Towards a Project Centric Metadata Model and Lifecycle for Ontology Mapping Governance
Download as PDF, PPTX1 like676 views
The document discusses a project-centric approach to ontology mapping governance, highlighting challenges in current metadata and lifecycle models that are mostly artefact-centric. It proposes an ontology mapping lifecycle and metadata model that integrates various project phases and stakeholder activities, facilitating better governance and transparency. Future work includes tool support development and enhanced user involvement from diverse stakeholders.
![www.adaptcentre.ieContext
• Though the field of ontology matching and mapping is
around for over a decade, Euzenat and Shvaiko iden@fied
several challenges that s@ll have to be tackled (2013).
• Infrastructure and support for ontology alignment
management: “[t]he challenge is to provide convenient and
interoperable support, on which tools and, more
importantly, on which applica@ons, can rely in order to
store and share alignments. This involves using standard
ways to communicate alignments and retrieve them. Hence
alignment metadata and annota@ons should be properly
taken into account.”
• But…](https://image.slidesharecdn.com/iiwas-151213102735/85/Towards-a-Project-Centric-Metadata-Model-and-Lifecycle-for-Ontology-Mapping-Governance-3-320.jpg)
![www.adaptcentre.ieRelated work
Ontology Mapping Metadata [AF, EDEOL, OM2R, …]
• Focuses on represen@ng alignments or designed to facilitate
discovery and reuse.
• Metadata models inappropriately represent the domain; for
instance, “requirements are part of a mapping project” and “a
mapping project results in a mapping” instead of “requirements
are part of a mapping”.
Ontology Mapping Lifecycles [OISIN, Thomas et al, …]
• Likle related work.
• Artefact-centric and most contain no feedback loop.
• Reuse, sharing and publica@on of alignments (or mappings) are
considered separate ac@vi@es not integrated in a mapping’s
lifecycle or workflow.
Provenance informaMon of generated data via mappings
• Not covered in this presenta@on.](https://image.slidesharecdn.com/iiwas-151213102735/85/Towards-a-Project-Centric-Metadata-Model-and-Lifecycle-for-Ontology-Mapping-Governance-5-320.jpg)
![www.adaptcentre.ieMapping Governance
• Data Governance is defined as “a control that ensures that
the data entry […] meets precise standards, such as […]
data integrity constraints in the data model. The data
governor uses data quality monitoring […] to communicate
errors in data […]” (Wikipedia)
• OWL 2 is good for publishing data on the LD Web and
adopts the Open World Assump@on. However, our
platorm would need to adopt a close(r) world assump@on
as to meet our constraints.
• For our tools, this is covered by our database schema and
applica@on logic.
• What if data comes from elsewhere?](https://image.slidesharecdn.com/iiwas-151213102735/85/Towards-a-Project-Centric-Metadata-Model-and-Lifecycle-for-Ontology-Mapping-Governance-11-320.jpg)
More Related Content
Similar to Towards a Project Centric Metadata Model and Lifecycle for Ontology Mapping Governance (20)
Towards a Project Centric Metadata Model and Lifecycle for Ontology Mapping Governance
- 1. Towards a Project Centric Metadata Model and Lifecycle for Ontology Mapping Governance Christophe Debruyne, Brian Walshe and Declan O’Sullivan ADAPT Centre, Trinity College Dublin The ADAPT Centre is funded under the SFI Research Centres Programme (Grant 13/RC/2106) and is co-funded under the European Regional Development Fund.
- 2. www.adaptcentre.ieContext • Seman@c heterogeneity on the Linked Data Web. How can this problem be tackled? With ontology matching and mapping. • Ontology Matching: detec@ng the correspondences that cons@tute an alignment. • Ontology Mapping: interpre@ng these correspondences to create an executable mapping (wrt requirements) Image from Shvaiko and Euzenat’s "Ontology Matching: State of the Art and Future Challenges", IEEE Transac@ons on Knowledge & Data Engineering, vol.25, no. 1, pp. 158-176, Jan. 2013.
- 4. www.adaptcentre.ieProblem • Management ac@vi@es are artefact-centric and concerned with storing and sharing. Related work is ofen limited to (annota@ons of) produced ar@facts. • We argue a project-centric approach should be adopted. • Mappings are created for a purpose • Progress is the result of agreement processes between the community of stakeholders • Generate metadata as the project progresses • Not only formulate queries about alignment, mappings, AND projects: “What are the most debated aspects in a par@cular ontology mapping project?”
