Using ontological engineering to Support the development of collaborative learning applications An Integrated Framework for Fine-Grained Analysis and Design of Group Learning Activities Seiji Isotani Riichiro Mizoguchi The Institute of Scientific and Industrial Research Osaka University, Japan

Agenda 1. The Problems: CSCL Design and Analysis 2. Ontological Approach 3. Background - Interaction Patterns (Inaba et al., 2003a) - Learner’s Growth Model (Inaba et al., 2003b) - Limitations 4. GMIP – Growth Model Improved by Interaction Patterns - Ontological Structure to Describe Learning Theories - Benefits - CHOCOLATO: Concrete Helpful Ontology-aware Collaborative Learning Authoring Tool 5. Conclusions

Group Teacher/Instructor How to design the best learning activities? Which Theory? What kind of benefits? How to analyze? learning by observation learning by guiding learning by apprenticeship Learner A Learner B Learner C LearnerSupported theoryLearning strategyRole for learnerProposed activitiesdesired Benefit Learner A Cognitive apprenticeship Learning by guidingMasterGuide learner B Skill development Learner B Cognitive apprenticeship Learning by apprenticeship ApprenticeImitate learner A Skill development Learner C Observational learning Learning by observation Observer Observe the interactions between A and B Knowledge acquisition Selection of an appropriate set of learning theories Cognitive Apprenticeship Theory Observational Learning Theory The Problems: CSCL Design and Analysis Each learning theory is “unfold” into a detailed pedagogical approach Apprentice role Master role Observer role

 Learning theories  hard to understand  too complex & ambiguous  There is not a common vocabulary to describe them  Different point of views, levels of aggregation, perspective and emphasis  How to “unfold” the theories into a set of activities for a group?  How to develop programs to help the design and analysis of group activities supported by learning theories? The Problems

learning theories Use ontological engineering to represent theories for CSCL Ontology Model Use the ontology model to support the development of ontology-aware systems users teacher/instructor/designer The systems help users to:  design group activities;  estimate benefits;  analyze, etc… Our Approach

Some of learning theories that support advantages of CL –Sociocultural theory [Vygotsky, 1930] –Observational learning [Bandura, 1971] –Peer tutoring [Endlsey,1980] –Cognitive Flexibility [Spiro et al., 1988] –Cognitive apprenticeship [Collins, 1991] –Legitimate Peripheral Participant [Lave & Wenger, 1991] –Anchored Instruction [CTGV, 1992] –Distributed cognition [Salomon, 1993] Background

Interaction Pattern [Inaba et al., 2003a] What is: patterns for expected interactions based on learning theories. Main Benefit: offer a model and vocabulary to describe learner-to-learner interaction process Usefulness: possibility to compare real interaction process with interaction patterns helping teacher to estimate educational benefits. Background: Previous Models Final Start Teaching the way to solve a problem Prompting Understanding Showing the way To solve a probem Prompting Acceptance Showing the way to solve a problem Showing problematic issues Request to show the way to solve problem Interaction expected to “master” role-player Interaction expected to “apprentice” role-player Necessary Transitions Desired Transitions Cognitive Apprenticeship

[Stages of Knowledge acquisition] nothing (0) tuning (2) accretion (1) Background: Previous Models Learner’s Growth Model [Inaba et al., 2003b] What is: graph representation (in a simplified way) of learner’s knowledge acquisition process (Rumelhart et al, 1978) and skill development process (Anderson, 82). Main Benefit: offer a way to represent the learner’s development (growth) and clarify the learning goals (stages). Usefulness: guideline to select learning strategies appropriate for desired learning goal. S(0,2) S(4,2) S(3,3) S(4,3) S(0,0) S(0,1) S(0,3) S(2,0) S(1,0) S(3,0) S(4,0) S(1,1) S(2,1) S(3,1) S(4,1) S(1,2) S(2,2) S(3,2) S(1,3) S(2,3) [Stages of Skill development] nothing (0) rough cognitive stage (1) explanatory cognitive stage (2) associative stage (3) autonomous stage (4) restructuring (3)

Limitations of the Previous Models 1.We cannot identify clearly the relationship among interactions and learner’s development. 2.We cannot explain the learner’s development through a set of interactions (events). 3.It is difficult to facilitate the collaboration (interaction) while a session are taking place. For example, in a CSCL session if a learner, who plays the role of anchored instructor, has a misunderstanding he will transfer it to the other learners (anchor holders) from the beginning until the end of the session.

