國
立
交
通
大
學
資訊工程學系
博
士
論
文
基於
SCORM 標準的智慧型學習內容管理系統之研製
Design and Implementation of an Intelligent Learning
Content Management System based on SCORM Standard
研 究 生：蘇俊銘
指導教授：蔡文能 教授
指導教授：曾憲雄 教授
基於
SCORM 標準的智慧型學習內容管理系統之研製
Design and Implementation of an Intelligent Learning
Content Management System based on SCORM Standard
研 究 生：蘇俊銘 Student：Jun-Ming Su
指導教授：蔡文能 Advisor：Wen-Nung Tsai
指導教授：曾憲雄
Advisor：Shian-Shyong Tseng
國 立 交 通 大 學
資 訊 工程 學 系
博 士 論 文
A DissertationSubmitted to Department of Computer Science College of Computer Science
National Chiao Tung University in partial Fulfillment of the Requirements
for the Degree of Doctor of Philosophy
in
Computer Science July 2006
Hsinchu, Taiwan, Republic of China
中華民國九十五年七月
基於
SCORM 標準的智慧型學習內容管理系統之研製
學生：蘇俊銘 指導教授：蔡文能 教授 指導教授：曾憲雄 教授 國立交通大學資訊學院 資訊工程系摘要
隨著網際網路的快速發展，網路學習(E-Learning)系統已廣為流行。為了解決 教材無法在不同網路學習系統間分享與再利用之問題，國際組織已提出許多的國 際標準格式，包含:ADL的SCORM、IMS的CP與QTI、IEEE LTSC的LOM、AICC 的CMI等等。而SCORM在近幾年已成為最受廣泛使用的標準。SCORM為因應隨時 隨地學習之需求，而提供可發展、包裝與傳遞高品質教育及訓練教材的教材標準。 雖然SCORM具有分享、再利用、及重新組裝之優點，但對於製作、擷取與管理具 個人化學習導引機制的SCORM教材來說，仍是相當困難。此外，如提供所有學習 者，相同的學習課程與策略，則學習成效將無法有效提升。於是近幾年來，可根 據不同學習者的學習能力與評量結果來提供不同學習課程的適性化學習環境便漸 受重視。故對於基於SCORM標準的智慧型網路學習系統而言，如何有效地建立與 管理具客製化學習導引與教學策略的SCORM課程、如何根據個人的學習歷程資 訊、學習能力及教學策略，自動化提供學習者適當的學習活動、與如何評估及分 析學習歷程資料來了解學習者的迷失概念等，便是本論文所關心的研究問題。目前IEEE LTSC組織提出一稱為學習科技系統架構(Learning Technology System
Architecture, LTSA)的參考模型，用來定義學習科技系統中的關鍵互操作性介面。
而 為 了 支 援 分 散 式 網 路 學 習 系 統 的 互 操 作 性(Interoperability) 與 延 伸 性
Framework, ELF)規劃出具分層概念的網路學習系統模型，其每ㄧ層皆根據網路學 習系統不同的需求來定義其不同的功能。
此外，基於知識管理的概念，如何有效管理適性化網路學習系統中的各種資 源與資訊，就如同於有效的管理不同的知識。因此，基於知識管理與具分層概念
的LTSA 架構，在此論文中，提出了智慧型學習內容管理系統(Intelligent Learning
Content Management System, ILCMS)，來智慧地管理大量的學習內容與提供學習
者適性化的學習策略，並藉由有效地學習歷程分析，做進ㄧ步的策略精練。ILCMS
的分層架構具備6 個知識模組: (1)知識表示(KR):使用 SCORM 標準、本論文提出
的教學活動模型(IAM)與物件導向活動模型(OOLA)來表示與管理學習內容及活
動、(2)知識資源(KRes):儲存學習活動、學習物件、試題、應用程式與學習歷程等
學習資源於所屬之資源庫中、(3)知識管理(KM):應用叢集(Clustering)技術與負載平
衡策略，提出階層式內容管理機制(Level-wise Content Management Scheme, LCMS)
來有效管理大量的學習物件、(4)知識擷取(KA):提供教師有用的工具來製作
SCORM 與 OOLA 相容的課程與活動，其包含應用高階派翠網路(High Level Petri Nets, HLPN)來分析 SCORM 導引規則而提出的物件導向課程朔模(Object Oriented Course Modeling, OOCM)機制、(5)知識控制(KC):智慧地根據學生的學習成效來提
供客製化的學習內容、服務、與試題，以進行適性學習、及(6)知識探勘(KMin): 應用資料探勘技術來分析學習歷程資料以建構適性化學習課程與自動地建構學習 概念圖。最後，為評估 ILCMS，針對每ㄧ知識模組，發展各個相對應的系統功能 與實際進行實驗驗證。而藉由實驗結果可證實ILCMS 所架構的知識模組確實是可 行的，且有益於學習者與教師進行有效的學習與教學。 關鍵字: 共享內容物件參考模型(SCORM)、網路學習、知識管理、學習內容管理、 適性化學習環境、資料探勘、學習物件、學習歷程分析、概念圖建立。
Design and Implementation of an Intelligent Learning
Content Management System based on SCORM Standard
Student: Jun-Ming Su Advisor: Prof. Wen-Nung Tsai Advisor: Prof. Shian-Shyong Tseng
Department of Computer Science College of Computer Science National Chiao Tung University
Abstract
With the vigorous development of the Internet, e-learning systems have become more and more popular. Currently, in order to solve the issue of sharing and reusing teaching materials in different e-learning systems, several standard formats, including SCORM of ADL, CP and QTI of IMS, LOM of IEEE LTSC, CMI of AICC, etc., have been proposed by international organizations. Among these international standards, the Sharable Content Object Reference Model (SCORM) has become the most popular standard in recent years.
SCORM is a set of specifications for developing, packaging and delivering high-quality education and training materials whenever and wherever they are needed. Although SCORM has many advantages of reusing, sharing, and recombining teaching materials among different standards, it is difficult to create, retrieve, and manage the SCORM compliant course with personalized learning sequences. Moreover, if the same teaching materials are provided to all learners based on predefined strategies, the leaning efficiency will be diminished. Thus, in recent years, adaptive learning
environments have been proposed to offer different teaching materials for different students in accordance with their aptitudes and evaluation results. Therefore, for the intelligent e-learning system based upon SCORM standard, how to efficiently create and manage the SCORM compliant learning contents with desired learning sequencing and teaching strategies, how to automatically generate appropriate learning activity for learners according to individual learning portfolio, personal aptitude, and teaching strategies, and how to evaluate the historical learning portfolio for understanding the mis-concept of learners are our concerns.
Currently, the IEEE’s LTSC proposed a Learning Technology System Architecture (LTSA) which is as a reference model and identifies the critical interoperability interfaces for learning technology systems. In addition, in order to support the interoperability and scalability of distributed e-learning system, IMS Abstract Framework (AF) and E-Learning Framework (ELF) propose the e-learning system models with layering concept, each layer of which defines different functionalities according to the different requirements of an e-learning system.
