Evaluation of Multi Modal Biosignals and Biometrics in e-Learning(e ラーニングにおけるマルチモーダル生体測定の評価)

学位の種類 博士(工学)
学位記番号 博甲第572号
学位授与の日付 平成23年3月25日
学位論文題目 Evaluation of Multi Modal Biosignals and Biometrics in e-Learning (e ラーニングにおけるマルチモーダル生体測定の評価)
 主査 教授 福村 好美
 副査 教授 中村 和男
 副査 教授 三上 喜貴
 副査 准教授 湯川 高志
 副査 産学融合特任准教授 野村 収作

平成22(2010)年度博士論文題名一覧] [博士論文題名一覧]に戻る.

Table of contents

Acknowlegement p.iv

Abstract p.v

Description and Abbreviations of Technical Terms p.viii

1.Introduction p.1
1.1 Background p.6
1.2 Motivation p.7
1.3 Aim p.7

2.Theory and Background p.9
2.1 Learning and vivid Definition of Learning p.9
2.2 Traditional theories of leaning p.9
2.3 Summary p.14

3.Literature Survey p.15
3.1 Adaptation in E-learning p.15
3.2 Personalisation and Adaptation in Adaptive E-learning p.29
3.3 Applying Adaptive E-learning p.30
3.3.1 identifying Components, Functions and Interactions p.36
3.4 Summary p.59

4.Proposed Architecture p.60
4.1 The General Architecture:Specifications and Descriptions p.61
4.1.1 RLO Repository - Reusable Learning Objects Repository p.63
4.1.2 Course / Lesson Meta Data p.63
4.1.3 Course or Lesson Schema p.64
4.1.4 Adaptation Engine p.64
4.2 The User Modeler:Specifications and Descriptions p.65
4.2.1 Preference Profiler p.67
4.2.2 Behaviour Profiler p.67
4.2.3 Psychophysiological Profiler p.67
4.3 Summary p.68

5.Transcribing the Physiological Response for E-learning systems p.69
5.1 Introduction p.69
5.2 Method p.70
5.2.1 Subject p.70
5.2.2 Procedure p.70
5.2.3 Measurements p.74
5.2.4 Data Analysis p.74
5.3 Results p.76
5.3.1 Results of experiment 1 p.76
5.3.2 Results of experiment 2 p.79
5.4 Discussion p.82
5.5 Summary p.84

6.Biometrics and future adaptive E-learning systems p.86
6.1 Biometrics and E-leaning p.86
6.2 Face Recognition p.89
6.3 General Architecture of the System p.91
6.4 Summary p.96

7.Summary and Conclusion p.99

References p.102

List of figures p.111

 The use of E-learning can help students or learners to fit studying into their busy work and family lives, and will allow them to work at their own pace and place. The E-learning is experiencing an unprecedent growth and will continue to do so for the foreseeable future. On the other hand, bored students are dropping out of online classes while pleading for richer, adaptive and more engaging online learning experinece. Mnay adaptive system have been proposed and tested with E-learning system for providing better personalised online interactive learning environments. Howeferm, these adaptive E-learning systems, hitherto, have failed to address the core elemnt surrounding the online learning environments; that is the psychophysiological background of learner.

 E-learning is a complex human learning phenomenon. It is not only an educational phenomenon but also a comlex phenomenon that involves psychophysiological processes. Further, the E-learning is not jus the delivery of contents, rather, it is seen as the transition and building of knowledge.

