1 / 17

USC Graduate Student Day Columbia, SC March 2006

Understanding Ontologies for Web Service Coordination Jingshan Huang, Rosa Laura Zavala Gutiérrez, Benito Mendoza García, and Michael N. Huhns. Presented by: Jingshan Huang Computer Science & Engineering Department University of South Carolina PhD Student Research Area: Ontology, SOC, MAS

muniya
Download Presentation

USC Graduate Student Day Columbia, SC March 2006

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Understanding Ontologies forWeb Service Coordination Jingshan Huang, Rosa Laura Zavala Gutiérrez, Benito Mendoza García, and Michael N. Huhns Presented by: Jingshan Huang Computer Science & Engineering Department University of South Carolina PhD Student Research Area: Ontology, SOC, MAS Advisor: Dr. Michael N. Huhns USC Graduate Student Day Columbia, SC March 2006

  2. Q: What is Web service? A: A functionality that can be engaged over the Web Q: What is Ontology? A: - A computational model of some portion of the world - A declarative knowledge representation model - A list of nodes + properties + relationships Background Knowledge Jingshan Huang - Computer Science & Engineering Department

  3. I need some publications before graduation  • Different Web services need to understand each other • Ontologies help in this mutual understanding • Impractical to have a global ontology for all Web services • Challenge is then to merge/align different ontologies • We present a new merging algorithm: PUZZLE Motivation Jingshan Huang - Computer Science & Engineering Department

  4. Schema-level merging • Completely automatic • Considers both linguistic and contextual features • Incorporates WordNet • Integrates heuristic knowledge • Three reasoning rules Characteristics of PUZZLE Jingshan Huang - Computer Science & Engineering Department

  5. Goal: construct a merged ontology from numerous ones • Input to PUZZLE: G1, G2, …, Gn • Merge G1 and G2 into G12 • Merge G12 and G3 into G123 • … • Output of PUZZLE: a merged G Overview of PUZZLE System Jingshan Huang - Computer Science & Engineering Department

  6. Merge G1 and G2 = relocate every node in G1 into G2 • Relocate G1’s root into G2’s root • Traverse G1 in a breadth-first order • Based on the new position(s) of each node’s parent(s), relocate that node into G2 How We Merge Two Ontologies Jingshan Huang - Computer Science & Engineering Department

  7. relocate into P’ P’ P’ P’ B C A B C B A B A/C A C P P’ 4 mutually exclusive outcomes of the relocation B C A Relocate a Node into the Target Ontology Jingshan Huang - Computer Science & Engineering Department

  8. Concept meaning = linguistic feature + contextual feature • Linguistic feature = concept name • Contextual feature = property list + relationships with other concepts Details of Relocation — Overview Jingshan Huang - Computer Science & Engineering Department

  9. relocate into P P’ Revisit the example shown before Where to put A in the left graph? B C A All the candidate concepts should be considered • For an exact match or synonym, put it into A’s candidate equivalent-class list • For an anti-postfix or hyponym, put it into A’s candidate subclass list • For a postfix or hypernym, put it into A’s candidate superclass list Details of Relocation – Linguistic Matching Jingshan Huang - Computer Science & Engineering Department

  10. Consider all the combinations between properties from 2 lists • Find the total-match first • Then the name-match • Lastly the datatype-match The similarity v between 2 property lists depends on the numbers found above: v = (v1* w1 + v2 * (w2 + w2’ * f1) + v3 * (w3 + w3’ * f2))/n1 with correcting factors f1 = v1/n1 and f2 = (v1 + v2)/n1 Details of Relocation – PropertyMatching Jingshan Huang - Computer Science & Engineering Department

  11. semantic bridge A C B i2 j2 k2 i1 j1 k1 Three rules are applied to determine relationships • Rule 1: superclass/subclass relationship of a class is transferable to its equivalent class(es) • Rule 2: if two concepts share the same parent, they are either equivalent-class, superclass/subclass, or sibling • Rule 3: an extension of Rule 2, and embodies the idea of a semantic bridge Details of Relocation – Determine Relationships Jingshan Huang - Computer Science & Engineering Department

  12. Experiment Purpose • Determine whether a correctly merged ontology is generated Experiment Data • A set of ontologies in the domains of “Building” • Constructed by graduate students in computer science and engineering Jingshan Huang - Computer Science & Engineering Department

  13. Characteristics of Ontology Schemas in our Experiment Jingshan Huang - Computer Science & Engineering Department

  14. Evaluation of PUZZLE - 1 Precision and Recall Measurements of Resultant Ontology Jingshan Huang - Computer Science & Engineering Department

  15. Evaluation of PUZZLE – 2 Merging Convergence Experiment Jingshan Huang - Computer Science & Engineering Department

  16. Adopt machine learning techniques • Take into account partOf, hasPart, causeOf, and hasCause relationships • Test our system with more general ontologies Future Work Jingshan Huang - Computer Science & Engineering Department

  17. Suggestions? • Comments? • Questions? Thank you!!! Jingshan Huang - Computer Science & Engineering Department

More Related