Furthermore, based on the Knowledge Management concept, how to efficiently manage the different resources and information in an adaptive e-learning system is similar to efficiently manage diverse knowledge. Therefore, based on this concept and LTSA with layering concept, in this dissertation, an Intelligent Learning Content Management System (ILCMS) is proposed to intelligently manage a large number of learning contents and offer learners an adaptive learning strategy which can be refined by means of efficient learning portfolio analysis.
The layered architecture of ILCMS consists of six knowledge modules: 1) Knowledge Representation (KR), which uses SCORM standard, and proposed Instructional Activity Model (IAM) and Object Oriented Learning Activity (OOLA) model to represent and manage the learning content and activity, 2) Knowledge
Resources (KRes), which stores related learning resources including Learning Activity,
Learning Object, Test Item, Application Program, and Learning Portfolio in respective repositories, 3) Knowledge Manager (KM), which includes a Level-wise Content Management Scheme (LCMS), applying clustering approach and load balancing strategies, to efficiently manage a large number of learning resources, 4) Knowledge Acquirer (KA), which provides teachers with useful tools to create the SCORM and OOLA compliant learning content and activity by means of proposed Object Oriented Course Modeling approach based on High Level Petri Nets and OOLA model, 5) Knowledge Controller (KC), which intelligently delivers the desired learning contents,
services, test sheet to learners according to her/his learning results and performance, and 6) Knowledge Miner (KMin), which applies data mining techniques to analyze the
learning portfolio for constructing the adaptive learning course and the learning concept map automatically. Finally, in order to evaluate ILCMS, system implementations and experiments have been done for each knowledge module. Also, the experimental results shows that proposed knowledge modules of ILCMS are workable and beneficial for learners and teachers.
Keywords: Sharable Content Object Reference Model (SCORM), E-Learning, Knowledge Management, Learning Content Management, Adaptive Learning Environment, Data Mining, Learning Object, Learning Portfolio Analysis, Concept Map Construction.
誌
謝
本論文能順利完成，首先需要感謝的便是我的指導教授 蔡文能教
授與 曾憲雄教授。如沒他們在我博士求學生涯中，遭遇最艱困的時
候，適時的伸出援手，並給予我支持與鼓勵，則今日我無法順利完成
此論文。
尤其更加感謝 曾憲雄教授，在曾教授不厭其煩的指導下，讓我對
此博士論文的研究領域從陌生到熟悉，從疑惑到了解。曾教授對我的
指導並非僅止於論文研究，他豐富的實務經驗與做人處事態度，著實
令我獲益良多，因此，在這博士求學過程中，我學到的不僅僅是研究
的方法，更獲得許多寶貴的經驗與生活的價值體認。雖然感謝二字不
足以形容整個過程，但是，心中最真誠的感謝仍想要藉此機會表達。
此論文的完成，也非常感謝從校內口試到校外口試過程中，一路
給予我許多論文修改建議的 李素瑛教授與 孫春在教授；以及在校
外口試中，給予我寶貴意見的中央大學 陳國棟教授、台灣師範大學
葉耀明教授、中央大學 楊鎮華教授與清華大學 張智星教授，讓此
論文能夠更加完整與契合。
當然，也不能忘了在這論文研究其間，知識工程實驗室所有和我
一起研究、打拼的研究夥伴們，亦讓我深切體認團隊合作的重要與價
值。其中，尤其感謝瑞鋒在此論文研究上的辛苦幫忙與協助，讓此研
究更加順利與扎實。
最後，家人的體諒與支持，是讓我能夠順利且安心完成博士學業
的最大後盾，雖然此段求學過程，相當漫長與坎坷，遇到不少挫折與
挑戰，如無父母親無怨無悔的支持與鼓勵，相信今日我難以順利完成
此論文研究，順利取得博士學位。當然，更加感謝我那美麗賢慧、善
良體貼的女友 蕙瑜，ㄧ路陪伴我走過這漫長與艱困的博士求學生涯，
多虧她的付出與支持，讓本論文充滿了更多的愛與活力，實在衷心感
謝。
僅將此份倫文，獻給所有支持我及我愛的家人、師長、女友與朋友
們。
Table of Contents
Abstract (In Chinese)………...………I Abstract (In English)……….………III Acknowledgement………VI Table of Contents………...………VII List of Figures………….………...………IX List of Tables………….………..……..………XII List of Algorithms………….………..………XIV Chapter 1 Introduction...1
Chapter 2 Related Works ...7
2.1 Intelligent Tutoring System and Adaptive Learning Environment...7
2.2 International Standards in E-Learning System ...9
2.3 The SCORM Compliant Authoring System ...12
2.4 Applying Petri Nets in E-Learning System...13
2.5 Structured Document Management...15
2.6 Learning Portfolio Analysis ...15
2.7 Concept Map Construction ...17
Chapter 3 Intelligent Learning Content Management System (ILCMS) ...19
3.1 The Layered Model of IEEE LTSA ...19
3.2 The Architecture of ILCMS...21
Chapter 4 Knowledge Representation (KR)...26
4.1 Sharable Content Object Reference Model (SCORM) ...26
4.2 Instructional Activity Model (IAM)...30
4.3 Object Oriented Learning Activity (OOLA) Model ...48
Chapter 5 Knowledge Acquirer (KA)...51
5.1 Transformation of Traditional Teaching Material ...51
5.2 Object Oriented Course Modeling (OOCM)...57
5.3 Object Oriented Learning Activity Authoring Tool...75
Chapter 6 Knowledge Manager (KM) ...78
6.1 Level-wise Content Management Scheme (LCMS) ...78
6.2 Searching Process of LCMS ...92
Chapter 7 Knowledge Controller (KC)...97
7.2 The Learning Process of OOLA based Learning Activity...98
Chapter 8 Knowledge Miner (KMin) ...100
8.1 Learning Portfolio Analysis Using Data Mining Approach ...100
8.2 Two-Phase Concept Map Construction (TP-CMC)...114
Chapter 9 Implementation and Experimental Results ...131
9.1 Learning Content Editor (LCE) in KA Module ...131
9.2 OOLA Authoring Tool in KA Module ...138
9.3 Learning Object Repository Manager in KM Module...145
9.4 Learning Portfolio Analyzer (LPA) in KMin Module ...150
List of Figures
Figure 2.1: The LTSA System Components... 11
Figure 3.1: The Layered Model of IEEE LTSA...20
Figure 3.2: Applying Knowledge Management Concept to Layered Model of LTSA ..20
Figure 3.3: The Layered Architecture of Intelligent Learning Content Management System (ILCMS)...22
Figure 4.1: SCORM Content Packaging Scope and Corresponding Structure of Teaching Materials...27
Figure 4.2: An Example of Activity Tree (a) with Clusters (b)...29
Figure 4.3: The Concept of Modularizing an AT ...31
Figure 4.4: The Diagram of IAM...32
Figure 4.5: The Flowchart of AT Selecting Process...34
Figure 4.6: The Diagram of Remedy Course Process...37
Figure 4.7: The Example of IAM...37
Figure 4.8: An Example of IAM with Pedagogical Theories...42
Figure 4.9: IAM Mapping to Discrete Structure, Linear Structure, and Hierarchical structure ...44
Figure 4.10: IAM Mapping to Spiral Curriculum and Lattice Curriculum. ...44
Figure 4.11: The Design of IAM in Part of “Introduction to Computer” ...47
Figure 4.12: The Diagram of OOLA Model ...50
Figure 4.13: An Example of Representing an Adaptive Learning Activity by OOLA ..50
Figure 5.1: Traditional Teaching Material and Concept of Learning Object...53
Figure 5.2: Diagram of Three Modes for Standardized Transformation Scheme of Traditional Teaching Material...55
Figure 5.3: Flowchart of Content Transformation Scheme (CTS)...56
Figure 5.4: The idea of Object Oriented Course Modeling (OOCM)...57
Figure 5.5: The Flowchart of Object Oriented Course Modeling (OOCM) ...58
Figure 5.6: The Diagram of HLPN of OOAT ...61
Figure 5.7: The Five Sequencing Components of OOATs...64
Figure 5.8: The Structure of Sequencing Rules ...66
Figure 5.9: An Example of modeling Skip Action in Sequencing Rules by Conditional Choice OOAT ...67
Figure 5.10: The Process of Objective Reference...68
Figure 5.11: The Structure of Rollup Rules ...69
Figure 5.12: The Rollup Model in OOATs...69
Figure 5.13: An Example of PN2AT and AT2CP Process ...71
Figure 5.14: The HLPNs Model and AT Structure of Course “PhotoShop”...74
Figure 6.1: The Flowchart of Level-wise Content Management Scheme (LCMS)...80
Figure 6.2: The Corresponding Content Tree (CT) of the Content Package (CP) by CP2CT process ...82
Figure 6.3: The Flowchart of Level-wise Content Clustering Algorithm (LCCAlg) ....86
Figure 6.4: An Example of Creating Level-wise Content Clustering Graph (LCCG)...89
