diff --git a/classifier/README.txt b/classifier/README.txt
index 2407089..5a88272 100644
--- a/classifier/README.txt
+++ b/classifier/README.txt
@@ -128,6 +128,15 @@ that may be identified as being candidates for classification.
 So we are now ready to have the whole stuff run in one call.
 
   >>> classifier.process(r1)
+  >>> list(sorted([c.title for c in r1.getConcepts()]))
+  [u'Collection One', u'Contract', u'External File', u'Gerald Webb', u'im Editors']
+
+  >>> for name in rnames[1:]:
+  ...     classifier.process(resources[name])
+  >>> len(webbg.getResources())
+  4
+  >>> len(webbg.getResources((concepts['ownedby'],)))
+  3
 
 
 Fin de partie
diff --git a/classifier/base.py b/classifier/base.py
index 014b4ce..8feda3d 100644
--- a/classifier/base.py
+++ b/classifier/base.py
@@ -57,13 +57,16 @@ class Classifier(AdapterBase):
             infoSet.update(extractor.extractInformationSet())
         analyzer = component.getAdapter(self, name=self.analyzer)
         statements = analyzer.extractStatements(infoSet)
+        defaultPredicate = self.context.getConceptManager().getDefaultPredicate()
         for statement in statements:
             if statement.subject is None:
                 statement.subject = resource
+            if statement.predicate is None:
+                statement.predicate = defaultPredicate
             self.assignConcept(statement)
 
     def assignConcept(self, statement):
-        pass
+        statement.object.assignResource(statement.subject, statement.predicate)
 
 
 class Extractor(object):