automated ontology matching in the architecture ... · automated ontology matching in aec/fm domain...
TRANSCRIPT
© G.F. Schneider
Automated Ontology Matching in the Architecture, Engineering and Construction Domain - A Case Study
Georg Ferdinand Schneider
Technische Hochschule Nürnberg, Nürnberg, Germany
Fraunhofer Institute for Building Physics IBP, Nürnberg, Germany
7th Linked Data in Architecure and Construction Workshop (LDAC)
19 – 21 June 2019 | Lisbon, Portugal
© G.F. Schneider
Agenda
• Introduction and Motivation
• Methodology for Case Study
• Manually Defined/ Revised Alignments
• Results from Automated Ontology Matching
• Conclusion and Future Work
19/06/2019 2
© G.F. Schneider
Agenda
• Introduction and Motivation
• Methodology for Case Study
• Manually Defined/ Revised Alignments
• Results from Automated Ontology Matching
• Conclusion and Future Work
19/06/2019 3
© G.F. Schneider
Introduction and MotivationHeterogeneity of the domain
19/06/2019 4
Geometry
Process/ Project
Elements & Equipment
MaterialBuilding
Automation
Conservation
Product Data/ eCommerce
…
• Model-based information exchange in the built environment: Multi-* challenge
– Multiple domains
– Multiple stakeholders
– Multiple information silos
– Multiple tools and formats
Linked Data & Semantic Web Technologies to address these challenges1G. F. Schneider, A. Bougain, P. S. Noisten, and M. Mitterhofer. Information Requirement Definition for BIM: A Life Cycle Perspective. In Proceedings of the 11th European Conference on Product and Process Modelling (ECPPM), Limassol, Cyprus, September 7-9 2016.
Adapted from: 1
© G.F. Schneider
Introduction and MotivationPlethora of domain ontologies
19/06/2019 5
• Proliferating development of domain ontologies in AEC/FM domain
• Ontology reuse as stipulated by ontology engineering methods often neglected
• 'Thus, merely using ontologies, like using XML, does not reduce heterogeneity: it raises heterogeneity problems to a higher level.‘1
• Defining manually alignments is almost as difficult as creating ontologies from scratch
Strong need to use automated ontology
matching methods
Figure source: Rasmussen, M. H., Pauwels, P., Hviid, C. A., & Karlshøj, J. (2017). Proposing a central AEC ontology that allows for domain specific extensions. In 2017 Lean and Computing in Construction Congress.
1Euzenat, J., & Shvaiko, P. (2007). Ontology matching (Vol. 18). Heidelberg: Springer.
BOT
ThinkHome
BRICK
BRICK DogOnt
IFC
EEPSA
SimModel gbXML
SAREF4Bldg
© G.F. Schneider
Introduction and MotivationOntology Matching, Correspondences and Alignments
19/06/2019 6
‚Ontology matching is the process of finding relations between the entities of different ontologies, e.g., identifying the entities that represent the same (or similar) semantics in these ontologies.‘1
‘Correspondence is the expression of a relation holding between entities (classes,
properties, or individuals) of different ontologies.’ 1
‘Alignment is a set of correspondences between two ontologies. The alignment is the output of the ontology matching process.’1
1Kovalenko, O., & Euzenat, J. (2016). Semantic matching of engineering data structures. In Semantic web technologies for intelligent engineering applications (pp. 137-157). Springer, Cham.
© G.F. Schneider
Introduction and MotivationAutomated Ontology Matching in AEC/FM Domain
19/06/2019 7
• Demand for automated matching methods to facilitateinteroperability
• Knowledge-based data integration in smart city domain, Automated method ‚OntoPhil‘
• Reoccuring ontology design patterns identified, smart citydomain
No dedicated study to AEC/FM domain on automatedontology matching
Costin, A. & Eastman, C.
Otero-Cerdeira et al.
Bellini et al.
Gyrard et al.
Espinoza-Arias et al.
