Presentation is loading. Please wait.

Presentation is loading. Please wait.

1 Ontology Development Ecoinformatics International Technical Collaboration Seattle, Washington, USA January 27, 2010 Bruce Bargmeyer Lawrence Berkeley.

Similar presentations


Presentation on theme: "1 Ontology Development Ecoinformatics International Technical Collaboration Seattle, Washington, USA January 27, 2010 Bruce Bargmeyer Lawrence Berkeley."— Presentation transcript:

1 1 Ontology Development Ecoinformatics International Technical Collaboration Seattle, Washington, USA January 27, 2010 Bruce Bargmeyer Lawrence Berkeley National Laboratory Tel:

2 Topics F Overview of ontology development project F Overview of ontology results F Some observations

3 Scope F General applications u Facilitate information discovery, search, retrieval, assessment, interoperation, reasoning, modeling, simulation, analysis and integration across organizations and domains n Databases, sensors, simulations, documents, etc. u Focus on terminology/concepts that are of use for Simulations, Analysis and Modeling with an emphasis on integrated modeling u General application: advance the state-of-the-art of monitoring and assessment u We focused the effort on a particular Use Case 3

4 4 limited semantics fuller semantics Concepts and relations expressed in an ontology representation language that provides formal semantics (i.e., specifies logical inferences). Formal Ontology Taxonomy/Thesaurus Terms (possibly with definitions) & relations between terms Glossary Terms associated with definitions (concepts) Conceptual Model Concepts and relations among them in a modeling language ER/UML OWL, Common Logic Increasing semantics expressivity Keywords Terms Ontology Development

5 Semantics Management Discipline F While we skipped over some of the past technologies, we Must keep in mind the discipline and lessons learned with prior technologies. u Keywords: Term list, may be undefined u Glossary: Terms defined in natural language, limited relation between terms (may be considered concepts) u Thesaurus/Taxonomy: Terms defined, some relations (broader than, narrower than, related to, isA) F We need new disciplines to manage new capabilities. u Manage ontology concepts, properties, relations, axioms, natural language definitions, formal definitions, computable meaning (sameAs, disjointWith), description logic u Common Logic: Full first order logic. 5

6 Methodology F There is no Ontological Engineering equivalent to Knowledge or Software Engineering. There are no standard methodologies for building ontologies. F We had to develop and evolve our own approach and methodology as the purpose, skills, and capabilities emerged. F.General approach u Collaborate between ontologists (mostly geeks) and Subject Matter Experts (SMEs) to identify and enter content u Set up tools for collaboration u Weekly telecons to interact between participants 6

7 Ontology Issues F Different use cases call for different ontology content and structure F A single ontology may not address all desired uses F It may be more practical to develop and manage specialized ontologies that are related F Definitions of concepts and relations u Initially, we used somewhat loose natural language definitions (or none). Depending on use cases, it will desirable in the future to develop more formal definitions. u At the beginning, to get a quick start, we did not enforce discipline on the development of definitions. As we proceed, we want to utilize ISO 704:2000 as guidelines for developing definitions for concepts and relations. We will also utilize ISO/IEC Part 4, for data definitions. 7

8 Content organization 8 W5HUnitsPlacesImported Ontologies & SAM-WIKI Generic Industry Domain AbstractionsSEMATECH & CHIP-WIKI Domain Individuals KB Occurants (events) Framework Semiconductor Device Mfg. Industry Structure of the ontology Extraction Sensors

9 Domain Knowledge Shared Upper Knowledge Reference Systems FacilitiesGeoSpatialSpectroscopy Geo-Reference Temporal-Reference Remote Sensing Imagery Sensors Physical GeographySpectroscopyGeoSpatialFacilities Processing Chemical Processing Manufacturing Processing

10 Content Construction Content of the ontology F At the core of the framework, we used the four concepts in Ucore: Who, What, When, and Where. u We extended these to cover Why and How (causal graphs). u Ucore does not currently have this represented as an ontology, but has them represented in UML. We created an ontology based on the Ucore material. u Event is the central concept in our ontology and Who, What, When, Where, How and Why are properties of Event. F Identifying concepts and relations u Concepts and relations were identified by a variety of techniques. These include contributions from existing vocabularies, content from the use case, products of analysis and extraction activities, and independent contributions from SMEs. F Special work was required to bridge the gap between the material provided by the SMEs and the content needed for the Use Case. 10

