X oY) and (Yo Z)} then (XoZ). 3. Pseudo transitivity: If {(XoY) and (Y||ZoW)} then (X||ZoW). 4. Additivity: If {(XoY) and (XoZ)} then (XoY||Z). 5. Accumulation: If {(XoY||Z) and (ZoT||W)} then (XoY||T||W). 6. Dissection: If (XoY||Z) then XoY and XoZ. processes, resorted to the definitions of owl:property present for planning. However, in this study, we propose to use RAAs to infer the property relations compatible among materials, defining them as “owl: ObjectProperty. The transitivity rule definition is shown in Table 4 (see the lines 1-5) and the usage of inference rules below for constitution of relations among materials was planned as we prepared a work plan. Therefore, in this work, RAAs have been applied to the types of owl:ObjectProperty as mentioned before (Covered_by, Covers, Previous and Next Materials) in order to detect the order relations among materials (just as for work processes) and constitute a material-based work plan for student needs. According to a scenario, a user enters into the LMS system with his/her ID number and chooses the code of the lesson for which a work plan is needed. LMS can reach all information related to both the course and the student from DE DB. 389 TABLE IV. TABLE V. OBJECT PROPERTY VARIOUS MATERIAL RESOURCES THE ONTOLOGICAL DEFINITION OF SOME RAAS <!-- Object Properties of Tasks.Owl--> 1 <owl:ObjectProperty rdf:ID="hasTransitivity"> 2 <rdf:type rdf:resource="&owl;TransitiveProperty"/> 3 <rdfs:domain rdf:resource="#Material"/> 4 <rdfs:range rdf:resource="# Material "/> 5 </owl:ObjectProperty> ASSERTIONS ARE DEFINED FOR 1 2 3 4 5 <ObjectPropertyAssertion> <ObjectProperty IRI="#Covered_by"/> <NamedIndividual IRI="#01_Math101_01_005"/> <NamedIndividual IRI="#01_Calc102_01_001"/> </ObjectPropertyAssertion> 6 <owl:ObjectProperty rdf:ID="hasPseudoFactorization"> 7 <rdf:type rdf:resource="&owl;TransitiveProperty"/> 8 <rdfs:domain rdf:resource="#Material"/> 9 <rdfs:range rdf:resource="# Material "/> 10 </owl:ObjectProperty> 6 7 8 9 10 <ObjectPropertyAssertion> <ObjectProperty IRI="#NextMaterial"/> <NamedIndividual IRI="#01_Math101_01_005"/> <NamedIndividual IRI="#01_Math101_02_001"/> </ObjectPropertyAssertion> 11 <owl:ObjectProperty rdf:ID="hasPseudoTransitivity"> 12 <rdf:type rdf:resource="&owl;TransitiveProperty"/> 13 <rdfs:domain rdf:resource="#Material " /> 14 <rdfs:range rdf:resource="#Material" /> 15 </owl:ObjectProperty> 11 12 13 14 15 <ObjectPropertyAssertion> <ObjectProperty IRI="#Covered_by "/> <NamedIndividual IRI="#01_Math101_01_014"/> <NamedIndividual IRI="#02_Math207_01_001"/> </ObjectPropertyAssertion> For instance, performance of the student is lower than the minimum permissible value for Unit 1 of Math 101. LMS searches for materials relevant to units which student has shown inferior performance. Additionally, it will also capture other related materials which are about the same unit. Thus gathered is a repository of relevant materials. In order to carry out inference operation, LMS consults material.owl capturing ontology information on the relevant material. As seen below in Figure 2, the property ‘NextMaterial’ of “owl:ObjectProperty” numbered 005 which belongs to Unit 1 of the first grade Math 101 is connected to the material numbered 001 which belongs to Unit 2 of the first grade Math 101 course. This relation is associated semantically in Table 5, (see lines 6-10). 