Figure 6.5: An Example of LCCG Maintaining Process ...91
Figure 6.6: The Searching Process in LCMS...94
Figure 6.7: The Diagram of Near Similarity According to the Query Threshold Q and Clustering Threshold T ...95
Figure 7.1: The Diagram of Rule Inference Process in KC Module...98
Figure 7.2: The Learning Process of Running OOLA Model...99
Figure 8.1: The Flowchart of LPM ...102
Figure 8.2: Learning Pattern Extraction Phase ...106
Figure 8.3: Maximal frequent sequential pattern mining algorithm ...107
Figure 8.4: Mining process of modified GSP algorithm withα= 6 ...108
Figure 8.5: The decision tree based upon the learner profiles in Table 6 ... 111
Figure 8.6: The algorithm of personalized activity tree creation (PATC)... 112
Figure 8.7: The result of PATC algorithm based upon cluster 2 ... 113
Figure 8.8: The Flowchart of Two-Phase Concept Map Construction (TP-CMC)... 115
Figure 8.9: The given membership functions of each quiz’s grade... 118
Figure 8.10: The Fuzzification of Learners’ Testing Records... 118
Figure 8.11: Fuzzy Item Analysis for Norm-Referencing (FIA-NR)... 119
Figure 8.12: Look ahead Fuzzy Association Rule Mining Algorithm (LFMAlg) ...121
Figure 8.13: The Mining Process of Large 2 Itemset...121
Figure 8.14: The Transforming of Association Rules. ...125
Figure 8.15: Concept Map Constructing (CMC) Algorithm...126
Figure 8.16: The Process of Concept Map Constructing Algorithm...128
Figure 8.17: The (a) and (b) created based up analyzing L-L rule type only. The (c) and (d) are created based upon Anomaly Diagnosis and analyzing L-L rule type only. The (e) and (f) created by our approach...129
Figure 9.1: The Index Page of CTS System...132
Figure 9.2: The Process of PowerPoint File Transformation...132
Figure 9.3: The Prototypical Architecture of OOCM Authoring Tool...134
Figure 9.4: The Screenshot of the OOCM Authoring Tool...135
Figure 9.5: The Screenshot of Course “PhotoShop” Executed on SCORM RTE 1.3. 136 Figure 9.6: The Histogram of the Time Cost ...137
Figure 9.7: The Implementation of OOLA Authoring Tool ...139
Figure 9.8: The Screenshot of Constructing a complex learning activity by OOLA Tool ...140
Figure 9.9: Scaffolding Instruction by OOLA ...141
Figure 9.10: The F-measure of Each Query...146
Figure 9.11: The Executing Time Using LCCG-CSAlg ...146
Figure 9.12: The Comparison of SLCAlg and LCCAlg with Cluster Refining ...147
Figure 9.13: The Screenshot of LCMS Prototypical System in KM Module...148
Figure 9.14: The Results of Accuracy and Relevance in Questionnaire (10 is the highest) ...149
Figure 9.15: The learning process of training system to acquire learners’ learning behavior ...151
Figure 9.16: The SCORM learning course executed on SCROM run time environment (RTE) ...151
Figure 9.17: The concept maps (a), (b), and (c) with Discrimination 0.0, 0.3, and 0.5 are created by TP-CMC approach respectively. (Support=50, Confidence=0.85) ...156
List of Tables
Table 2.1: Relative Skills Frequency...17
Table 4.1: The Definitions of Related Symbols in IAM ...33
Table 4.2: The Related Values of AT1 and AT2...38
Table 4.3: The Symbol Definitions of Pedagogical Theory in IAM ...40
Table 4.4: Learning style and logical organization of each AT. ...42
Table 4.5: Selecting Criteria for Each Activity Tree. ...43
Table 4.6: The Content-Contribution Relationship Table of Course...46
Table 4.7: The weight matrix of contribution B...46
Table 4.8: The weight matrix of contribution C...46
Table 5.1: The Arc Expression Function E(a) and its Related Token Color. ...61
Table 5.2: The Related SDM definition of OOAT...63
Table 5.3: The Action Types and Corresponding OOATs of Precondition in Sequencing Rules...66
Table 5.4: The Action Types and Corresponding OOATs of Postcondition in Sequencing Rules...66
Table 8.1: The Learning Characteristics of Learners ...105
Table 8.2: The Learning Sequences of 10 Learners ...105
Table 8.3: The set of maximal frequent learning patterns (MF)...108
Table 8.4: The result of feature transforming process...109
Table 8.5: The result of applying ISODATA clustering algorithm... 110
Table 8.6: The learner profiles with cluster labels ... 111
Table 8.7: Test Item–Concept Mapping Table... 114
Table 8.8: Sorted Fuzzified Testing Grade on Q4... 119
Table 8.9: Difficulty and Discrimination Degree of Each Quiz ...120
Table 8.10: The Mining Results (Confi > 0.8)...122
Table 8.11: Prerequisite Relationship of Association Rule ...123
Table 8.12: The Explanations of Rule Types...124
Table 8.13: Result by Analyzing the Prerequisite Relationships in Table 8.11...124
Table 8.14: Relative Quizzes Frequency...127
Table 8.15: The Result of Computing the Influence Weight of Concept-Pair (B, D) in Figure 8.16.f...127
Table 9.1: The pretest-posttest of learning achievement ...142
Table 9.2: The one-group pretest-posttest t-test ...142
Table 9.3: The pretest-posttest of learning achievement of high grade group ...143
Table 9.4: The one-group pretest-posttest t-test of high grade group ...143
Table 9.5: The pretest-posttest of learning achievement of low grade group ...144
Table 9.7: t-Test of the test results (α=0.05)...153 Table 9.8: The Related Statistics of Testing Results in Physics Course...155 Table 9.9: Concepts List of Testing Paper in Physics Course ...155
List of Algorithms
Algorithm 4.1: AT Selection Algorithm...36
Algorithm 5.1: PN2AT Algorithm ...71
Algorithm 5.2: AT2CP Algorithm...72
Algorithm 5.3: OOLA Model to NORM Rule (OOLA2NORM)...77
Algorithm 6.1: Content Package to Content Tree Algorithm (CP2CTAlgo)...83
Algorithm 6.2: Single Level Clustering Algorithm (SLCAlg) ...87
Algorithm 6.3: Level-wise Content Clustering Algorithm (LCCAlg)...89
Algorithm 6.4: LCCG Maintaining Algorithm...92
Algorithm 6.5: LCCG Content Searching Algorithm (LCCG-CSAlg) ...96
Algorithm 8.1: Modified GSP Algorithm ...107
Chapter 1 Introduction
As internet usage becomes more and more popular over the world, e-learning system including online learning, employee training courses, and e-book in the past ten years has been accepted globally [6] [13] [20] [52] [53] [61] [69] [86] [88] [99] [130] [131] [138] [143]. In 2000, Urdan et al. [128] considered that e-learning is defined more narrowly than distance learning and defined it as “the delivery of content via all electronic media, including the Internet, intranets, extranets, satellite broadcast, audio/video tape, interactive TV, and CD-ROM“. E-learning system can make learner conveniently study at any time and any location. However, because the teaching materials in different e-learning systems are usually defined in specific data format, the sharing of the materials among these systems becomes difficult, resulting in increasing the cost of creating teaching materials. In order to solve the issue of the uniform teaching materials format, several standard formats including SCORM (Sharable Content Object Reference Model) of ADL [100], CP (Content Packaging) and QTI (Question & Test Interoperability) of IMS [56], CMI (Computer-Managed Instruction) of AICC [1], LOM (Learning Objects Metadata) of IEEE LTSC [79], etc. have been proposed by international organizations. By these standard formats, the teaching materials in different learning management systems can be shared, reused, and recombined.