11 Content Construction F To the extent possible, we incorporated existing ontologies identified by SMEs. u Reuse or adapt existing ontologies u Save time u Gain benefit of broader consensus u Facilitate interoperation F Issues raised by integration of existing ontologies u Ontology alignment u Ontology merging u Poor documentation n Implicit assumptions u Choices between alternative ontologies covering same domain 11

12 Content Input for Ontology Construction F Glossaries identified as potentially useful by SMEs F Ontologies identified as potentially useful by SMEs F Term lists (some with definitions) developed by SMEs and Assessment Teams F Content entered into a Wiki by SMEs F UML models for Who, What, Where, When from Ucore F Content extensions needed for Use Case 12

13 Interrelated Ontologies Developed F Interrelated Ontologies u Broader, multi-level framework with a domain independent core plus industry specific content n The core of the framework is based on W5H u SME terms incorporated into the ontologies u “Lightning strike” ontology that goes from general level to data instance level to demonstrate an ontology approach to the Use Case n Small knowledgebase developed for Use Case u Ontologies extended by importing or otherwise incorporating externally developed ontologies 13

14 Tools F After review, we decided to use Protégé 3.x to create the ontology. The newly released Protégé 4.x is missing some of the useful functions of 3.x. We will watch continued development of Protégé to see when to make the switch. Protégé 4.x provides support for OWL 2, so we will want to go there. F To avoid everyone having to install Protégé on their local computers we used Web Protégé. F We built an initial version of a Semantic MediaWiki for collaborative work on ontology development. F For future efforts, want to check out Collaborative Protégé. This is an extension of the existing Protégé system that supports collaborative ontology editing as well as annotation of both ontology components and ontology changes. F We used various visualization tools to display the ontology 14

15 Ontology Output F SME term lists translated into ontologies u facilities-modeling.owl u integrated-modeling.owl u advanced-spectroscopy.owl 15

16 Ontology Output (Cont.) F ChipO.owl F ChipWiki.owl F MicroelectronicsLevel 1.owl F MicroelectronicsLevel 2.owl F SEMATECH.owl F Imports the SAMO ontologies: F Advanced-spectroscopy.owl F Elements.owl F Faciltities_modeling.owl F Integrated-modeling.owl F Intell.owl F Materials.owl F Ogc-gml.owl F SAMO.owl F SAMWiki.owl F Sensor-data.owl F Skos-owl1.dl.rdf F Things.owl F Time-entry.owl F W5H.owl 16

17 Ontology Output (Cont.) F Imports the NASA SWEET ontologies: F Material_thing.owl F Numerics.owl F Phenomena.owl F Process.owl F Property.owl F Space.owl F Substance.owl F Sunrealm.owl F Time.owl F Units.owl F Biosphere.owl F Data.owl F Earthrealm.owl F Human_activities.owl 17

18 Bangkadi Knowledge Base F bangkadi.owl - A small knowledge base for demonstration of the use case 18

19 Visualizations 19

20 Ontology Overview 20

21 Ontology Hierarchy 21

22 Imported SKOS 22

23 Ontology Elephants There is no single real elephant There must be an upper elephant An elephant is abstract An elephant is very abstract There must be a purpose for an elephant: use cases? An elephant is really very simple An elephant is the result of consensus Open vs. Closed Elephant There are only distributed elephants & their mappings Source: Leo Obrst, MITRE

24 Acknowledgements F Kevin Keck, LBNL F Craig Blackhart, LANL F Helen Cui, LANL F Ryan Hohimer, PNNL F Leo Obrst, MITRE 24


Download ppt "1 Ontology Development Ecoinformatics International Technical Collaboration Seattle, Washington, USA January 27, 2010 Bruce Bargmeyer Lawrence Berkeley."

Similar presentations


Ads by Google