16 17 18 19 20 <ObjectPropertyAssertion> <ObjectProperty IRI="#NextMaterial"/> <NamedIndividual IRI="#01_Math101_02_001"/> <NamedIndividual IRI="#01_Math207_01_004"/> </ObjectPropertyAssertion> 21 22 23 24 25 <ObjectPropertyAssertion> <ObjectProperty IRI="# Covered_by "/> <NamedIndividual IRI="#01_Math101_02_001"/> <NamedIndividual IRI="#02_Math207_01_001"/> </ObjectPropertyAssertion> The same semantic relation is defined between the material numbered 001 which belongs to the Unit 2 of the first grade Math 101 course and the material numbered 004 which belongs to the Unit 1 of the first grade Math 207 course (see Table 5 at lines 16-20). Therefore, the common units of two courses and the content-based relation between them are defined semantically. Figure 2. ‘Next Material’ of “owl:ObjectProperty” feature defined for two materials (individual) in the Material.owl The semantic material search agent will be able to understand that after the material 01_Math101_01_005, the material 01_Math101_02_001 needs to be processed, and later bring about a work plan assuming that the material 01_Math207_01_004 is necessary after the material 01_Math101_02_001. For this, it will need a transitive inference rule. It will use the transitive axiom: P (X, Y) and P (Y, Z) imply P (X, Z) in order to detect the material order. ‫ܔ܉ܑܚ܍ܜ܉ۻܜܠ܍ۼ‬ ૙૚̴‫ܐܜ܉ۻ‬૚૙૚̴૙૚̴૙૙૞ ሳልልልልልልልሰ ૙૚̴‫ܐܜ܉ۻ‬૚૙૚̴૙૛̴૙૙૚ ‫ܔ܉ܑܚ܍ܜ܉ۻܜܠ܍ۼ‬ ૙૚̴‫ܐܜ܉ۻ‬૚૙૚̴૙૛̴૙૙૚ ሳልልልልልልልሰ ૙૚̴‫ܐܜ܉ۻ‬૛૙ૠ̴૙૚̴૙૙ then, ‫ܔ܉ܑܚ܍ܜ܉ۻܜܠ܍ۼ‬ ૙૚̴‫ܐܜ܉ۻ‬૚૙૚̴૙૚̴૙૙૞ ሳልልልልልልልሰ ૙૚̴‫ܐܜ܉ۻ‬૛૙ૠ̴૙૚̴૙૙૝ 390 Therefore, many axioms will give information about the ordering among materials by being determined in the suitable places in ontology, and make it possible to comment on the work plan actively on the orderings. Jena (JENA RDF API), one of the semantic web technologies developed for semantic information inference from ontology [15], and PELLET OWL DL Reasoner, one of the semantic technologies for the reasoner operation based on facts [16], are use commonly. Those SW technologies are useful to extract data from ontology knowledgebase and infer new facts from existing facts for material order and quality. For instance, a student may require a material that contains at least 20 solved questions. Furthermore, a rank value about the materials is computed based on various information such as number of questions, number of answered example questions, the person who prepared the material, quality of instruction, picture, video, text or audio content by investigating the “owl:DataProperty” properties. The quality of materials searched and found, and the carriage of them to the upper ranks in the work plans will increase the student achievement and satisfaction. Thus, through defining unitbased relations among materials of courses in the ontology files, and by doing necessary ordering and filtration for constituting a qualitative work plan devoted to the needs via inferences from these definitions, it is possible to present the work plan needed by users. The system can prepare and suggest a study plan considering the student’s term performance, and it is possible to comment on the student’s performance and present a plan of materials if necessary. We need to consider students term performance and define a good study plan because DE-based LMS systems have a lot registered students. This is crucial for such systems to guide and assist their students who are normally out of campus. This may be achieved through a smart module or part of such system [13]. open for improvements and extension. In short, retrieving right materials and producing a complex study plan were made possible by using ontologies. REFERENCES [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] IV. CONCLUSION In this article, we proposed a new system called semantic-based Course Material Browsing Agent (CMBA) which produces study plan by finding materials in an LMS system semantically. We took the Online Education Management System of Istanbul Aydın University as base on