package edu.stanford.nlp.pipeline;
import edu.stanford.nlp.ie.NERClassifierCombiner;
import edu.stanford.nlp.ling.CoreAnnotations;
import edu.stanford.nlp.ling.CoreLabel;
import edu.stanford.nlp.sequences.ColumnTabDocumentReaderWriter;
import edu.stanford.nlp.util.CoreMap;
import junit.framework.TestCase;
import java.io.StringReader;
import java.util.Iterator;
import java.util.List;
import java.util.Properties;
/**
* @author Angel Chang
* @author John Bauer
*/
public class NERCombinerAnnotatorITest extends TestCase {
public static final String NER_3CLASS = DefaultPaths.DEFAULT_NER_THREECLASS_MODEL;
public static final String NER_7CLASS = DefaultPaths.DEFAULT_NER_MUC_MODEL;
public static final String NER_MISCCLASS = DefaultPaths.DEFAULT_NER_CONLL_MODEL;
private static NERCombinerAnnotator nerAnnotator = null;
private static AnnotationPipeline unthreadedPipeline = null;
private static AnnotationPipeline threaded4Pipeline = null;
/**
* Creates the tagger annotator if it isn't already created
*/
@Override
public void setUp()
throws Exception
{
synchronized(NERCombinerAnnotatorITest.class) {
if (nerAnnotator == null) {
nerAnnotator = new NERCombinerAnnotator(false, NER_3CLASS, NER_7CLASS, NER_MISCCLASS);
Properties props = new Properties();
props.setProperty("ner.applyNumericClassifiers", "false");
props.setProperty("ner.useSUTime", "false");
props.setProperty("ner.model", NER_3CLASS);
NERClassifierCombiner ner = NERClassifierCombiner.createNERClassifierCombiner("ner", props);
NERCombinerAnnotator threaded4Annotator = new NERCombinerAnnotator(ner, false, 4, -1);
threaded4Pipeline = new AnnotationPipeline();
threaded4Pipeline.addAnnotator(new TokenizerAnnotator(false, "en"));
threaded4Pipeline.addAnnotator(new WordsToSentencesAnnotator(false));
threaded4Pipeline.addAnnotator(threaded4Annotator);
NERCombinerAnnotator unthreadedAnnotator = new NERCombinerAnnotator(ner, false, 1, -1);
unthreadedPipeline = new AnnotationPipeline();
unthreadedPipeline.addAnnotator(new TokenizerAnnotator(false, "en"));
unthreadedPipeline.addAnnotator(new WordsToSentencesAnnotator(false));
unthreadedPipeline.addAnnotator(unthreadedAnnotator);
}
}
}
public void testPipelineAnnotator() {
Annotation document = new Annotation(TEXT);
unthreadedPipeline.annotate(document);
verifyAnswers(ANSWERS, document);
}
public void testThreadedAnnotator() {
Annotation document = new Annotation(TEXT);
threaded4Pipeline.annotate(document);
verifyAnswers(ANSWERS, document);
document = new Annotation(TEXT + TEXT + TEXT);
threaded4Pipeline.annotate(document);
verifyAnswers(ANSWERS, document);
}
public static void verifyAnswers(String[][] expected, Annotation document) {
int sentenceIndex = 0;
for (CoreMap sentence : document.get(CoreAnnotations.SentencesAnnotation.class)) {
List<CoreLabel> tokens = sentence.get(CoreAnnotations.TokensAnnotation.class);
assertEquals(expected[sentenceIndex % expected.length].length, tokens.size());
int token = 0;
for (CoreLabel word : sentence.get(CoreAnnotations.TokensAnnotation.class)) {
assertEquals(expected[sentenceIndex % expected.length][token], word.ner());
++token;
}
++sentenceIndex;
}
}
public static final String TEXT = "John Bauer used to work at Stanford. He worked there for 4 years. John left in August 2014. ";
public static final String[][] ANSWERS = {
{ "PERSON", "PERSON", "O", "O", "O", "O", "ORGANIZATION", "O" },
{ "O", "O", "O", "O", "O", "O", "O" },
{ "PERSON", "O", "O", "O", "O", "O" }
};
private static Iterator<Annotation> getTestData(String inputString, boolean includeAnswer)
{
ColumnTabDocumentReaderWriter<CoreMap> colReader = new ColumnTabDocumentReaderWriter<>();
if (includeAnswer) {
colReader.init("word=0,tag=1,answer=2");
} else {
colReader.init("word=0,tag=1");
}
StringReader strReader = new StringReader(inputString);
return colReader.getDocIterator(strReader);
}
private static void checkAnnotation(String goldInputString) throws Exception
{
// Use separate sets for gold and test since the NER annotator may write stuff to the AnswerAnnotation
Iterator<Annotation> goldDocs = getTestData(goldInputString, true);
Iterator<Annotation> testDocs = getTestData(goldInputString, false);
int k = 0;
while (testDocs.hasNext()) {
Annotation goldDoc = goldDocs.next();
Annotation testDoc = testDocs.next();
nerAnnotator.annotate(testDoc);
List<CoreLabel> goldTokens = goldDoc.get(CoreAnnotations.TokensAnnotation.class);
List<CoreLabel> testTokens = testDoc.get(CoreAnnotations.TokensAnnotation.class);
assertEquals("token number", goldTokens.size(), testTokens.size());
for (int i = 0; i < goldTokens.size(); i++) {
CoreLabel goldToken = goldTokens.get(i);
CoreLabel testToken = testTokens.get(i);
//System.err.println("POS: " + testToken.get(CoreAnnotations.PartOfSpeechAnnotation.class));
String goldNer = goldToken.get(CoreAnnotations.AnswerAnnotation.class);
String testNer = testToken.get(CoreAnnotations.NamedEntityTagAnnotation.class);
//System.err.println("Ner tag for token " + i + " doc " + k +", GOLD: " + goldNer + ", TEST:" + testNer);
assertEquals("Ner tag for token " + i + " (\"" + testToken.word() + "\") doc " + k, goldNer, testNer);
}
k++;
}
}
public void testCombinedAnnotation() throws Exception
{
StringBuilder sb = new StringBuilder();
sb.append("EU\tNNP\tORGANIZATION\n");
sb.append("rejects\tVBZ\tO\n");
sb.append("German\tNNP\tMISC\n");
sb.append("call\tNN\tO\n");
sb.append("to\tTO\tO\n");
sb.append("boycott\tVB\tO\n");
sb.append("British\tNNP\tMISC\n");
sb.append("lamb\tNN\tO\n");
sb.append(".\t.\tO\n");
sb.append("Peter\tNNP\tPERSON\n");
sb.append("Blackburn\tNNP\tPERSON\n");
sb.append("BRUSSELS\tNNP\tLOCATION\n");
sb.append("1996-08-22\tCD\tDATE\n");
sb.append("It\tPRP\tO\n");
sb.append("is\tVBZ\tO\n");
sb.append("bright\tJJ\tO\n");
sb.append("during\tIN\tO\n");
sb.append("the\tDT\tDATE\n");
sb.append("day\tNN\tDATE\n");
sb.append(".\t.\tO\n");
sb.append("It\tPRP\tO\n");
sb.append("was\tVBZ\tO\n");
sb.append("2\tJJ\tDURATION\n");
sb.append("days\tIN\tDURATION\n");
sb.append("before\tDT\tO\n");
sb.append("the\tNN\tO\n");
sb.append("meeting\tNN\tO\n");
sb.append(".\t.\tO\n");
checkAnnotation(sb.toString());
}
}