SEMANTIC WEB TECHNOLOGIES FOR PERSONALIZED LEARNING AND COLLABORATIVE

SEMANTIC WEB TECHNOLOGIES FOR PERSONALIZED LEARNING AND COLLABORATIVE TEACHING Apple W P Fok Centre for Innovative Applications of Internet and Multimedia Technologies (AIMtech) Image Computing Group, Department of Computer Science City University of Hong Kong

Outline • Goal & Motivation: Personalized Education (PE) • Conceptual Framework of Personalized Education System (PES) • PES Realization: Personalized Agents Team (PEAs) • Personalized Education Ontology (PEOnto): An Integration of multiple ontologies for PES • Application of PEOnto: Personalized Instruction Planner (PIP) • WELNET: A Collaborative Blended Learning Community for Personalized Learning and Collaborative Teaching

Education Reform and IT in Education

Personalized Education Personalization in e-commerce: – capture & retain customers’ loyalty – Building a meaningful one-to-one relationship. – Riecken D. – Delivering appropriate content and services to fulfill user’s needs. – Monica Bonett – Understanding where and when to recommend the “right” things. – Oracle Personalization - Cater to individual learning differences (ability & needs) - Machine learning and updating of student profiles - Intelligent educational content search & filtering - Automatic individualized study plan generation PES framework [Fok & Ip 2004]

Personalization in Education A supportive learning platform should: • Monitor and manage individual student profile • Provide a common structure for educational content annotation & indexing • Search and recommend materials relevant to individual learning needs • Intelligently sequence learning materials to meet individual learning objectives • Support education research through collecting and analyzing usage data of students and teachers (e. g. data-mining) • Adapt to student’s needs through analysis of learning progress (eg. adaptive educational hypermedia)

Personalization in Education [Fok & Ip 2004]

The Framework of the PES Fok & Ip, 2004

Architecture of PES • Built upon Tsinghua University “Smart Platform” – – Asynchronous communication Support Publish-and-subscribe model Loosely-coupled Parallel Execution Fok & Ip, 2005

Run time structure of PES Dual-citizenship web server!

PE Agents’ Design Fok & Ip, 2005

Emerging Technologies for Educational Resources Indexing & Re-use • IEEE Learning Object Metadata: An Ontological Representation - Conlan, O. , Hockemeyer, C. , Lefrere, P. , Wadde, V. , Albert, A. , 2001, Extending Educational Metadata Schemas to describe Adaptive Learning Resources, ACM ISBN 1 -59113420 -7/01/0008 - Qin, J. & N. Hernandez. (2004). Ontological representation of learning objects: building interoperable vocabulary and structures. WWW 2004, May 17 -22, 2004, New York, 348 -349. New York: ACM Press. - Recker, M. M. , Wiley, D. A. , 2000, A non-authoritative educational metadata ontology for filtering and recommending learning objects - Scime, A. , and Kerschberg, L. , 2000, Web. Sifter: An Ontology -based Personalizable Search Agent for the Web, International Conference on Digital Libraries: Research and Practice, Kyoto Japan, 2000 - Kerschberg, L. , Kim, W. , and Scime, A. , 2000, Web. Sifter II: A Personalizable Meta-Search Agent based on Semantic Weighted Taxonomy Tree

Educational Ontology Semantic Web Technologies for describing content that are readable and can be processed by machine (eg. software search agent) Extending Semantic Web to the Educational community: • Emerging standards for defining learning contents: • describing “structure” of learning objects [LOM] • describing “packaging, sequencing and presenting” reusable learning objects [SCORM] • Mechanism to relate different educational concepts to facilitate search of learning objects [Educational Ontology, OWL]

Semantic Metadata

Personalized Education Ontology Fok & Ip, 2006 (PEOnto) • An Educational Ontology • A fundamental component of PE • The development of a semantic web for educational resources • Facilitate personal epistemology in discovering, selecting, organizing and using relevant educational resources. • Incorporate FIVE interrelated educational ontologies – – – People Ontology Language Ontology Curriculum Ontology Pedagogy Ontology PEA Ontology

The Roles of PEOnto • Strengthen agents communication and performances – Ontological commitments – Automatic messages/parameters generations • Understand LO in a semantic way – Relevant for a particular task/activity – Fulfill a particular learning objective type – Sequence in relation to different LOs • Understand Discover implicit information for further analyze – The relations between the instructional design (LO) and students’ learning – Different learning paths for different students’ learning needs (i. e. Cognitive, Skills or Affective Domain development) – Different teaching/learning styles and learning patterns

PEOnto Components 2005 Fok & Ip, ICCE

PEOnto – cont. • People Ontology (Pe. Onto) – The structure of school education, people, schools and the activities perform between them – Construct the User Profiles based on the IMS Learner Information Package Specification and further extended the taxonomy for in-depth classification and mining purposes

