@Override protected String print(Relation t) { final StringBuilder sb = new StringBuilder(); writeLine(sb, "Value", t.getValue()); writeLine(sb, "Type", t.getRelationshipType() + " [" + t.getRelationSubType() + "]"); writeLine(sb, "Source", t.getSource()); writeLine(sb, "Target", t.getTarget()); return sb.toString(); } }
br.setType(relation.getRelationshipType()); br.setSubType(relation.getRelationSubType()); br.setValue(relation.getValue()); br.setBegin(relation.getBegin()); br.setEnd(relation.getEnd());
@Test public void testRelations() throws Exception { BaleenCollectionReader bcr = getCollectionReader( Re3dReader.PARAM_FOLDER, tmpDir.toAbsolutePath().toString(), Re3dReader.PARAM_ENTITIES, true, Re3dReader.PARAM_RELATIONSHIPS, true, Re3dReader.PARAM_RANDOM_DATES, true); assertTrue(bcr.doHasNext()); bcr.getNext(jCas.getCas()); AnnotationIndex<Entity> entityIndex = jCas.getAnnotationIndex(Entity.class); Entity source = Streams.stream(entityIndex).filter(e -> e.getBegin() == 12).findAny().get(); Entity target = Streams.stream(entityIndex).filter(e -> e.getBegin() == 9).findAny().get(); AnnotationIndex<Relation> relationIndex = jCas.getAnnotationIndex(Relation.class); contains(relationIndex, r -> r.getBegin() == 11); contains(relationIndex, r -> r.getEnd() == 12); contains(relationIndex, r -> r.getValue().equals("")); contains(relationIndex, r -> r.getRelationshipType().equals("HasAttrOf")); contains(relationIndex, r -> r.getSource().equals(source)); contains(relationIndex, r -> r.getTarget().equals(target)); bcr.close(); }
.append(FIELD_SOURCE_TYPE, r.getSource().getType().getShortName()) .append(FIELD_SOURCE_TYPE_FULL, r.getSource().getType().getName()) .append(FIELD_VALUE, r.getValue()) .append(FIELD_TARGET_VALUE, r.getTarget().getValue()) .append(FIELD_TARGET_TYPE, r.getTarget().getType().getShortName())
r.getSource().getCoveredText() + " " + r.getValue() + " " + r.getTarget().getCoveredText()) r -> { assertTrue(r.getSource().getCoveredText().matches("Alphabet Limited|Larry Page")); assertEquals("chief executive officer", r.getValue()); assertTrue(r.getTarget().getCoveredText().matches("Alphabet Limited|Larry Page")); });
r.getSource().getCoveredText() + " " + r.getValue() + " " + r.getTarget().getCoveredText()) r -> { assertTrue(r.getSource().getCoveredText().matches("Alphabet Limited|Larry Page")); assertEquals("chief executive officer", r.getValue()); assertTrue(r.getTarget().getCoveredText().matches("Alphabet Limited|Larry Page")); });
@Test public void testLanguageSentenceGivesRelation() throws AnalysisEngineProcessException, ResourceInitializationException { jCas.setDocumentText("Jon, at last, visits London."); final Person person = Annotations.createPerson(jCas, 0, 3, "Jon"); final Location location = Annotations.createLocation(jCas, 21, 27, "London", ""); final uk.gov.dstl.baleen.types.structure.Sentence s = new uk.gov.dstl.baleen.types.structure.Sentence(jCas); s.setBegin(0); s.setEnd(jCas.getDocumentText().length()); s.addToIndexes(); addWordTokens(); processJCas(); final List<Relation> relations = new ArrayList<>(JCasUtil.select(jCas, Relation.class)); Assert.assertEquals(1, relations.size()); final Relation r = relations.get(0); assertEquals(person, r.getSource()); assertEquals(location, r.getTarget()); assertEquals("", r.getRelationshipType()); assertEquals("Jon, at last, visits London.", r.getValue()); assertEquals(0, r.getSentenceDistance()); assertEquals(5, r.getWordDistance()); assertEquals(-1, r.getDependencyDistance()); }
@Test public void testStructureSentenceGivesRelation() throws AnalysisEngineProcessException, ResourceInitializationException { jCas.setDocumentText("Jon, at last, visits London."); final Person person = Annotations.createPerson(jCas, 0, 3, "Jon"); final Location location = Annotations.createLocation(jCas, 21, 27, "London", ""); final uk.gov.dstl.baleen.types.structure.Sentence s = new uk.gov.dstl.baleen.types.structure.Sentence(jCas); s.setBegin(0); s.setEnd(jCas.getDocumentText().length()); s.addToIndexes(); addWordTokens(); processJCas(SentenceRelationshipAnnotator.PARAM_VALUE_STRATEGY, ValueStrategy.BETWEEN); final List<Relation> relations = new ArrayList<>(JCasUtil.select(jCas, Relation.class)); assertEquals(1, relations.size()); final Relation r = relations.get(0); assertEquals(person, r.getSource()); assertEquals(location, r.getTarget()); assertEquals("", r.getRelationshipType()); assertEquals(", at last, visits", r.getValue()); assertEquals(0, r.getSentenceDistance()); assertEquals(5, r.getWordDistance()); assertEquals(-1, r.getDependencyDistance()); }
