private List<ConLLWord> addWordResourcesToModel(
OntModel outputModel,
List<ConLLWord> wordObjectsOfSentence,
Individual sentenceResource,
Individual contextResource) {
for (ConLLWord word : wordObjectsOfSentence) {
// generate URI differently
String uri =
contextResource.getURI().substring(0, contextResource.getURI().lastIndexOf("=") + 1)
+ word.getStart()
+ ","
+ word.getEnd();
Individual wordResource =
outputModel.createIndividual(
uri, outputModel.createClass(NIFOntClasses.RFC5147String.getUri()));
wordResource.addOntClass(NIFOntClasses.Word.getOntClass(outputModel));
wordResource.addOntClass(NIFOntClasses.String.getOntClass(outputModel));
wordResource.addProperty(
NIFDatatypeProperties.beginIndex.getDatatypeProperty(outputModel), word.getStart() + "");
wordResource.addProperty(
NIFDatatypeProperties.endIndex.getDatatypeProperty(outputModel), word.getEnd() + "");
wordResource.addLiteral(
NIFDatatypeProperties.anchorOf.getDatatypeProperty(outputModel),
outputModel.createLiteral(word.getWordString()));
wordResource.addProperty(
NIFDatatypeProperties.posTag.getDatatypeProperty(outputModel), word.getPos());
if (tagsetKnown) addOliaPos(wordResource, word.getPos(), outputModel);
wordResource.addProperty(
NIFObjectProperties.referenceContext.getObjectProperty(outputModel), contextResource);
// TODO: we may need something different here
if (!word.getPosFine().equals("_"))
wordResource.addProperty(
NIFDatatypeProperties.posTag.getDatatypeProperty(outputModel), word.getPosFine());
if (!word.getLemma().equals("_"))
wordResource.addProperty(
NIFDatatypeProperties.lemma.getDatatypeProperty(outputModel), word.getLemma());
// TODO: add genus, numerus etc here
wordResource.addProperty(
NIFObjectProperties.sentence.getObjectProperty(outputModel), sentenceResource);
word.setResource(wordResource);
sentenceResource.addProperty(
NIFObjectProperties.word.getObjectProperty(outputModel), wordResource);
sentenceResource.addProperty(
NIFObjectProperties.referenceContext.getObjectProperty(outputModel), contextResource);
}
return wordObjectsOfSentence;
}
public void transform(OntModel inputModel, OntModel outputModel, NIFParameters nifParameters) {
this.prefix = nifParameters.getPrefix();
String uri = this.prefix + "char=0,";
// only supporting RFC5147 string atm
contextResource =
outputModel.createIndividual(
uri, outputModel.createClass(NIFOntClasses.RFC5147String.getUri()));
contextResource.addOntClass(NIFOntClasses.Context.getOntClass(outputModel));
contextResource.addOntClass(NIFOntClasses.String.getOntClass(outputModel));
contextResource.addProperty(
NIFDatatypeProperties.beginIndex.getDatatypeProperty(outputModel), "0");
if (!nifParameters.getOptions().has("informat")) {
log.warn("informat parameter empty, please choose informat=file or informat=text");
}
if (!nifParameters.getOptions().has("tagset")) {
log.warn(
"No tagset chosen, please choose an OLiA tagset from: https://github.com/NLP2RDF/software/blob/master/java-maven/vocabularymodule/OLiA/src/main/java/org/nlp2rdf/vm/olia/models");
} else {
loadTagset(nifParameters.getOptions().valueOf("tagset").toString());
}
if (nifParameters.getOptions().valueOf("intype").equals("file")) {
if (nifParameters.getOptions().valueOf("informat").equals("text")) {
File input = new File(nifParameters.getOptions().valueOf("i").toString());
FileReader reader = null;
try {
reader = new FileReader(input);
this.transformConLL(reader, inputModel, outputModel, nifParameters);
reader.close();
} catch (FileNotFoundException fnf) {
log.error("Could not open file " + nifParameters.getOptions().valueOf("i").toString());
} catch (IOException e) {
log.error("Could not read file " + nifParameters.getOptions().valueOf("i").toString());
}
}
} else if (nifParameters.getOptions().valueOf("intype").equals("url")) {
log.error("URL input not yet supported");
} else {
if (nifParameters.getOptions().valueOf("informat").equals("text")) {
StringReader reader = new StringReader(nifParameters.getOptions().valueOf("i").toString());
this.transformConLL(reader, inputModel, outputModel, nifParameters);
reader.close();
}
}
}
private Individual addSentenceResourceToModel(
OntModel outputModel, String sentence, int startOffset, Individual context) {
int endOffset = startOffset + sentence.length();
String uri =
context.getURI().substring(0, context.getURI().lastIndexOf("=") + 1)
+ startOffset
+ ","
+ endOffset;
// only supporting RFC5147 string atm
Individual sentenceResource =
outputModel.createIndividual(
uri, outputModel.createClass(NIFOntClasses.RFC5147String.getUri()));
sentenceResource.addOntClass(NIFOntClasses.Sentence.getOntClass(outputModel));
sentenceResource.addOntClass(NIFOntClasses.String.getOntClass(outputModel));
sentenceResource.addProperty(
NIFDatatypeProperties.beginIndex.getDatatypeProperty(outputModel), startOffset + "");
sentenceResource.addProperty(
NIFDatatypeProperties.endIndex.getDatatypeProperty(outputModel), endOffset + "");
sentenceResource.addLiteral(
NIFDatatypeProperties.anchorOf.getDatatypeProperty(outputModel),
outputModel.createLiteral(sentence));
return sentenceResource;
}