SCORM is a set of specifications for developing, packaging and delivering high-quality education and training materials whenever and wherever they are needed. SCORM-compliant courses leverage course development investments by ensuring that compliant courses are Reusable, Accessible, Interoperable, and Durable (RAID). Although SCORM has many advantages of reusing, sharing, and recombining teaching materials among different standards, it is difficult to create, retrieve, and manage the
SCORM compliant course with personalized learning sequences based on the pedagogical theory. For example, the work to create the SCORM compliant teaching materials is still hard, even using the authoring tools. This leads to that teachers or editors may be unwilling to use it.
As we know, if the same teaching materials are provided to all learners based on the predefined strategies or the predefined learning maps, the leaning efficiency will be diminished. Thus, in recent years, adaptive learning environments [22] [43] [98] [111] [113] [122] [123] [140] have been proposed to offer different teaching materials for different students in accordance with their aptitudes and evaluation results. After students learn the teaching materials through the adaptive learning environment, the teachers can further analyze the historical learning records and then refine or reorganize the teaching materials and tests if needed. Therefore, more and more attention has been paid to the research of personalized instruction in computer education environment.
Moreover, because sequencing can help to generate teaching materials which can match the learner’s needs, (semi-)automatic sequencing of course materials also becomes an important research issue. However, although the personalized instruction scheme has been emphasized in most of existing e-learning systems, these systems, unfortunately, may not show good personalized and intelligent abilities.
Therefore, to sum up above, for the intelligent e-learning system, the following issues are needed to be solved.
z How to propose a scheme to efficiently create and manage the SCORM compliant learning contents with the desired learning sequencing.
z How to propose a scheme to efficiently create and manage the teaching strategies. z How to propose an intelligent approach which can automatically generate
appropriate learning activity for learners according to the individual learning portfolios, personal aptitudes, and teaching strategies.
z How to propose an efficient approach to evaluate the historical learning portfolio for understanding the mis-concept of learners.
At present, the international organization, IEEE LTSC, analyzed the basic requirements of e-learning system to propose a Learning Technology System Architecture (LTSA) [80] which is as a reference model and identifies the critical interoperability interfaces for the learning technology systems. LTSA, including 4 processes and 2 stores, that is, Learner Entity, 2) Coach, 3) Delivery, 4) Evaluation, 5) Learner Record, and 6) Learning Resource, can provide learners with an adaptive
learning environment.
In addition, in order to support the interoperability and scalability of distributed e-learning system, IMS Abstract Framework (AF) [55] proposes a layered model, which defines the interface definition set. Also, E-Learning Framework (ELF) [35] also proposes a layered model, each layer of which defines different functionalities according to the different requirements of an e-learning system. Therefore, based on the layered models of IMS AF and ELF, LTSA reference model can be reorganized into 4 layers: Resources, Common Services, Learning Services, and Application, according to the functions of its components.
Besides, because IEEE LTSA is as a reference model of building an e-learning system in support of adaptive learning, it does not clearly specify and define the data format of learning content and activity. Therefore, in order to solve the issue of uniform data format among e-learning systems, how to define the data representation format of learning content and activity is a very important issue.
Furthermore, based on the Knowledge Management concept [39], how to efficiently manage the different resources and information in an adaptive e-learning system is similar to efficiently manage diverse knowledge. Therefore, based on this concept and
IEEE LTSA [80] with layering concept, in this dissertation, an Intelligent Learning Content Management System (ILCMS) is proposed to intelligently manage a large number of learning contents and offer learners an adaptive learning strategy which can be refined by means of efficient learning portfolio analysis. The layered architecture of ILCMS consisting of six knowledge modules in corresponding layer respectively, i.e., 1) Knowledge Representation (KR), which uses SCORM standard, and new proposed Instructional Activity Model (IAM) [115] and Object Oriented Learning Activity (OOLA) model to represent and manage the learning content and activity, 2) Knowledge Resources (KRes), which stores all related learning resources in
repositories, 3) Knowledge Manager (KM), which efficiently manage a large number of learning resources in repositories, 4) Knowledge Acquirer (KA), which provide teachers with useful tools to create the SCORM and OOLA compliant learning content and activity, 5) Knowledge Controller (KC), which intelligently deliver the desired learning contents, services, test sheet to learners according to her/his learning results and performance, and 6) Knowledge Miner (KMin), which analyzes the learning portfolio to analyze the learning portfolio for constructing the adaptive learning course and the learning concept map automatically.
