Telecon Mapping Session Seacoos+ GCMD, CF, AOSN

MMI hosted a Web Conference for vocabularies integration via ontologies, the June 14th 2005 at 10:00 am pacific time. The goal of the meeting was to show in more detail how is the vocabulary mapping being done at MMI, by showing how to create relations using an ontology mapping tool created at MMI.

Introduction and Goal

MMI hosted a Web Conference for vocabularies integration via ontologies, the June 14th 2005 at 10:00 am pacific time. The meeting lasted for one hour. There were about 10 participants from different organizations: Mike Godin (AOSN), Donna Cote (TAMUS), Matthew Arrot (LOOKING), Robert Arko (LDEO), Lola Olsen (GCMD), Maryln Drewery(UAH - SCOOP), Cindy Chandler (WHOI), Luis Bermudez (MMI).

The goal of the meeting was to show in more detail how is the vocabulary mapping being done at MMI, by showing how to create relations using an ontology mapping tool created at MMI. Only air - water and due temperature terms were mapped from SEACOOS vocabulary to GCMD, CF and AOSN. 11 direct mappings were created where 30 new relations can be inferred.

Related links:

The mapping tool

Screen shots of the tool used are presented in the figures 1 and 2. The xml file product of the mapping is presented in Figure 3.

Figure 1. MMI mapping tool - View mapping area

Figure 2. MMI mapping tool - View of mapping results

Figure 3.Mappings in XML

Vocabulary inference

Because of the inference capabilities of OWL (Web Ontology Language), 41 total relations were created. For example air_temperature:seacoos was mapped with other three terms (See the three first relations in the Figure 2). The right of Figure 3 shows the related terms of Air_temperature: gcmd. It has 4 sameAs:owl relations. All of these were inferred due to the transitive and inverse properties of the sameAs: owl relation ( A=B -> B=A and also A=A). Air_temperature:gcmd is the sameAs :owl air_temperature:cf, even do it was not explicitly stated.

Feedback

  • Versioning control : Need to keep track of versions. How to relate previous terms with new ones? How to handle if a term is deleted?
  • For the user interface is good to know which terms have been mapped, and which ones are not yet mapped.
  • Present the relations in a tree view. Maybe a tree for broader relations, for narrower etc...
  • Sometimes relations created need to be reviewed. Need to think about a publishing workflow mechanism.
  • MMI should recommend authoritative terms, based on pre reviewed mappings made by scientists.
  • A visualization tool (star type, like concept vista of visual thesaurus) is the best way to display relations. This will be used by the reviewers to navigate the ontologies and accept or discard the mappings.
  • Circles and inconsistencies need to be checked. For example a term cannot be broader than and narrower-than to itself.
Posted June 14th, 2005 by luisBer