What are they

Vocabularies are lists (or hierarchies) of terms with associated codes. In AnnoCultor we assume that they are represented in RDF and are queried with SPARQL.

Object and property conversion rules allow restructuring source data in XML or SQL query result into some dat model represented in RDF. This allows processing data, originally represented in diferent formats, in a uniform way.

Vocabulary lookup rules alow enriching the data with links to terms from vocabularies. For example, all Paris, Parijs, and Parigi refer to the same city with Geonames code 2988507 that stands for... yes, Paris. AnnoCultor allows loading a vocabulary, extracting the map between labels and codes, and looking up certain fields (places, categories, people) in vocabularies.

Loading a vocabulary

To do vocabulary lookup we need to create a map between term labels that may occur in the data and their codes.

Let us look at the example:

 <ac:vocabularies rdf:parseType="Collection">
   <ac:VocabularyOfTerms rdf:about="terms">
     <ac:comment>Custom vocabulary</ac:comment>
     <ac:file>thesaurus.rdf</ac:file>
     <ac:sparqlQuery>
       SELECT ?code ?label 
       WHERE 
       {
        ?code &lt;http://www.w3.org/2008/05/skos#prefLabel&gt; ?label
       }
     </ac:sparqlQuery>
     <ac:listeners rdf:parseType="Collection"> 
       <ac:OnNormaliseLabel>
         <ac:java>
            return label.toLowerCase();
         </ac:java>
       </ac:OnNormaliseLabel>
     </ac:listeners>
   </ac:VocabularyOfTerms>
 </ac:vocabularies>

Here vocabulary is stored in file thesaurus.rdf, and there are RDF resources that correspond to vocabulary codes, and they have property skos:prefLabel holding term preferred labels. The SPARQL query gets the map code-label used to look these terms up.

Custom queries should always return pairs code-label.

In the example we realize that the case of labels from the vocabulary differ from the case of terms used in the dataset. We fix this by creating a listener OnNormaliseLabel that converts them all to lower case.

How to get fun

In progress.