AnnotationSet lookups = doc.getAnnotations("EXT").get("Lookup");
assertEquals(0,lookups.size());
// run the gazetteer on the document
eg.setDocument(doc);
eg.execute();
AnnotationDiffer differ = new AnnotationDiffer();
differ.setSignificantFeaturesSet(new HashSet<String>());
AnnotationSet keys = doc.getAnnotations().get("Lookup");
System.out.println("Lookups old: "+keys.size());
AnnotationSet responses = doc.getAnnotations("EXT").get("Lookup");
System.out.println("Lookups new: "+responses.size());
differ.calculateDiff(keys, responses);
int correct = differ.getCorrectMatches();
int falsePositives = differ.getFalsePositivesStrict();
int missing = differ.getMissing();
System.out.println("Diff: correct="+correct+" false positives="+falsePositives+" missing="+missing);
File outFile = new File(testingDir,"news1pre_procBE"+backendNr+".xml");
FileUtils.writeStringToFile(outFile, doc.toXml(),"UTF-8");
// the 33 false Positives come from duplicates that are introduced from several list files
// and which are not removed by the ExtGaz