combining data mining and ontology engineering to enrich ontologies and linked data

15
Combining Data Mining and Ontology Engineering to enrich Ontologies and Linked Data Mathieu d’Aquin Knowledge Media Institute (Kmi), The Open University, UK (@mdaquin) Gabriel Kronberger University of Applied Science Upper Austria, School for Informatics, Communications and Media Mari Carmen Suárez-Figueroa Ontology Engineering Group, Departamento de Inteligencia Artificial, Facultad de Informática, Universidad Politécnica de Madrid

Upload: mathieu-daquin

Post on 20-Jan-2015

1.665 views

Category:

Technology


3 download

DESCRIPTION

Presentation at the "first international workshop on Knowledge Discovery and Data Mining Meets Linked Open Data" (Know@LOD) at ESWC 2012

TRANSCRIPT

Page 1: Combining Data Mining and Ontology Engineering to enrich Ontologies and Linked Data

Combining Data Mining and Ontology Engineering to enrich Ontologies and

Linked DataMathieu d’Aquin

Knowledge Media Institute (Kmi), The Open University, UK (@mdaquin)

Gabriel KronbergerUniversity of Applied Science Upper Austria, School for Informatics, Communications and Media

Mari Carmen Suárez-FigueroaOntology Engineering Group, Departamento de Inteligencia Artificial, Facultad de Informática,

Universidad Politécnica de Madrid

Page 2: Combining Data Mining and Ontology Engineering to enrich Ontologies and Linked Data

The Knowledge Discovery Process

Page 3: Combining Data Mining and Ontology Engineering to enrich Ontologies and Linked Data

The Knowledge Discovery Process

With Linked Data

and Ontologies…

?

Ontologies?populatedBy/modelling/characterising/structuring?

Ontology Patterns?

??

??

??

Page 4: Combining Data Mining and Ontology Engineering to enrich Ontologies and Linked Data

The Knowledge Discovery Process

Linked

?

Page 5: Combining Data Mining and Ontology Engineering to enrich Ontologies and Linked Data

The Knowledge Discovery Process

Linked

?

Page 6: Combining Data Mining and Ontology Engineering to enrich Ontologies and Linked Data

The Knowledge Discovery ProcessGuided by

knowledge

?

Page 7: Combining Data Mining and Ontology Engineering to enrich Ontologies and Linked Data

The Ontology Engineering Process

Ellicitate knowledge

Model knowledge

Represent knowledge

Traditionallycompetency questions, key concepts, etc.

diagrams, etc.

OWL, RDFS, etc.

Ellicitate domain

In Linked Data

Reuse from others

Combine

through existing information systems, etc

find commonly used vocabularies

align, fill the gaps, etc.

In both cases, it is expected that the data will somehow fit the ontology, that the ontology will support relevant applications, and support the inference of new information

Page 8: Combining Data Mining and Ontology Engineering to enrich Ontologies and Linked Data

DataData

Knowledge Engineering and Knowledge Discovery: a co-evolution process?

Represent knowledge/Combine

Model knowledge/Reuse

Ellicitate knowledge/domain

Pre-process

Mine

Interpret

Data

Ontologies/Knowledge

Page 9: Combining Data Mining and Ontology Engineering to enrich Ontologies and Linked Data

Knowledge Engineering and Knowledge Discovery: a co-evolution process?

Page 10: Combining Data Mining and Ontology Engineering to enrich Ontologies and Linked Data

Major (new) issues 1/4Ontology-based filtering, checking and interpretation of DM results

Mine

DataDataData

Ontologies

Results ??

Zablith et al., Using Ontological Contexts to Assess the Relevance of Statements in Ontology Evolution, EKAW 2010

Ontology

DocsText

Analysis

Relation Discovery New concepts

New relations

Page 11: Combining Data Mining and Ontology Engineering to enrich Ontologies and Linked Data

Major (new) issues 2/4

Mining from Linked and Ontology based data

Mine

DataData

Data

Ontologies

Results??

Ontology

Ontology

Nikolov et al., Unsupervised Learning of Link Discovery Configuration, ESWC 2012

Data DataGenetic

Algorithm

Similarity Configuration

Link Discovery

Links

Page 12: Combining Data Mining and Ontology Engineering to enrich Ontologies and Linked Data

Major (new) issues 3/4

Ontology-guided data mining

Mine

Data

Ontologies

Results??

d’Aquin and Motta, Extracting Relevant Questions to an RDF Dataset Using Formal Concept Analysis, K-CAP 2011

Ontology

RDF Data

Inference + Formal Context

Generation

Formal Context

Formal Concept Analysis

Lattice

Interpretation

Prominent questions/qu

eries

Page 13: Combining Data Mining and Ontology Engineering to enrich Ontologies and Linked Data

Major (new) issues 4/4Versioning and consistency

Mine

DataDataData

Ontologies

Results

MineMine

OntologiesOntologies

Results

Results

??

Requires keeping track of the different models and their versions, the agreement and disagreement between them, as well as the areas of consensus and controveries(d’Aquin, Formally Measuring Agreement and Disagreement in Ontologies, K-CAP 2009)

Lead to the notion of ontology convergence

Page 14: Combining Data Mining and Ontology Engineering to enrich Ontologies and Linked Data

Conclusion

• Many existing works have considered the connection between data mining and ontology engineeing

• A large scale, web of linked data and ontologies make the related challenges more prominent…

• … and need real interactions between the two approaches, not as disconnected components.

• Need to investigate and exploit the colateral benefits of ontology engineering and knowledge discovery…

• … coming up with new techniques for enriching knowledge from mined data, and guiding the extraction of further data wit ontological knowledge

Page 15: Combining Data Mining and Ontology Engineering to enrich Ontologies and Linked Data

Thank you

[email protected]

http://people.kmi.open.ac.uk/mathieu

@mdaquin