- 5. www.adaptcentre.ieRelated work Ontology Mapping Metadata [AF, EDEOL, OM2R, …] • Focuses on represen@ng alignments or designed to facilitate discovery and reuse. • Metadata models inappropriately represent the domain; for instance, “requirements are part of a mapping project” and “a mapping project results in a mapping” instead of “requirements are part of a mapping”. Ontology Mapping Lifecycles [OISIN, Thomas et al, …] • Likle related work. • Artefact-centric and most contain no feedback loop. • Reuse, sharing and publica@on of alignments (or mappings) are considered separate ac@vi@es not integrated in a mapping’s lifecycle or workflow. Provenance informaMon of generated data via mappings • Not covered in this presenta@on.
- 7. www.adaptcentre.ieApproach Our proposed approach • Adopt a project-centric view that • takes into account all mapping project phases and • adequately records the ac@vi@es by the stakeholders • resul@ng in a traceable and transparent process. • Many more artefacts can be related to mapping projects, including scopes, requirements, decisions, discussions, etc. • How? 1. Propose an ontology mapping lifecycle 2. Propose an ontology mapping metadata model 3. Elaborate on ontology mapping governance ac@vi@es
- 8. www.adaptcentre.ieOntology Mapping Lifecycle ReuseCharacterizationStaging Scope Data Discovery Requirements Analysis Data Analysis Discover Evaluate Reuse? Matching Evaluate Execute Plan No Align and Map Create Alignment Prepare Render Mapping Application Yes Source Data Target Data Candidate Correspondences Alignments Mapping Stakeholders Staging Need for communi@es CharacterizaMon Analyze ontologies wrt to scope and requirements Reuse Par@al or complete reuse? Matching Discovering correspondences Mapping Refine correspondences for alignment and create mapping ApplicaMon Monitor mappings
- 9. www.adaptcentre.ieOntology Mapping Metadata Model Ontology Mapping Metadata Model • First conceptualized with Object Role Modelling. • In this paper and presenta@on, we adopted a “UML-like” graphical presenta@on. • The ontology adopts PROV-O and is available as an OWL 2 ontology. hkps://raw.githubusercontent.com/CNGL-repo/mapping- provenance-published/master/mapping-provenance.owl
- 10. www.adaptcentre.ieOntology Mapping Metadata Model OR rdfs:subClassOf 0..1 prov:wasInformedBy 1..1 mg:Ontology Analysis 0..* prov:used 1..1 mg:Evaluation mg:Plan mg:Requirement mg:Reuse Evaluation mg:Matcher Configuration 1..* prov:generated 1..* 0..* prov:used 1..* prov:Entity mg:Matcher 0..* mg:withMatcher 1..1 prov:SoftwareAgent rdfs:subClassOf mg:Reuse 0..* prov:used 1..1 xsd:string 0..* mg:withParameter 1..1 mg:Resource 0..* mg:withResource 1..1 rdfs:subClassOf Planning the matching phase
- 11. www.adaptcentre.ieMapping Governance • Data Governance is defined as “a control that ensures that the data entry […] meets precise standards, such as […] data integrity constraints in the data model. The data governor uses data quality monitoring […] to communicate errors in data […]” (Wikipedia) • OWL 2 is good for publishing data on the LD Web and adopts the Open World Assump@on. However, our platorm would need to adopt a close(r) world assump@on as to meet our constraints. • For our tools, this is covered by our database schema and applica@on logic. • What if data comes from elsewhere?
- 12. www.adaptcentre.ieMapping Governance • Because of the OWA, OWL axioms do not work. • We thus adopt an approach proposed by Tao et al. (2010), where basically constraints are translated into SPARQL SELECT or ASK queries to detect problems !"#$% ⊓ ¬ 1!"#ℎ! . !"#$%&' ⊑⊥ ∃!"#ℎ. !ℎ!"!#$%"&'% ⊓ ¬ ∃!"#ℎ. !"#$% ⊑⊥ ASK WHERE { ?s a mg:Stage. NOT EXISTS { ?project mg:with ?s. } } ASK WHERE { ?p a mg:Project. ?p mg:with ?c. ?c a mg:Characterize. NOT EXISTS { ?p mg:with ?s. ?s a mg:Stage. } }
- 13. www.adaptcentre.ieConclusions and Future Work Conclusions • Crea@on and management of ontology mappings is far from trivial. Metadata models focus on the representa@on of artefacts, on the support for discovery and reuse and are of limited expressivity. • We proposed a new ontology mapping lifecycle and metadata model that adopts a project- centric view and relates artefacts, stakeholders and an ontology mapping project. Facilita@ng mapping governance. Future Work • Build tool support for the method and gather data for evalua@on. • Structure the discussions, adop@ng SIOC and IBIS • Leverage user involvement of the mul@ple different stakeholders (Ramy et al., 2015) • Adopt the Design Intent Ontology (Solanki, 2015), which is an Ontology Design Pakern to structure and “reason” over the requirements.