Upload
pilis
View
36
Download
0
Embed Size (px)
DESCRIPTION
Using Schema Analysis for Feedback in Authoring Tools for Learning Environments. Harrie Passier* & Johan Jeuring** Faculty of Informatics * Open University of the Netherlands ** Open University of the Netherlands and University of Utrecht. Overview. Introduction Context Feedback - PowerPoint PPT Presentation
Citation preview
Using Schema Analysis for Feedback in Authoring Tools for Learning Environments
Harrie Passier* & Johan Jeuring**Faculty of Informatics
* Open University of the Netherlands** Open University of the Netherlands and University of Utrecht
AIED 2005 Harrie Passier, OUNL 2
Overview• Introduction
– Context– Feedback– Lack of feedback – Research goal – Ontology based feedback
• Using Schema Analysis for Feedback in Authoring Tools– Schemata– Schema representations– Schemata: abstract interpretations– Schema analysis– Two examples: completeness and synonyms
• Questions and discussion
AIED 2005 Harrie Passier, OUNL 3
Context
• Faculty of Informatics of the OUNL
• Research interest: Generating Feedback– Feedback to students
• Design education like modelling (UML – class and object diagrams)
• Mathematics courses (solving systems of linear equation)
– Feedback to authors• Course development
• Information from student phase to author phase: optimisation of e-course (sub project of Alfanet project –Audit module)
AIED 2005 Harrie Passier, OUNL 4
Feedback
• Definition– Comparison of actual performance with some set standard (norm)
– Assess progress, correct errors and improve performance
• An essential element needed for effective learning
AIED 2005 Harrie Passier, OUNL 5
Lack of feedback
• Student side: there is a frequently lack of (semantically rich) feedback in eLearning systems (Mory, 2003)
• Author side: eLearning systems are often complex tools. There is a high probability of mistakes. To improve the quality, authoring tools should include mechanisms for checking the authored information (Murray, 1999)
AIED 2005 Harrie Passier, OUNL 6
Research goal
• Develop generic, domain and task independent feedback mechanisms that produce semantically rich feedback to learners and authors
• Three types of feedback– To a student during learning
– To an author during course authoring
– From a group of learners who study a course to an author
• Ontologies are arguments of the general feedback engine– Reusability, flexibility and adaptability of knowledge structures
(Aroyo, 2004)
AIED 2005 Harrie Passier, OUNL 7
Ontology based feedbackFunctional architecture
Domain ontology
Model language ontology
Task ontology
Education ontology
Feedback ontology
eLearning system
Player
Author tool
Feed
back
en
gine
AIED 2005 Harrie Passier, OUNL 8
Ontologies as norms
Examples
• Author perspective:– Domain ontology (communication technology)– Course structure (IMS Learning Design – IMS LD) – Task ontology (steps to develop a course)– Education (inductive and deductive learning)– Feedback (preventive and corrective feedback)– …
• Student perspective:– Domain ontology (communication technology)– Model language ontology (UML)– ..
AIED 2005 Harrie Passier, OUNL 9
Using Schema Analysis for Feedback in Authoring Tools
Scope:
• Authoring
• Structural aspects– Course structure
– Domain structure
AIED 2005 Harrie Passier, OUNL 10
Schema
• An ontology specifies the objects in a domain of interest together with their characteristics in terms of attributes, roles and relations. Many aspects can be represented, such as categories (taxonomic hierarchy), time, events and composition.
• A schema is a certain type of ontology. It describes the structure of a composite object. A composite object contains objects related to other objects using ‘has_part’ or ‘uses’ relations.
AIED 2005 Harrie Passier, OUNL 11
Schema representations
Two schemata:
• Domain schema: RDF <resource, property, value>
• Course structure: IMS LD (= Document Type Defintion -DTD)– Addition of specific annotations to content and structure:
• New elements: Definition and Example• New attribute: Educational-strategy (Inductive | Deductive) • In practice many elements can be added
wheel
rimspoke
has_part
AIED 2005 Harrie Passier, OUNL 12
Example IMS LD definition
<!ELEMENT Activity %Activity-model; ><!ATTLIST Activity
… Educational-strategy (Inductive | Deductive) ><!ENTITY %Activity-model "(Metadata?, …, Activity-description)" ><!ELEMENT Activity-description (Introduction?, What, How?, …, Feedback-description?) ><!ELEMENT What %Extra-p; ><!ENTITY %Extra-p "(…| Figure | Audio | Emphasis | List | … | Example | Definition)*" >
<!ELEMENT Definition (Description, Concept, RelatedConcept+) ><!ATTLIST Definition Id ID #REQUIRED Name CDATA #REQUIRED><!ELEMENT Example (Description, Concept, RelatedConcept+) ><!ATTLIST Example Id ID #REQUIRED Name CDATA #REQUIRED Belongs-to-definition IDREFS #REQUIRED>
AIED 2005 Harrie Passier, OUNL 13
Schemata: abstract interpretations
Possible properties of a course:
• Completeness: Are all concepts that are used in the course defined somewhere?
• Correctness: Does the definition of a concept used in the course correspond to the definition of the concept in the ontology?
• Timely: Are all concepts used in a course defined on time?– Use of educational strategy attribute (inductive, deductive)
• Recursive concepts: Are there concepts defined in terms of it self?
• Synonyms: Are there concepts with different names but exactly the same definition?
• Homonyms: Are there concepts with multiple, different definitions?
AIED 2005 Harrie Passier, OUNL 14
Schema analysis
• The analyses take schemata as input
• We perform two types of analyses– The analysis of structural properties of one schema, for example
the recursive property
– The comparison of a schema with one or more other schemata, for example to test on correctness
AIED 2005 Harrie Passier, OUNL 15
Some definitions (I)
Suppose o = Ont [(a, [b,c]), (b, []), (c, [d,e]), (d, []), (e, [])] :: Ontology, where the letters represent concepts
a
b c
d e
AIED 2005 Harrie Passier, OUNL 16
Some definitions II
Then
• terminalConcepts = [(b, []), (d, []), (e, [])]
• nonTerminalConcepts = [(a, [b,c]), (c, [d,e])]
• allConcepts = [(a, [b,c]), (b, []), (c, [d,e]), (d, []), (e, [])]
• reachable nonTerminalConcepts allConcepts
= [(a, [b,c,d,e]), (b, []), (c, [d, e]), (d, []), (e, [])]
• reachableTerminals nonTerminalConcepts nonTerminalConcepts
= [(a, [b,d,e]), (c, [d,e])]
NB. Functions based on fixpoint calculations (grammar analyses)
a
b c
d e
AIED 2005 Harrie Passier, OUNL 17
Example I: Completeness
• Definition: are all concepts used in the course defined somewhere?– Within a course – Within an domain ontology– Between a course and an domain ontology
• Steps (within a course)– Determine the set of used concept id’s
• in the right- and left hand sides of concepts within examples• in the right hand side of concepts within definitions
– Determine the set of defined concept id’s• in the left-hand side of concepts in definitions
– Check that each of the used concepts appears in the set of defined concepts
AIED 2005 Harrie Passier, OUNL 18
Example II: Synonyms
• Concepts with different names may have exactly the same definition– Within an ontology
• Example– Concept a (a, [c,d]) and concept b (b, [c,d]), are synonyms
• Formal definition: Given a set of productions, two concepts x and y are synomyms if their identifiers are different, Idx Idy, and
(reachableTerminals productions x) equals (reachableTerminals productions y)
• Steps– Determine for all concepts in the ontology all reachable terminal concepts
– Collect the concepts with the same reachable terminal concepts and different concept id’s