Semantic Web Mini-Tutorial: Data integration of life science data

Once we have our data in RDF format, we would like to link the RDF names to our own concepts in OWL. We call this semantic annotation. This requires us to link RDF types to OWL types. Once we have done that, we can use the OWL names in our queries. The method that we describe here makes use of RDFS subPropertyOf.

In addition to theirDataModel, we also have constructed a personalized knowledge model that describes our domain in familiar concepts called myModel. This model serves as a collection of our own semantics and allows us to use our own terms to access the data. In this example, we have defined a ChromosomeRegion that is a DNAregion and that a ChromosomeRegion isOnChromosome Chromosome. Chromosome has the property chromosomeIdentifier (that should be a string) that we use for linking myModel to the property chrom of theirDataModel.

To map a semantic model to a data model, we make use of the subPropertyOf statement (see Mapping Model for the complete file.):

<owl:DatatypeProperty rdf:ID="chrom"> <rdfs:subPropertyOf rdf:resource="http://www.somewhere.org/myModel.owl#chromosomeIdentifier"/> </owl:DatatypeProperty>

In a Sesame RDFS repository (with RDFS inferencing), source and target nodes of a subProperty are automatically seen as subtypes of corresponding nodes in the target triple:

We might be inclined to store the subPropertyOf statements directly in the data model. However, this would pose a problem in a scenario where we do not control theirDataModels and therefore cannot directly add mapping statements. We therefore have defined the mapping statements in an independent "Bridging" or Mapping Model that is be stored in a separate file. We believe that this allows for a more explicit mapping and provides greater control over the mapping itself.

After transforming the data to RDF and constructing the different data and knowledge models, the whole set is uploaded to the RDFS repository of Sesame (version 1.2.4) and ready for querying. We use the SeRQL-S option to query our example data set using a query defined in terms of myModel:

Table 3: SeRQL-S query

SELECT *
FROM {x} myModel:chromosomeIdentifier {y}

USING NAMESPACE
   myModel = <http://www.somewhere.org/myModel.owl#>
               

The inference functionality of Sesame's RDFS repository allows us to automatically find all the data values associated with the chrom property using our own terms. The result can be returned as an HTML page or RDF file, depending on what you want. In Sesame the HTML page gives the results as links to entries in the repository for further data exploration.