As mentioned above, the relationship of six knowledge modules in ILCMS are described as follows. First of all, KRes Module consists of five types of learning resources, i.e., Learning Activity, Learning Object, Test Item, Application Program, and Learning Portfolio, which are described by data formats: SCORM and OOLA model defined in KR Module and stored in their respective repositories. Then, KA module includes a Learning Content Editor (LCE) and an Object Oriented Learning Activity (OOLA) authoring tool [81]. In LCE, for reusing the existing traditional teaching materials, such as HTML and PPT file format, a Content Transformation Scheme (CTS) [114] has also been proposed. CTS approach can
divide a traditional teaching material into separate learning objects with SCORM metadata and then package them into one SCORM course. Moreover, in order to edit SCORM 2004 compliant learning contents, an Object Oriented Course Modeling (OOCM) [117] approach based upon High Level Petri Nets (HLPN) theory [59] [60] [62] [70] [71][73] [82] [84] has been proposed. OOCM can provide teachers or editors with an authoring tool to efficiently construct the SCORM compliant course with desired sequencing behaviors. Furthermore, OOLA authoring tool can help teachers construct an OOLA learning activity with desired teaching strategy. Moreover, KM module includes a Learning Object Repository (LOR) Manager, where we apply clustering approach and load balancing strategies to propose a management approach, called Level-wise Content Management Scheme (LCMS) [116], to efficiently maintain, search, and retrieve the learning contents in SCORM compliant LOR. When learners initiate a learning activity, the Learning Activity Controller (LAC) in KC module will retrieve the appropriate learning objects in LOR, testing sheets in Testing Item Bank (TIB), or application program (AP) in AP Repository (APR) according to the personalized learning activity in Learning Activity Repository (LAR) for learners. As mentioned above, the learning contents, test sheet, and AP will be retrieved and triggered according to the specific learning strategy. Those strategies are created by teachers using the authoring tools in KA module. Furthermore, KMin module includes a Learning Portfolio Analyzer (LPA), which consists of Learning Portfolio Mining (LPM) [118] and Two-Phase Concept Map Construction (TP-CMC) [110]
algorithms. According to learners’ characteristics, the former applies the clustering and decision tree approach to analyze the learning behaviors of learners with high learning performance for constructing the adaptive learning course. The latter applies Fuzzy Set Theory and Data Mining approach to automatically construct the concept map by learners’ historical testing records. Therefore, after the learners finished the learning
activities, teachers can use LPA module to analyze the learning portfolios of learners for refining their teaching strategies and contents.
The rest of this dissertation is organized as follows. Chapter 2 surveys the background knowledge of this work. Chapter 3 describes the layered architecture of LTSA and introduces the six modules of ILCMS. From Chapter 4 to Chapter 7, the details of Knowledge Representation (KR), Knowledge Acquirer (KA), Knowledge Manager (KM), Knowledge Controller (KC), and Knowledge Miner (KMin), are described. The system implementation and experimental results of ILCMS are shown in Chapter 9, and finally conclusion and future work are given in Chapter 10.
Chapter 2 Related Works
2.1 Intelligent Tutoring System and Adaptive Learning
Environment
In 1989, Johnson et al. [61] proposed a software design and development research program called Microcomputer Intelligence for Technical Training (MITT). In order to organize system knowledge and operational information for enhancing the operator performance, Vasandani et al. also developed an intelligent tutoring system [130] [131]. Furthermore, Hwang proposed an intelligent tutoring environment to detect the on-line behaviors of students [52]. Afterward, many related articles had also been proposed to develop the tutoring systems and learning environments [53] [69] [88][138] [143].
In adaptive learning environment, Shang [111] proposed an intelligent environment for active learning to support the student-centered, self-paced, and highly interactive learning approach. The learning environment can use the related learning profile of student, e.g., learning style and background knowledge, to select, organize, and present the customized learning materials for students. Trantafillou [122] also proposed an adaptive learning system, called AHS, in which Learners can be divided into two groups with Field Independence (FI) and Field Dependence (FD) respectively according to their cognitive styles. Then, the AHS system can provide appropriate strategy and learning materials for different groups. Moreover, according to learning styles and learning experience of learners, Gilbert [43] applied the Case Based Reasoning (CBR) technique to assign a new learner to the most similar one of four groups. Based upon the learning experience in group selected by CBR, the proposed system can offer the new learner an adaptive learning material. However, in all systems mentioned above, the information and approaches used to represent and group learners respectively are too easy to provide learners with personalized learning materials.
Carchiolo et al. [22] had proposed adaptive formative paths for e-learning environments. They constructed a domain database and student profiles to obtain personalized learning paths. During the learning process, the learning paths can be dynamically modified according to student needs and capabilities. Although this system has some advantages, including consideration of each student’s prior knowledge and generation of an adaptive learning path, it does not take pedagogical theory into account, and it is not yet compatible with the SCORM standard. Sheremetov and Arenas [98] also proposed a system, called EVA, for developing a virtual learning space at the National Technical Institute in Mexico. EVA consists of five virtual learning spaces: 1. the Knowledge Space, in which all necessary information exists; 2. the Collaborative Space, in which real or virtual companions get together to learn; 3. the Consulting Space, in which the teachers or tutors (also real or virtual) guide learning and provide consultation; 4. the Experimentation Space, in which the practical work is done by the students in the virtual environment; and 5. the Personal Space, in which records of users are stored. The model of knowledge is represented in the form of graph, where each node, the basic element of the knowledge structure, is a unit of learning material (ULM). ULMs with a related knowledge concept can be grouped into a POLIlibro (or Multi-Book) along the learning trajectory (path), depending on the students. However, the relations between ULMs are not sufficient to express the structure of the knowledge model, and the attributes of a ULM are insufficient for mining the behaviors of students. The authors also proposed some methods for planning trajectories and scheduling learning activities based on the agent technology. However, how to generate a learning path was not discussed.
Therefore, the development of intelligent tutoring system (ITS) or adaptive learning system (ALE) has become an important issue in both computer science and education.
2.2 International Standards in E-Learning System
However, most existing e-learning systems represent student profile, learning management data, test bank and subject contents with different formats, which results in the difficulties of sharing, reusing, and recombining those e-learning resources. Therefore, several international organizations have proposed teaching material standards, such as SCORM proposed by IMS, Simple Sequencing Specification and Content Packaging proposed by IMS, and LOM proposed by IEEE LTSC.
2.2.1 IMS (Instructional Management System)
In 1997, the IMS Project [56], which is part of the nonprofit EDUCAUSE [33], started its work and developed open, market-based standards including specifications of learning resource metadata for online learning. In the same year, the NIST (National Institute for Standards and Technology) and the IEEE P.1484 group, which now is the IEEE Learning Technology Standards Committee (LTSC) [79], also started to do a similar effort. Then, the IMS collaborated with NIST and ARIANDE project [3]. In 1998, IMS and ARIADNE submitted a joint proposal and specification to the IEEE, which is the basis of current IEEE Learning Object Metadata (LOM) base document. Currently, the IMS project have proposed many standard specifications including learning metadata specification, content packaging specification, learner profiles specification, question and test interoperability, and simple sequence specification, etc.
2.2.2 IEEE LTSC
The international organization, IEEE LTSC [79], proposed the Learning Technology System Architecture (LTSA) and Learning Objects Metadata (LOM), described as follows.
LOM (Learning Objects Metadata):
The IEEE’s Learning Objects Metadata (LOM) [79] describes the semantics of learning object metadata. Here, a learning object is defined as any entity, including multimedia content, instructional content, and instructional software, which can be used, reused, shared, and recombined. To allow learning objects to be managed, located, and evaluated, the LOM standard makes efforts in the minimal set of properties needed.
