Missouri Botanical Garden Open Conference Systems, TDWG 2013 ANNUAL CONFERENCE

Font Size: 
Using Darwin Core as a Model: An Ontologically Minimalist Approach to Publishing Occurrence Data in RDF
Joel Sachs

Building: Grand Hotel Mediterraneo
Room: Sala dei Continenti
Date: 2013-11-01 09:02 AM – 09:11 AM
Last modified: 2013-10-07

Abstract


There are a number of ontologies for representing biodiversity occurrence data in RDF. These include Darwin-SW [1], Taxonconcept [2], DwC-FP [3], and the emerging Biological Collections Ontology [4]. These ontologies overlap in some areas, and are incompatible in others. One approach to this multiplicity of models would be to construct a meta-model or an upper ontology capturing the conceptual space of all of the “lower-level” models. This approach, common in the early days of the semantic web, has the drawback of decreasing cognitive (and, frequently, computational) tractability. An alternative approach is to use a “lower ontology” to share data, which can then be ingested into any of the competing models, as necessary, for computation. For biodiversity data, the obvious choice for a lower ontology is Darwin Core itself.

This talk will use data generated at the TDWG Technobioblitz in 2010 to demonstrate an ontologically minimalist approach to publishing Darwin Core occurrence records in RDF, in a way that preserves the intended semantics. We will take a tour of the Technobioblitz dataset [5], which will include a description of the Darwin Core terms used; examples of RDF records for occurrences and identifications; examples of background scientific knowledge expressed in RDF; and examples of SPARQL queries combining all of the above.

As we walk through the above, we will see that heavy-duty ontological apparatus is often not necessary for knowledge sharing and integration. We will then look at use cases (drawn primarily from [6] and [7]) that show various ways in which ontological commitment can, in fact, be a barrier to data integration.

I will conclude with the claim that ontology should, as much as possible, be kept out of Darwin Core, and that the less ontology there is in the shared Core, the easier it will be for different projects to build on it in an inter-operable manner.

References
1. http://code.google.com/p/darwin-sw/
2. http://www.taxonconcept.org/
3. http://filteredpush.org/ontologies/oa/dwcFP.owl
4. http://code.google.com/p/bco/
5. http://lists.tdwg.org/pipermail/tdwg-content/2011-January/002265.html
6. http://code.google.com/p/tdwg-rdf/wiki/UseCases
7. http://code.google.com/p/tdwg-rdf/wiki/CompetencyQuestions