 On the other hand, the learners hold the characteristics of different background, culture, interests, needs, experiences, prior-knowledge, talents, abilities, preferences and gender. Therefore in the present adaptive E-learning systems. personalisation is based upon the presumption that each learner follows a uniquely distinguishable strategy during learning process and shows varying preference in the consumption of learning objects. Further, the evaluated historical records of educational domain stipulate that learner psychophysiology is a crucial factor in the knowledge transition and knowlwdge building. Threore, it becomes very crucial to investigate the learner psychophysiology in the context of E-learning to provide a better adaptive environment for improved knowledge transition and better adaptation. In the present E-learning system, the learners are portrayed as models in orther to personalise or provide adapatation. These models are often known as student models. Given the importance of the E-learning, to provive an effective online learning environment, it is crucial to understand not only the browsing history and pattern of a learner but also the psychophysiological phenomenon surround the online learning students themselves and introduce the psychophysiological factor into student modeling. This dissertation evaluates multi modal biosignals to address the future psychophysiologically adaptive and interactive E-learning systems, and proposes a new architecture for the user modeling component of the future adaptive E-learning systems. Further, it is becoming very difficult for the tutors to precisely quantify the actual amount of time that a student spends on lessons. So far, the student log files have been the sole source of information that provides insight into the time that is spent by students on lessons. However, these log files do not always provide adequate and precise information; for an instance, when a student logs into the E-learning system and leaves the environment without logging out, the system still considers that time period as a genuine time that is spent on lessons. Therefore, the future adaptive E-learning systems should be able to automatically alert the learner and the tutor on occasions when the time that is spent on is considered inadequate.

 Moreover. Identification of learners in online examinations is also an unanswered puzzle in the context of E-learning. There are many readily available security measures, which can be utilized for access control mechanisms or the invigilation purpose. Nevertheless, the conventional mechanisms are not robust enough for the purpose of Invigilation; the username and password based access control is not appropriate for this purpose as these can be swapped with others. The high-level biometric access control mechanisms are neither a plausible choice as not everyone can afford a biometric hardware for a single occasion. Further, since the online examinations are conducted over a period of time of 1 hour or 2 hours, the access control system should be robust enough to guarantee the identity of the candidate throughout the examination session. It is also vital to conceive that the access control and invigilation system should be robust enough to automatically fuse the identity of the candidate without any intrusive method or user response during the course of the examination. In order to provide an answer and pave a new research path to the above problem domains, this thesis investigates multi modal biosignal and biometrics in E-learning.

 本論文は「Evaluation of Multi Modal Biosignals and Biometrics in e-Learning」と題して、7章より構成されている。第1章「Introduction」では、現状のeラーニングシステムにおける,受講者に即した環境構築に関する研究概況と課題を示すとともに、本研究の意義と目的を述べている。第2章では,学習システムに関する理論的な背景と,生体・心理学面での研究の重要性を提示し、第3章では、特にeラーニングにおける受講者適応型システムに関する研究状況を概観している。
 以上の研究状況を踏まえて、第4章では次世代のeラーニングシステムに関するアーキテクチャをモデル化し、本研究が対象とするeラーニングのユーザモデルを定義している。この中で、学習者の受講状態を識別するために、従来の学習ログに加えて、学習行動ログと生体情報収集・分析の必要性を述べている。第5章では、正規の授業として開発された非同期型eラーニングの教材コンテンツを用いて、受講者生体情報を実験により測定・分析した結果を述べている。すなわち、対話型教材(IM : Interactive Material)と、非対話型教材(N-IM : Non-Interactive Material)の2種類の異なる教材コンテンツを用いた授業実験により、生体情報の特徴的変化を抽出する。収集情報としては、皮膚電気抵抗値(SCL : Skin Conductance Level)、心電図(ECG:Electrocardiogram),血圧関連特性値(CO:Cardiac Output,BP:Blood Pressure,TPR:Total Peripheral resistance)を対象として,20名弱の被験者情報を収集した。その結果,SCL値についてはN-IMの場合には変化なくほぼ一定であるが,IMの場合には有意に低下傾向が確認できた。一方ECGに関しては両教材での差異はないことを確認した。すなわち,教材コンテンツの相違がSCLに影響していることを意味し,遠隔の学習者状態をSCL測定により推測できる。同様に,血圧関連値COの測定結果から,両教材間の差異が明示された。これらの結果を応用することにより,受講中の学習者の心理状態を判定して,学習のシーケンスを制御する,あるいは形成的評価として教材の改善が期待できる。