The LOM describes learning resources by using the following categories. z General: describe the general information of learning resource.
z LifeCycle: describe the history and current state of learning resource and its evolution information.
z Meta-MetaData: describe the specific information about the metadata record itself, e.g., who created this metadata record, etc.
z Technical: describe the technical requirements and characteristics of learning resource.
z Educational: describe the key educational or pedagogic characteristics of learning resource.
z Rights: describe the intellectual property rights and conditions of use for learning resource.
z Relation: define the relationships among this resource and other targeted resource.
z Annotation: provide comments on the educational use of learning resource, e.g., who created this annotation.
z Classification: describe classification criteria and hierarchy of learning resource.
Learning Technology System Architecture (LTSA):
IEEE LTSC analyzed the basic requirements of e-learning system to propose a Learning Technology System Architecture (LTSA) [80] which identify the critical interoperability interfaces for learning technology systems. LTSA mainly includes 4 processes and 2 stores, described as follows:
(1) Learner Entity: learners receive the multimedia learning contents delivered by system and the learning progress of learners will be tracked and recorded. (2) Coach: teachers provide learning system with teaching materials and evaluate
the learning performance of learners.
(3) Delivery: it is responsible for delivering the learning contents Coach indicates to learners.
(4) Evaluation: it evaluates the learning performance of learners and diagnoses the mis-concept.
(5) Learner Record: it records the learning behavior of learners, which can be used to analyze and track.
(6) Learning Resource: it stores the learning resources which were created by teachers and can be used to learn for learners.
Figure 2.1 illustrates the components of LTSA system.
2.2.3 SCORM (Sharable Content Object Reference Model)
Among those existing standards for learning contents, SCORM [100], which was proposed by the U.S. Department of Defense’s Advanced Distributed Learning (ADL) organization in 1997, is currently the most popular one. The SCORM specifications are a composite of several specifications developed by international standards organizations, including the IEEE LOM [79], IMS [56], AICC [1] and ARIADNE [3]. In a nutshell, SCORM is a set of specifications for developing, packaging and delivering high-quality education and training materials whenever and wherever they are needed. SCORM-compliant courses leverage course development investments by ensuring that compliant courses are "RAID:" Reusable: easily modified and used by different development tools, Accessible: can be searched and made available as needed by both learners and content developers, Interoperable: operates across a wide variety of hardware, operating systems and web browsers, and Durable: does not require significant modifications with new versions of system software [58]. The details of SCORM and its Sequencing & Navigation (SN) [109] will be described in Chapter 4.
2.3 The SCORM Compliant Authoring System
Recently, although many SCORM authoring tools have been developed by commercial companies, unfortunately, these tools support SCORM 1.2 only, for example, the Authorware 7 of Macromedia [64], Click2learn Unveils SCORM 1.2 Resource Kit [23], Seminar Author of Seminar Learning System [105], Elicitus Content Publisher [31], and more other SCORM 1.2 compliant authoring tools found in [32].
Because the complicated sequencing rule definitions of SN in SCORM 2004 make the design and creation of course hard, the article in [76] has proposed several document templates to construct SCORM compliant course according to the sequencing
definitions of SN. Teachers/authors can design their desired learning activities by modifying the sequencing definitions in document templates. Then, the SCORM course with sequencing definitions can be created by programming. However, for teachers/authors, creating the SCORM course with sequencing behavior rules by document templates is still hard. Moreover, it is time consuming and high cost to create SCORM course by programming.
Moreover, an open source tool, called Reload Editor, developed by [94] can be used to create the SCORM 2004 course. For setting the learning guidance, users have to edit the sequencing rules by clicking in the comboBox of sequencing rules. Although it offers the graphical user interface (GUI) to create SCORM course, the sequence of final course is hard to image and creating course is also time-consuming. Shih et al. [103] also proposed a collaborative courseware authoring tool to edit the SCORM compliant course which can support collaborative authoring and suggest an optimal learning sequence. They analyzed the metadata of SCA in SCORM 1.3 to design the activity rules which can be used to generate lecture sequencing. This tool also offers users the sequencing rules definition page to define the sequencing behavior of courseware. Besides, Yang et al. [144] developed a web-based authoring tool, called Visualized Online Simple Sequencing Authoring Tool (VOSSAT), to provide an easy-to-use interface for editing existing SCORM-compliant content packages with sequencing rules. Nevertheless, the disadvantages in [103][144] are the same as Reload Editor [94].
2.4 Applying Petri Nets in E-Learning System
Lin [70] applied Petri Nets theory to model online instruction knowledge for developing online training systems. Two-level specialized Petri nets including TP-net, which represents goal-oriented training plans, and TS-net, which represents
task-oriented training scenarios, are proposed. A Goal-Oriented Training Model Petri net (GOTM-net), which is combined by a TP-net and all TS-nets, is converted as a set of “if-then” rules representing the behaviors a learner may perform and the corresponding responses. However, GOTM-net may not be compatible with SCORM standard. Based on SCORM 1.2, Liu et al. [71] discussed meta-data structure which makes a base for reusing and aggregating learning resources in e-learning, and provided an aggregation model, called Teach net, based on High-Level Petri Nets (HLPN). Several routing constructs in workflow are also modeled by HLPN for flexible navigation. However, the Teach net is mainly used to model the content aggregation without considering course sequencing. Besides, the modeled routing constructs may be not sufficient for modeling sequencing definition in SCORM 2004.
2.5 Structured
Document Management
For fast retrieving the information from structured documents, Ko et al. [63] proposed a new index structure which integrates the element-based and attribute-based structure information for representing the document. Based upon this index structure, three retrieval methods including 1) top-down, 2) bottom-up, and 3) hybrid are proposed to fast retrieve the information from the structured documents. However, although the index structure takes the element and attribute information into account, it is too complex to be managed for the huge amount of documents.
How to efficiently manage and transfer document over wireless environment has become an important issue in recent years. The articles [75] [142] have addressed that retransmitting the whole document is expensive in faulty transmission. Therefore, for efficiently streaming generalized XML documents over the wireless environment, Wong et al. [133] proposed a fragmenting strategy, called Xstream, for flexibly managing the XML document over the wireless environment. In the Xstream approach, the structural characteristics of XML documents has been taken into account to fragment XML contents into an autonomous units, called Xstream Data Unit (XDU). Therefore, the XML document can be transferred incrementally over a wireless environment based upon the XDU. However, how to create the relationships between different documents and provide the desired content of document have not been discussed. Moreover, the above articles [63] [75] [133] [142] didn’t take the SCORM standard into account yet.
2.6 Learning Portfolio Analysis
In addition, for learning portfolio analysis, Chen [15][16] applied decision tree and data cube techniques to analyze the learning behaviors of students and discover the pedagogical rules on students’ learning performance from web logs including the
amount of reading article, posting article, asking question, login, etc. According to their proposed approach, teachers can easily observe learning processes and analyze the learning behaviors of students for pedagogical needs. However, although their proposed approaches can observe and analyze the learning behavior of students, they don’t apply education theory to model the learning characteristics of learners. Therefore, the learning guidance can not be provided automatically for the new learner. For providing the personalized recommendation from historical browser behavior in e-learning system, Wang [140] proposed a personalized recommendation approach which integrates user clustering and association-mining techniques. Based upon a specific time interval, they divided the historical navigation sessions of each user into frames of sessions. Then, a new clustering method, called HBM (Hierarchical Bisecting Medoids Algorithm) was proposed to cluster users according to the time-framed navigation sessions. In the same group, the association-mining technique was used to analyze those navigation sessions for establishing a recommendation model. Thus, this system can offer the similar students personalized recommendations. However, in this approach, the learning characteristics and sequential learning sequence of students were not considered, so that the personalized recommendation may be not appropriate. Of course, it doesn’t support SCORM 2004 standard yet.