Profile Structure and Its Related Information

Ontology-driven Profile Construction

PEOnto – cont. • Curriculum Ontology (Cur. Onto) – The structure of a curriculum design and its essential components and attributes – Represents the goal state of a user, a searching query, or classification of learning resources

Curriculum Ontology

PEOnto – cont. • Language Ontology (Lang. Onto) – The structure of a subject domain – Classify educational resources into different language learning items – Discover the relations between knowledge, skills and levels

Language Ontology (ESL)

Instances of Language Ontology

English Learning Objective Hierarchy

PEOnto – cont. • Pedagogy Ontology (Peda. Onto) – Describes the pedagogical approaches, instructional design procedures and the relations between educational resources and instructional events/activities. • Pedagogy Ontology • Instruction Ontology • Content Ontology – Helps to identify the usability of various resources and discover teaching/learning preferences/styles.

Peda. Onto Inner Ontologies Figure 6. 20

Pedagogy Ontology

Peda. Onto Overview

The Instructional Conditions, Instructional Methods and Instructional Outcomes of the Instruction Ontology. Figure 6. 29

Marco and Micro Views Figure 6. 30 Figure 6. 31

PEOnto Relations

Objective Links between different Ontologies Figure 6. 18

Objectives Hierarchy Figure 6. 17

Objective Classes

Verbs of Competencies

Material Information

PEOnto – cont. • PE Agents Ontology (PEAOnto) – Governs PEAs behaviors/duties – Describes the responsibilities of each PE agent and indicates the relations and communication path among the PEA team

PEAs Ontological Commitments

Application of PEOnto • Producing digitalized educational resources • Incorporating learning resources with appropriate pedagogies • Modifying, reusing, or improving existing educational resources effectively • Storing, retrieving and sharing educational resources as well as teaching experiences efficiently

Personalized Instruction Planner (PIP) Fok and Ip, ICME 2006 Personalized Instruction Planner Searching Tool Selecting Tool Organizing Tool Personalized Education Agents (PEAs) Crawling Agent Classification Agent Searching Agent Personalized Education Ontology (PEOnto) Curriculum Ontology Pedagogy Ontology People Ontology Schema Databases Personal/Content Profiles PIP Learning Objects PEOnto Schema and Metadata

Key Tasks of PIP • Personalization Search – Retrieve personalized search results in respect to the user profiles • Personalized Instruction Planning – Organize and structure instruction plan according to schoolbased curriculum or teaching preferences – Record all instruction designs and identify various uses of education resources. • Generating PE LOM resources – Incorporate educational vocabulary items (i. e. PEOnto) to label and annotate PE resources as LOM for improved interoperability and reusability

Ontology-driven Architecture for PIP

Steps of Materials Selections • Objective Statements; • Objective Classification; • Selection of instructional events; • Determining type of stimuli for each event; • Listing the candidate resources for each event; • Listing theoretically best resources for the events; • Recording final resources choices; • Generating a rationale for the decisions made and • Generating a prescription for each material in each event.

Personalized Instruction Planner

Instruction Plan Design

Personalized Instruction Planner

PIP – Global Search

PE Search Workflow Retrieve relevant educational resources from the Web-crawling Agent Internet 1 Databases 2 Classification Agent Filter and classify retrieved resources with respect to education goals, learning objectives, and instruction design principles 3 Education Ontology (PEOnto) Personalized Search Agent Response queries and collect feedbacks (i. e. usage results)

PIP – Global Search

PIP – Local Search

Customized Search • E. g. The message path of customized search request and response

PIP – Local Search

Personalized Instruction Planner

PES Performance Simulation • Stub Implementation – Run the service of planning, searching & filtering simultaneously – Assume each service per time costs 10 ms

The Past • A Conceptual framework for Personalized Education • The design and development of the PES • Central to the PE Framework is PEOnto - an integration of FIVE inter-related ontologies • PEOnto demonstrates the necessary attributes required in Personalized Education services delivery • Applied PEOnto in the development of PIP for English Second Language (ESL) Learning • PIP provides a testbed not only for evaluating the feasibility of PES, but more importantly, experiencing different mechanisms and strategies to realize our vision in education – Personalized Education

Present Authoring and Delivering Sharable, Reusable, Pedagogically Sound Education Resources • Further exploit PIP potentials in WELS – Better response time, higher automation, multiple subjects, Chinese encoding, better interface designs and so on… – Further adjust to fulfill more instructional design needs • Try out different approaches and develop/explore new Epedagogy approaches/models

Future Work The Personalized Education System and its PEAs • A user-friendly interface for teachers to annotate and deliver educational resources • Personalized Education Features • More subject domain ontologies • A localized intelligent education search engine • Experience and compare different agent design and algorithms so as to provide personalized e-learning experience to support teaching & learning through PES – Profiling and Mining – Task Performance Support

http: //wels. welnet. hk 71

Email: applefok@cs. cityu. edu. hk