Unifying the Models To overcome the difficulties addressed in the previous slide we aim to unify the previous models by: 1.Making tacit characteristics of learning theories explicit; 2.Identifying the relationships among interaction, learning strategies and learning goals; 3.Propose an ontological structure to describe learning theories for CL extending the Collaborative Learning Ontology (Inaba et al, 2000).

Collaborative Learning Ontology [inaba et al, 2000] LALA LCLC I-goal(L C ) I-goal(L B ) I-goal(L A ) W-goal({L A,L B }) W-goal({L A,L B,L C }) Y<=I -goal(L A <=L B ) Y<=I-goal (L B <=L A ) LBLB Learning Strategies Learning Goals Knowledge Acquisition: (accretion, tuning, …) Learning by Guiding Learning by Apprenticeship Cognitive Skill Development (cognitive, associative, …) Interaction Patterns HOW?

LALA I-goal(L B ) I-goal(L A ) Y<=I -goal(L A <=L B ) LBLB Interaction Influential I_L Events Instructor Event Learner Event Interaction Patterns for Learning Theories Anchored Instruction Peer Tutoring Distributed Cognition LPP Cognitive Constructivism Cognitive Flexibility Theory Sociocultural Theory Observational Learning Cognitive Apprenticeship Analyzing and Remodeling Interaction Patterns Interaction 3 Interaction 2 Interaction Interaction 4 Pattern Y<=I -goal(L B <=L A ) Learning Theory Instructor’s action Learner’s Action Necessary Interaction Desired Interaction Role

Ontological Structure to Describe Learning Theory Instructional event Influential I_L event I event I-goal Instructor Benefits for the Instructor Agent Action Instructional action L event Learner Agent Action Learning action Learning event I-goal Benefits for the Learner Interaction Pattern Y<=I-goal Learning Theory Teaching-Learning Process Learning Strategy I-goal I-role I-goal Agent You-role G Necessary Interaction Activity * * Influential I_L event Desired Interaction Activity Influential I_L event * *

Example: Cognitive Apprenticeship Instructor Event Learner Event

Cognitive Apprenticeship 3: Clarifying the problem* 4: Monitoring 5: Notifying how the learner is 9: Affirmative reaction 2: Demonstration how to solve a problem 8: Showing a solution 1:Setting up learning context 6: Instigating Thinking 7: Requesting problem's details  Interaction Pattern represented by Influential I_L Events  Learner’s development (Apprentice)

GMIP - Growth Model Improved by Interaction Patterns Cognitive Apprenticeship Learning by Apprenticeship x y Desired Interaction Necessary Interaction The dashed ellipses means that the interaction on the top/left must be followed by another interaction bottom/right. The ellipses means that the interaction on the top/left will be followed by another interaction bottom/right and vice-versa (cycle) [Stages of Skill development] nothing (0) rough cognitive stage (1) explanatory cognitive stage (2) associative stage (3) autonomous stage (4) [Stages of Knowledge acquisition] nothing (0) tuning (2) restructuring (3) accretion (1) [Interactions] 1.Setting up the learning context 2.Demonstrate how to solve a problem 3.Clarify the problem 4.Monitoring 5.Notifying how the learner is 6.Instigating thinking 7.Requesting problem’s details 8.Showing a solution 9.Affirmative reaction

Benefits of GMIP [Stages of Skill development] nothing (0) rough cognitive stage (1) explanatory cognitive stage (2) associative stage (3) autonomous stage (4) [Stages of Knowledge acquisition] nothing (0) tuning (2) restructuring (3) accretion (1) S(0,2) Cognitive Apprenticeship learning by apprenticeship Cognitive Apprenticeship learning by guiding Legitimate Peripheral Participant (LPP) Learning by Discussion