2.7 Concept Map Construction
In 1984, Novak [85] proposed Concept Map to organize or represent the knowledge as a network consisting of nodes (points/vertices) as concepts and links (arcs/edges) as the relations among concepts. Thus, a wide variety of different forms of concept maps have been proposed and applied in various domains [8][45][46]. In the adaptive learning environment, the Concept Map can be used to demonstrate how the learning status of a concept can possibly be influenced by learning status of other concepts and give learners adaptive learning guidance. Thus, Appleby proposed an approach to create the potential links among skills in math domain [5]. The direction of a link is determined by a combination of educational judgment, the relative difficulty of skills, and the relative values of cross-frequencies. Moreover, a harder skill should not be linked forwards to an easier skill. As shown in Table 2.1,
f
B
A represents the
amount of learners with wrong answers of skill A and right answers of skill B. If B
A B
A f
f 〉 , a skill A could be linked to a harder skill B, but backward link is not permitted.
Table 2.1: Relative Skills Frequency
A is right A is wrong B is right fAB f AB
B is wrong fAB fAB
Hsu also proposed a conceptual map-based notation, called Concept Effect Relationships (CER), to model the learning effect relationships among concepts [51]. In brief, for two concepts, Ci and Cj, if Ci is the prerequisite for efficiently learning the
may have multiple prerequisite concepts, and can also be a prerequisite concept of multiple concepts. Thus, based upon CER, the learning guidance of necessary concepts to enhance their learning performance can be derived by analyzing the test results of students. Later, based upon statistical prediction and approach of Hsu [51], a CER Builder was proposed by Hwang [49]. Firstly, CER Builder finds the test item that most students failed to answer correctly and then collects the other test items failed to answer by the same students. Thus, CER Builder can use the information to determine the relationships among the test items. Though the CER Builder is easy to understand, only using single rule type is not enough to analyze the prerequisite relationship among concepts of test items, which may decrease the quality of concept map.
Tsai proposed a Two-Phase Fuzzy Mining and Learning Algorithm [126]. In the first phase, Look Ahead Fuzzy Mining Association Rule Algorithm (LFMAlg) was proposed to find the embedded association rules from the historical learning records of students. In the second phase, the AQR algorithm was applied to find the misconcept map indicating the missing concepts during students learning. The obtained misconcept map as recommendation can be fed back to teachers for remedy learning of students. However, because the creating misconcept map, which is not a complete concept map of a course, only represents the missing learning concepts, its usefulness and flexibility are decreased. In addition, their approaches generate many noisy rules and only use single rule type to analyze the prerequisite relationship among learning concepts.
Chapter 3 Intelligent Learning
Content Management System
(ILCMS)
3.1 The Layered Model of IEEE LTSA
In order to provide learners with an adaptive learning environment, the Learning Technology System Architecture (LTSA) of IEEE LTSC [80] as a reference model identifies the critical interoperability interfaces for learning technology systems. In addition, in order to support the interoperability and scalability of distributed e-learning system, IMS Abstract Framework (AF) [55] proposes a layered model, which defines the interface definition set. Also, E-Learning Framework (ELF) [35] also proposes a layered model, each layer of which defines different functionalities according to the different requirements of an e-learning system. Therefore, based on the layered models of IMS AF and ELF, LTSA reference model can be reorganized into 4 layers: resources, common services, learning services, and application, according to the functions of its components. Figure 3.1 illustrates the layered LTSA model, where the module in higher layer will use the service provided from lower layer to offer more powerful and specific service. For example, the Delivery module in Common layer uses the resources in Resources layer to deliver to the learners.
Furthermore, based on the knowledge management concept [39], how to efficiently manage the different resources and information in an adaptive e-learning system is similar to efficiently manage diverse knowledge. Accordingly, each module in LTSA can be classified into five knowledge types according to its function, i.e., Knowledge Resources including learning resources and records, Knowledge Manager including the
Delivery, Knowledge Controller and Knowledge Acquirer including the Coach, and Knowledge Miner including the Evaluation, as shown in Figure 3.2.
Figure 3.1: The Layered Model of IEEE LTSA
3.2 The Architecture of ILCMS
As mentioned above, LTSA can be layered into 4 layers according to the service function of each layer and classified into 5 knowledge types based on knowledge management concept. However, because IEEE LTSA is as a reference model of building an e-learning system in support of adaptive learning, it does not clearly specify and define how to represent the learning content and activity. Therefore, in order to solve the issue of uniform data format among e-learning systems, how to define the data representation format of learning content and activity is a very important issue.
Therefore, in this dissertation, based on Knowledge Management concept [39] and layered IEEE LTSA [80], an Intelligent Learning Content Management System (ILCMS) is proposed to intelligently manage a large number of learning contents and offer learners an adaptive learning strategy which can be refined by means of efficient learning portfolio analysis. Figure 3.3 shows the layered architecture of ILCMS consisting of six knowledge modules in corresponding layer respectively, i.e., 1) Knowledge Representation, which uses SCORM standard, and new proposed Instructional Activity Model (IAM) and Object Oriented Learning Activity (OOLA) model to represent and manage the learning content and activity, 2) Knowledge Resources, which stores all related learning resources in repositories, 3) Knowledge
Manager, which efficient manages a large number of learning resources in repositories, 4) Knowledge Acquirer, which provides teachers with useful tools to create the
SCORM and OOLA compliant learning content and activity, 5) Knowledge Controller, which intelligently delivers the desired learning contents, services, test sheet to learners according to her/his learning results and performance, and 6) Knowledge Miner, which analyzes the learning portfolio for constructing the adaptive learning course and the learning concept map automatically.
Figure 3.3: The Layered Architecture of Intelligent Learning Content Management System (ILCMS)
Each knowledge module of ILCMS can be described in details as follows:
1. Knowledge Representation (KR): it includes 3 data models: SCORM, Instructional Activity Model (IAM) [115], and Object Oriented Learning
Activity (OOLA) [81], to represent the learning content and activity, respectively. As state previously, in order to share and reuse the contents among various learning systems, we use the popular SCORM standard to represent the teaching materials so that the issue of uniform content format can be solved. Moreover, in order to
efficient manage and reuse the large-scale Activity Tree (AT) with complex sequencing rules in SCORM. Therefore, we propose an Instructional Activity Model (IAM), which extends and modularizes the structure of AT with
inter-relation attributes by means of Pedagogical Theory and the concept of the Object Oriented Methodology, respectively. Furthermore, based on the modularized AT of IAM and object oriented concept, we further propose a model with sequencing rule definition, called Object Oriented Learning Activity (OOLA), to efficiently model a adaptive learning activity by means of three basic elements, that is, Content, Interaction, and Assessment. Thus, an adaptive learning activity can be easily created and offered to learners with a personalized learning contents, services, and assessment.
