Information-Flow-based Ontology Mapping

1
Information-Flow-based Ontology Mapping

• Yannis Kalfoglou, University of Southampton
• Marco Schorlemmer, University of Edinburgh

2
Semantic Web

• Need the ability to automatically
• MAP ontologies in the same domain
• MAP ontologies in different domains that overlap
• MERGE ontologies once they are mapped

3
MAPPING BARRIERS

• No public technical details available
• Present systems are so diverse and
  un-interoperable
• Mapping systems embedded in these diverse systems
• Previous mapping based on syntactical clues
  instead of semantics
• Lack theory, axioms, and rules

4
PURPOSE

• Build mathematics and an algorithm to
  semi-automatically MAP ontologies in RDF

5
OUTLINE

• Architectural overview
• Mathematical definitions and proposition
• IF-map example
• Basic mapping algorithm
• Final results
• Summary
• My Assessment

6
Architectural Overview
7
IF-map (Information-Flow-Based)
8
Mathematical Definitions and Proposition

• DEFINITIONS
• Local Logic
• Logic Infomorphism
• Ontology
• Populated Ontology
• Ontology Morphism
• PROPOSITION - Deriving an Ontology Morphism
  relationship from a REFERENCE (unpopulated
  ontology) to a LOCAL (populated ontology)

9
LOCAL LOGICTYPE vehicle carINSTANCE Cabrio
Coupe
10
SYMBOLS

• T- type, I-instance
• ? - set of types - disjunctively
• G - set of types - conjunctively
• Set of Sequents (G,?)
• every type, some type

11
LOGIC INFOMORPHISM (f)links 2 local logics (L,
L)f(Coupe) carf(Cabrio) vehicle
12
ONTOLOGY (O)
13
Populated Ontology (?)(Local ontology is
populated with instances x, classifications c
d, and relations r)
14
ONTOLOGY MORPHISMmapping the changeUnpopulated
to Populated
15
MAPPING Architecture
16
ONTOLOGY MORPHISMREFERENCE ONTOLOGY (unpop)to
LOCAL ONTOLOGY (pop)
17
IF-Map ExampleREFERENCE ONTOLOGYunpopulated
18
LOCAL ONTOLOGIES
19
CONSTRAINTS to LOGIC INFORMORPHISMS
20
Relate and Map
21
Relationships between typesREFERENCE - LOCAL
22
Completes ONE MAPPINGf(car) automobile(a)
arity (b) Infomorphism (c) complete
23
Basic Mapping Algorithm
24
Algorithm in Example

• Match types (MANUALLY?)
• REFERENCE - building
• LOCAL - house, cottage
• Do argument types match semantically?
• If YES partial map
• If NO use ISA hierarchy and check parents and
  children and list relations which might be merged

25
FINAL RESULTS (RDF)
26
SUMMARY

• Formalized definitions, mapping, linking
• Ontology
• Ontology morphism
• Logic Infomorphism
• FUTURE
• Ontology Merging
• Ontology Axioms
• Reasoning about Ontology Evolution

27
My Assessment

• Example semantic problem between British
  English and American English (cottage and house)
• Purpose of the paper changed from the beginning
  to the end WEAK EVALUATION
• Study of 5 CS Dept in the UK
• Mathematics and algorithm
• Hard to tell where Channel Theory left off and
  their Information Flow Theory began
• RDF Missing technical details of change during
  the IF-map