S(0,3) S(0,0) S(0,1) S(0,2) S(2,0) S(1,0) S(3,0) S(4,0) S(1,1) S(2,1) S(3,1) S(4,1) S(1,2) S(2,2) S(3,2) S(4,2) S(1,3) S(3,3) S(2,3) S(4,3) Benefits of GMIP Initial Stage Final Stage x y Desired Interaction Necessary Interaction The dashed ellipses means that the interaction on the top/left must be followed by another interaction bottom/right. The ellipses means that the interaction on the top/left will be followed by another interaction bottom/right and vice-versa (cycle). [Stages of Skill development] nothing (0) rough cognitive stage (1) explanatory cognitive stage (2) associative stage (3) autonomous stage (4) [Stages of Knowledge acquisition] nothing (0) tuning (2) restructuring (3) accretion (1)

Benefits of GMIP Anchored Instruction Learning by being Taught x y Desired Interaction Necessary Interaction The dashed ellipses means that the interaction on the top/left must be followed by another interaction bottom/right. The ellipses means that the interaction on the top/left will be followed by another interaction bottom/right and vice-versa (cycle). [Stages of Skill development] nothing (0) rough cognitive stage (1) explanatory cognitive stage (2) associative stage (3) autonomous stage (4) [Stages of Knowledge acquisition] nothing (0) tuning (2) restructuring (3) accretion (1) S(0,0) S(0,1) S(0,2) S(0,3) S(2,0) S(1,0) S(3,0) S(4,0) S(1,1) S(2,1) S(3,1) S(4,1) S(1,2) S(2,2) S(3,2) S(4,2) S(1,3) S(3,3) S(2,3) S(4,3) Initial Stage Final Stage Checking

Database Ontology Layer Author Learner’s Stage Identification System Learner’s Stage Identification System Theory-based Activities Designing Support System Theory-based Activities Designing Support System Learning Process Analysis System Learning Process Analysis System Theory-based Interaction Support System Learning Theory Ontology Learning Material Recommendation System Learning Material Recommendation System Learning Process Ontology Learning Material Authoring Interface CL ontology advice & recommendation selection, reuse & customization Learning Model CL Design Manager Ontology-aware Authoring System CHOCOLATO – Concrete Helpful Ontology-aware Collaborative Learning Authoring Tool MARI - Main Adaptive Representation Interface

CHOCOLATO – Concrete Helpful Ontology-aware Collaborative Learning Authoring Tool MARI Main Adaptive Representation Interface

CHOCOLATO – Concrete Helpful Ontology-aware Collaborative Learning Authoring Tool

Search for any goal stage Search for final goal stage CHOCOLATO – Concrete Helpful Ontology-aware Collaborative Learning Authoring Tool

 In our research we have been using ontologies to establish a common understanding of what a learning theory is by representing it in terms of its explicitness, formalism, concepts and vocabulary.  This makes theories understandable and sharable, both by computers and humans.  We use two previous models to clarify how interactions can affect learner’s development to propose another model, called GMIP - Growth Model Improved by Interaction Patterns.  Explicitly identify the relationships among interaction patterns, learning strategies and learning goals.  For users the GMIP allows the graphical visualization and use of learning theories. Thus, users can quickly interpret the theories, their benefits and can propose sequence of activities in compliance with them.  For computers, it provides a formal structure which allows systems to reasoning about the theories and the features (actions, roles, etc …) prescribed by them. Conclusions

 Support Group Interactions ( Learning Theory Ontology, Theory-based Interaction Support System, Theory- based Activities Designing Support System, Learning process Ontology)  Design Learning Environment (Learning Material Ontology, Learning Material Recommendation System, Learning Model)  Quality of Learning Process (Learning Model, Learner’s Stage Identification System, Learning Process Analysis System) Database Ontology Layer Author Learner’s Stage Identification System Learner’s Stage Identification System Theory-based Activities Designing Support System Theory-based Activities Designing Support System Learning Process Analysis System Learning Process Analysis System Theory-based Interaction Support System Learning Theory Ontology Learning Material Recommendation System Learning Material Recommendation System Learning Process Ontology Learning Material Authoring Interface CL ontology advice & recommendation selection, reuse & customization Learning Model CL Design Manager 16 Ontology-aware Authoring System  Our ultimate goal is develop a complete ontology-aware authoring system for CL based on well-grounded theoretical knowledge. Conclusions

Special Thanks Concluding Riichiro Mizoguchi Yusuke Hayashi Daniele Allard

Thank you! Concluding Seiji Isotani Riichiro Mizoguchi The Institute of Scientific and Industrial Research Osaka University, Japan