2. Knowledge Resources (KRes): it includes five types of learning resources, i.e., Learning Activity, Learning Object, Test Item, Application Program, and
Learning Portfolio, which are stored in their respective repositories and can be managed, reused, delivered, and analyzed by the sub-module of ILCMS in higher layers.
3. Knowledge Manager (KM): it includes a Learning Object Repository (LOR)
Manager, in which we analyze the content structure of SCORM and then apply clustering technique and load balancing strategies to propose a Level-wise Content
Management Scheme (LCMS)[117]. LCMS can automatically analyze the
SCORM compliant contents, group these related objects into a cluster, and then create the relation links among different clusters. Therefore, by means of LCMS, LOR manager can efficiently maintain, search, and retrieve the desired learning objects from the SCORM compliant LOR with a large number of learning objects. 4. Knowledge Acquirer (KA): it includes a Learning Content Editor (LCE) and an
Scheme (CTS)[114], which can efficiently transform the traditional teaching
materials, e.g., HTML and PPT file format, into SCORM compliant learning contents, and an SCORM 2004 compliant authoring tool with Object Oriented Course Modeling (OOCM) [117] approach based upon High Level Petri Nets (HLPN) theory [59] [60] [62] [70] [71] [73] [82] [84], which can help teachers or editors efficiently create the course with desired learning sequencing guidance of SCORM standard. These created SCORM compliant learning content will be stored in Learning Object Repository (LOR). In addition, in order to construct OOLA compliant learning activity, the latter is a user-friendly GUI authoring tool, by which teachers can efficiently edit desired learning activity with associated SCORM compliant course in LOR, test sheet in TIB, and application program (AP) in APR. AP like an interaction tool, e.g., chat room, browser, messenger, etc., offer learners to interact with other learners and teachers. These edited OOLA learning activities will be transformed into rule format and then stored in Learning Activity Repository (LAR).
5. Knowledge Controller (KC): includes a Learning Activity Controller (LAC), which includes a System Coordinator (SC) and an Inference Engine (IE) to provide learners with personalized learning contents, exercises, and test sheets according to different learner’s portfolios and teaching strategies.
6. Knowledge Miner (KMin): includes a Learning Portfolio Analyzer (LPA), which consists of Learning Portfolio Mining (LPM) [118] and Two-Phase Concept Map Construction (TP-CMC) [110] algorithm. According to learners’ characteristics, the former applies the clustering and decision tree approach to analyze the learning behavior of learners with high learning performance for constructing the adaptive learning course. The latter applies Fuzzy Set Theory and Data Mining approach to automatically construct the concept map by learners’
historical testing records. Therefore, after the learners finished the learning activities, teachers can use LPA module to analyze the learning portfolios of learners for refining their teaching strategies and contents.
After the explanation above, the relationship of five knowledge module in ILCMS are described as follows. First, LCE and OOLA authoring tool in KA module can offer teachers or editors to edit the new SCORM compliant learning contents or transforms existing traditional teaching materials into SCORM compliant ones, and construct an OOLA learning activity, respectively. Then, LOR Manager in KM module applies clustering approach and load balancing strategies to efficiently manage a large number of learning objects in LOR. When learners initiate a learning activity, the LAC in KC module will retrieve the appropriate learning objects in LOR, testing sheets in Testing Item Bank (TIB), or application program (AP) in APR according to the personalized learning activity in LAR for learners. As mentioned above, the learning contents, test sheet, and AP will be retrieved and triggered according to the specific learning strategy. Those strategies are created by teachers using the authoring tool in KA module. Besides, after the learners finished the learning activities, teachers can use the LPA in KMin module to analyze the learning portfolios of learners for refining their teaching strategies and contents.
The topics in this dissertation mentioned above will be detailedly discussed in following Sections.
Chapter 4 Knowledge Representation
(KR)
In this chapter, we describe the data format used to represent the learning resources in ILCMS. In order to share and reuse the contents among various learning systems, the popular SCORM standard is used to represent the teaching materials so that the issue of uniform content format can be solved. Moreover, in order to efficient manage and reuse the large-scale Activity Tree (AT) with complex sequencing rules in SCORM. Therefore, we propose an Instructional Activity Model (IAM), which extends and modularizes the structure of AT with inter-relation attributes by means of Pedagogical Theory and the concept of the Object Oriented Methodology, respectively. Furthermore, based on the modularized AT of IAM and object oriented concept, we further propose a learning activity model with sequencing rule definition, called Object Oriented Learning Activity (OOLA), to efficiently model a adaptive learning activity by means
of three basic elements, that is, Content, Interaction, and Assessment. Thus, an adaptive learning activity can be easily created and offered to learners with a personalized learning contents, services, and assessment. The details of SCORM, IAM, and OOLA model will be described below.
4.1 Sharable Content Object Reference Model (SCORM)
In SCORM specification, content packaging scheme is proposed to package the learning objects into standard teaching materials, shown in Figure 4.1. The content packaging scheme defines a teaching materials package consisting of 4 parts, that is, 1) Metadata: describes the characteristic or attribute of this learning content, 2) Organizations: describe the structure of this teaching material, 3) Resources: denote
Manifest: describes this teaching material consists of itself and another teaching
material. In Figure 4.1, the organizations define the structure of whole teaching material, which consists of many organizations containing arbitrary number of tags, called item, to denote the corresponding chapter, section, or subsection within physical teaching material. Each item as a learning activity can be also tagged with activity metadata which can be used to easily reuse and discover within a content repository or similar system and to provide descriptive information about the activity. Hence, based upon the concept of learning object and SCORM content packaging scheme, the teaching materials can be constructed dynamically by organizing the learning objects according to the learning strategies, students' learning aptitudes, and the evaluation results. Thus, the individualized teaching materials can be offered to each student for learning, and then the teaching material can be reused, shared, recombined.
Figure 4.1: SCORM Content Packaging Scope and Corresponding Structure of Teaching Materials
4.1.1 Sequencing and Navigation (SN)
of SCORM
At present, Sequencing and Navigation (SN) [109] in SCORM 1.3 (also called SCORM 2004) adopts the Simple Sequencing Specification of IMS [56] based on the concepts of learning activities, each of which may be described as an instructional event, as an event embedded in a content resource. The content in SN is organized into a hierarchical structure, namely, an activity tree (AT) as a learning map. An example of an AT is shown in Figure 4.2. Each learning activity, including one or more child activities, includes two data models: Sequencing Definition Model (SDM) including an associated set of desired sequencing behaviors of content designer and Tracking Status Model (TSM) including the information about a learner’s interaction with the learning objects within associated activities. SN uses information in SDM and TSM to control the sequencing, selection, and delivery of activities to the learner.
The sequencing behaviors describe how the activity or how the children of the activity are used to create the desired learning experience. SN places no restrictions on the structure, organization, or instruction of the activity tree. The tree and the associated sequencing definitions may be statically or dynamically created. Therefore, how to create, represent, and maintain the activity tree and associated sequencing definition, which is not specified, is an important issue. SN enables us to share not only learning contents but also intended learning experiences. It also provides a set of widely used sequencing methods so that the teacher could do sequencing efficiently. Accordingly, in this dissertation, SCORM standard is used to represent the learning contents associated with related learning object and sequencing rules.