© G.F. Schneider
Agenda
• Introduction and Motivation
• Methodology for Case Study
• Manually Defined/ Revised Alignments
• Results from Automated Ontology Matching
• Conclusion and Future Work
19/06/2019 8
© G.F. Schneider
Methodology
19/06/2019 9
Step 1
• Manual definition/ revision of class and object property levelalignments
• Six domain ontologies to BOT
• Subsumption
Step 2
• Use of ontology matching tool (AlignmentMakerLight)
• Match six domain ontologies to BOT
• Default settings
Step 3• Compare results
© G.F. Schneider
Agenda
• Introduction and Motivation
• Methodology for Case Study
• Manually Defined/ Revised Alignments
• Results from Automated Ontology Matching
• Conclusion and Future Work
19/06/2019 10
© G.F. Schneider
Manually Defined Alignments to BOTDERIRoom2BOT
19/06/2019 11
bot:Spacebot:Building
rdfs:subPropertyOf
rdfs:subPropertyOf
Figure source: http://vocab.deri.ie/rooms
bot:Site
rdfs:subClassOfrdfs:subClassOf
bot:Storey
rdfs:subClassOf
rdfs:subClassOf
rdfs:subClassOf
bot:Element
rdfs:subClassOf
bot:containsZone
Alignments for all classes and object properties of DERI rooms found!
© G.F. Schneider
Manually Defined Alignments to BOTSAREF4Building2BOT
19/06/2019 12
bot:Element
bot:Space
rdfs:subClassOf
bot:Building
rdfs:subClassOf
bot:containsZone
rdfs:subPropertyOf
bot:containsElement
rdfs:subPropertyOf
Figure source: http://ontoology.linkeddata.es/publish/saref4bldg/index-en.html
rdfs:subClassOf
© G.F. Schneider
Manually Defined Alignments to BOTOverview
19/06/2019 13
Total number of alignments does not qualify as metric that v0.3.0 is better
Number of object property level alignments increased
© G.F. Schneider
Agenda
• Introduction and Motivation
• Methodology for Case Study
• Manually Defined/ Revised Alignments
• Results from Automated Ontology Matching
• Conclusion and Future Work
19/06/2019 14
© G.F. Schneider
Automated AlignmentAlignmentMakerLight
19/06/2019 15
– Default settings
– Well-documented, open source
– Source ontology BOT
– Target ontology six domain ontologies
– Tool implements primary and secondary matching algorithms
© G.F. Schneider
Results from Automated Alignment
19/06/2019 16
© G.F. Schneider
Agenda
• Introduction and Motivation
• Methodology for Case Study
• Manually Defined/ Revised Alignments
• Results from Automated Ontology Matching
• Conclusion and Future Work
19/06/2019 17
© G.F. Schneider
Conclusion
• Findings:
– Revision of manual alignments to BOT (v0.3.0)
– Comparison between manually and automatically obtained alignments
• Limitations
– Only single matching tool with default settings
– Set of domain ontologies biased by authors expertise and interest
• Future work:
– Call for participation: Well-defined benchmark for AEC/FM domain -> OAEI
• Additional sub-domains
– What if ontologies (incrementally) evolve? -> “Hot-start” possibility of matching methods?
19/06/2019 18
>>> https://github.com/w3c-lbd-cg/bot/tree/AlignmentRevision <<<
© G.F. Schneider
AcknowledgementsThank you very much for your attention
19/06/2019 19
Parts of this research have been supported by initiative Mittelstand 4.0 by the German Federal Ministry for Economic Affairs and Energy. (https://www.kompetenzzentrum-planen-und-bauen.digital/)
Georg Ferdinand SchneiderTechnische Hochschule Nürnberg & Fraunhofer Institute for Building Physics IBPFürther Straße 25090429 NürnbergGermany
Phone: +49 911 56854-9145Mail: [email protected]: [email protected]: Georg_Schneider3ORCID: 0000-0002-2033-859X
Parts of this research have been supported by the MOEEBIUS project, a Horizon 2020 research and innovation program under grant agreement No 680517. (http://www.moeebius.eu/)
© G.F. Schneider
Backup
19/06/2019 20