Importing RDF

Revision as of 23:24, 5 February 2008

Wandora reads RDF(S) and N3 files. Import starts with File > Import > Simple RDF(S) Import... or File > Import > Simple N3 Import.... Optionally you can drag and drop RDF(S) files to layer stack. Layer stack automatically imports dropped RDF(S) file and creates a new layer for the file. Wandora converts imported RDF triplets to topics, associations and occurrences. Convert schema is very simple and pays no attention to semantics of RDF(S) file. Lets see the conversion process more detailed.

• A topic is always created for RDF subject and predicate. Topics created for the subject and predicate are typed with Wandora's predefined type topics.

• If object is RDF literal, an occurrence (text data) is created for the subject topic. Occurrence's type is the predicate topic and occurrence's value the RDF literal. Occurrence's scope is derived from lang attribute. If lang attribute is not found, scope is language independent.

• If object is not RDF literal, a topic is created for the object and the topic is associated with the subject topic. Association's type is the predicate topic. Both roles are Wandora's predefined topics. Object topic is typed with Wandora's predefined type topic.

Created topics doesn't contain base names or variant names. Created topics inherit one subject identifier from equivalent RDF resource. Subject identifier is the URI of equivalent RDF resource. Wandora employs Jena RDF framework to read RDF(S) files. Below is the Java code snippet used to handle RDF statements in Wandora.

   public void handleStatement(Statement stmt, TopicMap map,
                               Topic subjectType,
                               Topic predicateType,
                               Topic objectType) throws TopicMapException {
       
       Resource subject   = stmt.getSubject();     // get the subject
       Property predicate = stmt.getPredicate();   // get the predicate
       RDFNode object     = stmt.getObject();      // get the object
       String lan         = null;                  // language attribute
       
       Topic subjectTopic = getOrCreateTopic(map, subject.toString());
       Topic predicateTopic = getOrCreateTopic(map, predicate.toString());
       
       subjectTopic.addType(subjectType);
       predicateTopic.addType(predicateType);
      
       if(object.isLiteral()) {
           try { lan = stmt.getLanguage(); } catch(Exception e) { /* LANG ATTRIBUTE NOT FOUND! */ }
           if(lan==null || lan.length()==0) {
              subjectTopic.setData(predicateTopic,
                                getOrCreateTopic(map, occurrenceScopeSI),
                                                 ((Literal) object).getString());
           }
           else {
              subjectTopic.setData(predicateTopic,
                                getOrCreateTopic(map, XTMPSI.getLang(lan)),
                                                 ((Literal) object).getString());
           }
       }
       else if(object.isResource()) {
           Topic objectTopic = getOrCreateTopic(map, object.toString());
           Association association = map.createAssociation(predicateTopic);
           association.addPlayer(subjectTopic, subjectType);
           association.addPlayer(objectTopic, objectType);
           objectTopic.addType(objectType);
       }
       else if(object.isURIResource()) {
           log("URIResource found but not handled!");
       }        
   }

Post-processing the imported RDF(S)

To make the imported RDF(S) more topic mappish you may want to modify it after import. This chapter discusses about the post-processing techniques to make the RDF-imported topic map more convenient.

Constructing base names

First step is to add base names for the topics. RDF(S) originated topics contain no base name and only one subject identifier. You can create base name with topic's subject identifier using Make base name with SI tool found in topic table's context menu under Topics > Base names. Base name is automatically constructed using filename and anchor of the subject identifier URLs. If your topic map contains subject identifiers with identical filenames you should take extra care of these topics to prevent automatic merge of topics.

Second step is to clean up base names. You can use Topics > Base names > Regex replace... to filter out undesired parts of the base names.

Constructing variant names

Third step is to generate variant names from RDF label occurrences. Generally RDF document carries labels attached to RDF concepts. Labels may be language dependent. If such labels exists, a label occurrence is associated to RDF topic. To generate variant names from RDF label occurrences, select all RDF topics and use tool Topics > Variant names > Make display variants with occurrences.

Processing associations

Final step is to change roles of RDF(S) originated associations. By default these roles are

• http://www.wandora.net/core/rdf-subject
• http://www.wandora.net/core/rdf-object

You can not just edit these two role topics as all players share role topics. Instead you need to modify associations with Change association role... and Change association type... tools found in context menu association table. In general this step includes subtasks:

• For each association type
  • Open association type topic
  • Select all associations within the association table
  • For each role use tool Change association role...
  • Use tool Change association type... if necessary

Note that you may also need to create new role topics for each association type.