Problem-based learning supported by semantic
techniques
Esther Lozano, Jorge Gracia, Oscar Corcho
Ontology Engineering Group, Universidad Politécnica de Madrid. Spain{elozano,jgracia,ocorcho}@fi.upm.es
Outline
1. Introduction
2. System overview
3. Semantic grounding
4. Semantic-based feedback
5. Conclusions and Future Work
2
Introduction
3
“Engaging and informed tools for learning conceptual system knowledge”
Introduction
• Tries to capture human interpretation of reality
• Physical systems represented in models
• System behaviour studied by simulation
• Focused on qualitative variables rather than on numerical ones (eg., certain tree has a “big” size, certain species population “grows”, etc.)
4
Qualitative Reasoning
Introduction
• Core idea: “Learning by modelling”• Learning tools:
• Definition of a suitable terminology• Interaction with the model• Prediction of its behaviour
• Application examples:• “Study the evolution of a species
population when another species is introduced in the same ecosystem”
• “Study the effect of contaminant agents in a river”
• ....
5
Application: Learning of Environmental Sciences
Introduction
• “System for knowledge acquisition of conceptual knowledge in the context of environmental science”. It combines:• Model construction representing a system• Semantic techniques to put such models in relationship• Use of virtual characters to interact with the system
6
DynaLearn
Introduction
7
DynaLearn
QR Modelling
8
Entities
QR Modelling
9
Model fragments
Entity: The structure of the systemImported model fragment: Reuse within a model
Influence: Natality determines δSize
Quantity: The dynamic aspects of the system
Proportionality: δSize determines δNatality
QR Modelling
10
Running simulations
QR Modelling
• Based on a scenario, model fragments and model ingredient definitions
11
Simulations Results
Dependencies View of State 1 Value History
State Graph
Semantic Techniques
• To bridge the gap between the loosely and imprecise terminology used by a learner and the well-defined semantics of an ontology
• To put in relation to the QR models created by other learners or experts in order to automate the acquisition of feedback and recommendations from others
12
Semantic Techniques
Outline
1. Introduction
2. System overview
3. Semantic grounding
4. Semantic-based feedback
5. Conclusions and Future Work
13
System overview
14
List of suggestions
Grounding of learner model
Recommendation of relevant models
Generation of semantic feedback
Semantic repositoryOnline semantic resources
Learner Model
Grounded Learner Model
Reference Model
Learner
?
Outline
1. Introduction
2. System overview
3. Semantic grounding
4. Semantic-based feedback
5. Conclusions and Future Work
15
Semantic Grounding
16
http://dbpedia.org/resource/Mortality_ratehttp://dbpedia.org/resource/Population
http://www.anchorTerm.owl#NumberOf
Expert/teacher Learner
grounding
Semantic repository
Anchor ontology
Semantic Grounding
• Support the process of learning a domain vocabulary• Ensure lexical and semantic correctness of terms• Ensure the interoperability among models• Extraction of a common domain knowledge• Detection of inconsistencies and contradictions between
models• Inference of new, non declared, knowledge• Assist the model construction with feedback and
recommendations
17
Benefits of grounding
Semantic Grounding
18
Outline
1. Introduction
2. System overview
3. Semantic grounding
4. Semantic-based feedback
5. Conclusions and Future Work
19
Semantic-based feedback
List of suggestions
Ontology matching
Learner Model Grounding-based
alignment
Reference Model
Preliminary mappings
List of equivalences
Generation of semantic feedback
Terminology Discrepancies
Taxonomy Inconsistencies
QR structures Discrepancies
Grounding-based alignment
http://dbpedia.org/resource/Mortality_rate
Expert modelStudent model
grounding
Semantic repository
Preliminary mapping: Death_rate ≡ Death
Grounding-Based Alignment
• In the learner model:
• In the reference model:
• Resulting preliminary mapping:
22
Ontology Matching
• Ontology matching tool: CIDER
• Input of the ontology matching tool
• Learner model with preliminary mappings
• Reference model
• Output: set of mappings (Alignment API format)
23
Gracia, J. Integration and Disambiguation Techniqies for Semantic Heterogeneity Reduction on the Web. 2009
Terminology discrepancies
24
Discrepancies between labels
Learner model: Reference model:
equivalent terms with different label
Terminology discrepancies
25
Missing and extra ontological elements
Reference model:
missing term
Learner model:
equivalent termsextra term
subclass of
Taxonomic discrepancies
26
Inconsistency between hierarchies
Reference model:
Learner model:
equivalent terms
Disjoint classes
INCONSISTENT HIERARCHIES!
QR structural discrepancies
27OEG Oct 2010
Algorithm:
1. Extraction of basic units
2. Integration of basic units of the same type
3. Comparison of equivalent integrated basic units
4. Matching of basic units of the same type
5. Comparison of equivalent basic units
QR structural discrepancies
28OEG Oct 2010
External relationships
Internal relationships
Extraction of basic units
QR structural discrepancies
29OEG Oct 2010
Algorithm:
1. Extraction of basic units2. Integration of basic units of the same type
3. Comparison of equivalent integrated basic units
4. Matching of basic units of the same type
5. Comparison of equivalent basic units
QR structural discrepancies
30OEG Oct 2010
Integration of basic units by type
QR structural discrepancies
31OEG Oct 2010
Algorithm:
1. Extraction of basic units
2. Integration of basic units of the same type3. Comparison of equivalent integrated basic units
1. Missing instances in the learner model
2. Discrepancies in the internal relationships
4. Matching of basic units of the same type
5. Comparison of equivalent basic units
QR structural discrepancies
32OEG Oct 2010
Missing instances in the learner model
Missing quantity
Reference modelLearner model
QR structural discrepancies
33OEG Oct 2010
Discrepancies between internal relationships
Different causal dependency
Reference model Learner model
QR structural discrepancies
34OEG Oct 2010
Algorithm:
1. Extraction of basic units
2. Integration of basic units of the same type
3. Comparison of equivalent integrated basic units4. Matching of basic units
• Filter by MF (matching of MF first)• Matching based on the external relations
5. Comparison of equivalent basic units
QR structural discrepancies
35OEG Oct 2010
Matching of basic units
equivalent
equivalent
Reference model
Learner model
QR structural discrepancies
36OEG Oct 2010
Algorithm:
1. Extraction of basic units
2. Integration of basic units of the same type
3. Comparison of equivalent integrated basic units
4. Matching of basic units of the same type5. Comparison of equivalent basic units
1. Missing entity instances
2. Discrepancies in external relationships
QR structural discrepancies
37OEG Oct 2010
Missing entity instances
Missing entity instances
Reference model
Learner model
QR structural discrepancies
38OEG Oct 2010
Discrepancies in the internal relationships
Reference model
Learner model
Different causal dependencies
Feedback from the pool of models
39OEG Oct 2010
Algorithm:
1. Get semantic-based feedback from each model
2. For each generated suggestion, calculate agreement among models
3. Filter information with agreement < minimum agreement
4. Communicate information to the learner
Feedback from the pool of models
40OEG Oct 2010
Learner model
Example:
Feedback from the pool of models
41OEG Oct 2010
Example:
25%
75%
67%
67%
Interface
42OEG Oct 2010
Problem-based learning supported by semantic
techniques
Esther Lozano, Jorge Gracia, Oscar Corcho
Ontology Engineering Group, Universidad Politécnica de Madrid. Spain{elozano,jgracia,ocorcho}@fi.upm.es