which CMBA is grafted. Components of the offered CMBA system are: User, Distance Education Database (DE DB), Management Module of Distance Education Database (MMDE-DB), Ontology Knowledgebase (OKB), Inference Engine (IE) and CMBA the agent. CMBA has its own inference rules; Revised Armstrong’s Axioms (RAA) are used in planning and inference functions. In CMBA’s inference module, RAAs are currently applied on only objectType properties allowing the agent to perform subject ordering. Besides, “owl:DataProperty” properties are defined semantically in order to make quality choice among materials. Several use cases are considered in trying the developed system. The biggest advantage of this system and the created ontology knowledgebase is that it is [12] [13] [14] [15] [16] 391 The Institute for Higher Education Policy, Quality On The Line: Benchmarks for Success in Internet-Based Distance Education. Supported by National Education Association (NEA) and Blackboard Inc. April 2000. Hatanaka A, Okada A, Yuriyama M, Tarumi H, Kambayashi Y, “Functions of Distance Learning System: VIEW Classroom,” http://www.ineer.org/Events/ICEE2000/Proceedings/papers/MC21.pdfHatanaka, A., Okada, A., Yuriyama, M., Tarumi, H., Kambayashi, Y., Functions of Distance Learning System: VIEW Classroom. http://www.ineer.org/Events/ICEE2000/Proceedings/papers/MC21.pdf ,2000, last accessed November 2009. Protégé, OWL-S Ontology Editor CS/AI Department, University of Malta, 2004, http://owlseditor.semwebcentral.org/, last accessed November 2010. JBuilder software package, Borland JBuilder 2006, http://www.borland.com/jb2006, last accessed November 2010. T. Berners-Lee, J. Hendler, O. Lassila, The Semantic Web, Scientific American, 284(5), 34-43, 2001. Gruber, T. (n.d.). What is an ontology?. Retrieved June 15, 2007, http://wwwksl.stanford.edu/kst/what-is-an-ontology.html, last accessed November 2010. Ullrich, C., Description of an Instructional Ontology and its Application in Web Services for Education. Proceedings of Workshop on Applications of Semantic Web Technologies for Elearning SW-EL'04, 17-23, 2004. A. Elçi, A Metadata Model For E-Learning Coordination Through Semantic Web Languages, The Turkish Online Journal of Educational Technology – TOJET, ISSN: 1303-6521 volume 4 Issue 3 Article 3, pp: 12-17, July 2005. D. Çelik, A. Elçi, Provision of Semantic Web Services through an Intelligent Semantic Web Service Finder, Multiagent and Grid Systems - An International Journal, IOS Press, ISSN 1574-1702, 4(3) , 315-334, 2008. OWL Web Ontology Language Overview, W3C Recommendation, Online: http://www.w3.org/TR/owl-features/, last accessed November 2010. Grau, B.C. et. al. (2009) OWL 2-W3C Web Ontology Language. W3C Recommendation by Oxford University April 2008-October 2009 http://www.w3.org/TR/owl2-primer/ Accessed: November 2010. Armstrong, W. W., "Dependency Structures of Data Base Relationships", Information Processing, 74, North Holland, 1974. D. Çelik, A. Elçi, “Towards a semantic-based workflow model to composition of OWL-S based atomic processes through process similarity matching and inferencing techniques”, J. of Internet Technology, ISSN: 1607-9264, Published by: Taiwan Academic Network Executive Committee, 2010. Kang, D., Lee, S., Kim, K., & Lee, J.Y., An OWL-based semantic business process monitoring framework, Journal of Expert Systems with Applications, 0957-4174, Vol. 36, Issue 4, 7576-7580, 2009. Jena RDF API, Hewlett-Packard Development Company, 2001, http://www.hpl.hp.com/semweb/jena.htm, last accessed November 2010. E. Sirin, B. Parsia, PELLET - An OWL DL Reasoner, In International Workshop on Description Logics (DL2004), Whistler, Canada, 2004.