Collona final presentation

167 views

Published on

Final presentation of the project I developed at the IRSS2013 in Athens, July 2013

Published in: Education
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
167
On SlideShare
0
From Embeds
0
Number of Embeds
1
Actions
Shares
0
Downloads
2
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide
  • OntologyNetworkis a container for a set of ontologiesand a set of alignments. It makes it easy to retrievealignments tied to an ontology as well as to manipulate them as a network, i.e., traversing them,closing them, etcAlignmentis the main class of the API. An Alignmentis mostly made of a set of Cells and metadataabout the alignment, such as the aligned ontologies, the alignment arity, provenance metadata,and any other metadata that can be tied to analignment.Cellrepresents a correspondence: it relates two entities with a Relation. The entities may be anyidentified element of an Ontology orany construct from the expressive EDOAL language (see §4). In addition, Cell supports anytype of additional metadata (including confidencevalues).Relation represents the relation between two entities.The set and type of relations are extensible in theAlignment API and its implementation.
  • Collona final presentation

    1. 1. Collaborative Ontology Alignment Final Project Presentation Giorgos Papoutsakis Jaakko Lapplaianen IRSS13 | 4-31 July | NCSR Demokritos COLLONA
    2. 2. Outline Ontology Ontology Alignment COLLaborative ONtology Alignment Future Work IRSS13 | 4-31 July | NCSR Demokritos • • • • 2
    3. 3. Ontologies 101 IRSS13 | 4-31 July | NCSR Demokritos COLLONA
    4. 4. Ontologies - What • What kinds of things exist or can exist in the world? • what manner of relations can those things have to each other? • Ontology is more concerned with what is possible • Ontology in Computer Science “An ontology formally represents knowledge as a set of concepts within a domain, and the relationships between pairs of concepts.” IRSS13 | 4-31 July | NCSR Demokritos • Definition: “An ontology is defined as a formal explicit specification of a shared conceptualization” • model a domain • support reasoning about concepts. 4
    5. 5. Ontologies - Web Ontology Language Functional syntax XML syntax Manchester syntax RDF/XML syntax RDF/Turtle syntax <Ontology ontologyIRI="http://example.com/tea.owl" ...> Ontology(<http://example.com/tea.owl> <rdf:RDF <Prefix name="owl" ...> Declaration( Class( :Tea ) ) ) <http://example.com/tea.owl> rdf:type owl:Ontology . Ontology: <http://example.com/tea.owl> <owl:Ontology rdf:about=""/> IRI="http://www.w3.org/2002/07/owl#"/> :Tea rdf:typeTea Class: owl:Class . <owl:Class rdf:about="#Tea"/> <Declaration> <Class IRI="Tea"/> </Declaration> </rdf:RDF> </Ontology> IRSS13 | 4-31 July | NCSR Demokritos • • • • • 5
    6. 6. Ontologies - Example IRSS13 | 4-31 July | NCSR Demokritos 6
    7. 7. Ontology - Applications Semantic web Specify a data modeling representation Specify standard conceptual vocabularies Provide services for answering queries Publish reusable knowledge bases Offer services to facilitate interoperability across multiple heterogeneous systems and databases IRSS13 | 4-31 July | NCSR Demokritos • • • • • • 7
    8. 8. Ontology alignment IRSS13 | 4-31 July | NCSR Demokritos COLLONA
    9. 9. Alignment • Classes • Properties • 3 kind of mappings • Equivalence • Subsumption • Supersumption • Examples IRSS13 | 4-31 July | NCSR Demokritos • Sets of correspondences between elements of two ontologies • Human = Person • Lion < Animal 9
    10. 10. Alignment API • Framework for communication • Well supported • Representation Classes • • • • Ontology network Alignment Cell Relation IRSS13 | 4-31 July | NCSR Demokritos “The Alignment API is a set of abstractions for expressing, accessing and sharing ontology alignments” 10
    11. 11. Alignment applications • Use alignments as links between ontologies • Distributed information retrieval • At query transformation step • Understanding the semantics of the question • At result fusion step • Putting together the results • Human – Machine dialogues • Ontology evolution • Instance migration IRSS13 | 4-31 July | NCSR Demokritos • Avoid ontology merging 11
    12. 12. Our work IRSS13 | 4-31 July | NCSR Demokritos COLLONA
    13. 13. COLLONA • Automatic methods and humans need to collaborate to produce accurate results IRSS13 | 4-31 July | NCSR Demokritos • Automatic alignment methods are not very accurate • Humans can not handle heavy work load • Humans are slow 13
    14. 14. COLLONA – Core Tasks • Visualization of Ontological pairs. • Authoring of mappings by users. • Visualization of automatically produced mapping. IRSS13 | 4-31 July | NCSR Demokritos • Review of the main concepts regarding semi-automatic alignment systems and standard APIs. 14
    15. 15. COLLONA – Arch. & Data Flows IRSS13 | 4-31 July | NCSR Demokritos 15
    16. 16. COLLONA – Standard data formats • From ontologies files • From URI • From alignment file • Export • Alignment file IRSS13 | 4-31 July | NCSR Demokritos • Import 16
    17. 17. COLLONA – First problems IRSS13 | 4-31 July | NCSR Demokritos 17
    18. 18. COLLONA – Further problems IRSS13 | 4-31 July | NCSR Demokritos 18
    19. 19. COLLONA - Divide & Conquer • Panels for class information • Properties • Axioms • Panels for mappings • Class mappings • Property Mappings IRSS13 | 4-31 July | NCSR Demokritos • Graph shows only the class herierarchy 19
    20. 20. COLLONA – More features IRSS13 | 4-31 July | NCSR Demokritos 20
    21. 21. COLLONA today IRSS13 | 4-31 July | NCSR Demokritos 21
    22. 22. Future work IRSS13 | 4-31 July | NCSR Demokritos COLLONA
    23. 23. Future work - Synergies IRSS13 | 4-31 July | NCSR Demokritos • The tool does not distinguish between machines and users when importing/exporting mappings • The user becomes a new alignment method • This homogeneous feature allows • Machine alignment / user validator • Machine / user competition • Hybrid approaches, complex data flows for iterative mapping refinement • Alignment as a Service – Synergy with CLIC project • This is a step forward on ontology mapping quality 23
    24. 24. User validation Bad computer! Stupid Computer! IRSS13 | 4-31 July | NCSR Demokritos • Supervised alignment • Algorithm evaluation and refinement 24
    25. 25. Machine-user competition IRSS13 | 4-31 July | NCSR Demokritos • Expert evaluation and training 25
    26. 26. Hybrid approaches Custome Mappings r IRSS13 | 4-31 July | NCSR Demokritos • Combine human and machine workflows to iteratively refine ontology mappings 26
    27. 27. COLLONA @ CLIC Framework Customer Mappings New Mappings Mapping request Customer Negotiation layer Negotiation Alignment service 1 Alignment service 2 Alignment service 3 Service registration, data ALIGN!
    28. 28. Thank you!! IRSS13 | 4-31 July | NCSR Demokritos COLLONA

    ×