@Test public void testRegExRelationExtraction() throws AnalysisEngineProcessException, ResourceInitializationException { jCas.setDocumentText("Jon visits London. Steve visited Cheltenham"); final Person jon = Annotations.createPerson(jCas, 0, 3, "Jon"); final Person steve = Annotations.createPerson(jCas, 19, 24, "Steve"); final Location london = Annotations.createLocation(jCas, 11, 17, "London", ""); final Location cheltenham = Annotations.createLocation(jCas, 33, 43, "Cheltenham", ""); processJCas( PARAM_PATTERN, "(:Person:)\\s+(?:visit\\w*|went)\\s+(:Location:)", PARAM_TYPE, LOCATED_AT); final List<Relation> relations = new ArrayList<>(JCasUtil.select(jCas, Relation.class)); assertEquals(2, relations.size()); final Relation jlr = findRelationBetween(relations, jon, london); final Relation scr = findRelationBetween(relations, steve, cheltenham); assertEquals(LOCATED_AT, jlr.getRelationshipType()); assertEquals("Jon visits London", jlr.getValue()); assertEquals(0, jlr.getBegin()); assertEquals(17, jlr.getEnd()); assertEquals("", jlr.getRelationSubType()); assertEquals(LOCATED_AT, scr.getRelationshipType()); assertEquals("", scr.getRelationSubType()); assertEquals("Steve visited Cheltenham", scr.getValue()); assertEquals(19, scr.getBegin()); assertEquals(43, scr.getEnd()); }
@Test public void testBothOnlyGivesOneRelationAndDependencyValue() throws AnalysisEngineProcessException, ResourceInitializationException { jCas.setDocumentText("Jon, at last, visits London."); final Person person = Annotations.createPerson(jCas, 0, 3, "Jon"); final Location location = Annotations.createLocation(jCas, 21, 27, "London", ""); final Sentence s1 = new Sentence(jCas); s1.setBegin(0); s1.setEnd(jCas.getDocumentText().length()); s1.addToIndexes(); final uk.gov.dstl.baleen.types.structure.Sentence s = new uk.gov.dstl.baleen.types.structure.Sentence(jCas); s.setBegin(0); s.setEnd(jCas.getDocumentText().length()); s.addToIndexes(); addDependencyInformation(); processJCas(SentenceRelationshipAnnotator.PARAM_VALUE_STRATEGY, ValueStrategy.DEPENDENCY); final List<Relation> relations = new ArrayList<>(JCasUtil.select(jCas, Relation.class)); Assert.assertEquals(1, relations.size()); final Relation r = relations.get(0); assertEquals(person, r.getSource()); assertEquals(location, r.getTarget()); assertEquals("", r.getRelationshipType()); assertEquals("visits", r.getValue()); assertEquals(0, r.getSentenceDistance()); assertEquals(5, r.getWordDistance()); assertEquals(1, r.getDependencyDistance()); }
assertEquals("visits ", jlr.getValue()); assertEquals(0, jlr.getBegin()); assertEquals(17, jlr.getEnd()); assertEquals(LOCATED_AT, scr.getRelationshipType()); assertEquals(SUB, scr.getRelationSubType()); assertEquals("went to ", scr.getValue()); assertEquals(19, scr.getBegin()); assertEquals(43, scr.getEnd());
assertEquals("visits", jlr.getValue()); assertEquals(0, jlr.getBegin()); assertEquals(17, jlr.getEnd()); assertEquals(LOCATED_AT, scr.getRelationshipType()); assertEquals(SUB, scr.getRelationSubType()); assertEquals("went to", scr.getValue()); assertEquals(19, scr.getBegin()); assertEquals(43, scr.getEnd());
@Test public void test2() throws UIMAException { createPOS(); Person person = new Person(jCas, 0, 1); // H person.addToIndexes(); Money money = new Money(jCas, 59, 71); // $1.8 billion money.addToIndexes(); Temporal date = new Temporal(jCas, 75, 84); // September date.addToIndexes(); processJCas(); assertEquals(2, JCasUtil.select(jCas, Relation.class).size()); Relation r1 = JCasUtil.selectByIndex(jCas, Relation.class, 0); assertEquals("reckons", r1.getValue()); assertEquals(person, r1.getSource()); assertEquals("uk.gov.dstl.baleen.types.semantic.Entity", r1.getTarget().getTypeName()); assertEquals("the current account deficit", r1.getTarget().getCoveredText()); Relation r2 = JCasUtil.selectByIndex(jCas, Relation.class, 1); assertEquals("will narrow to", r2.getValue()); assertEquals("uk.gov.dstl.baleen.types.semantic.Entity", r2.getSource().getTypeName()); assertEquals("the current account deficit", r2.getSource().getCoveredText()); assertEquals(money, r2.getTarget()); }
@Test public void test1() throws UIMAException { createPOS(); Person person = new Person(jCas, 0, 2); // He person.addToIndexes(); Money money = new Money(jCas, 59, 71); // $1.8 billion money.addToIndexes(); Temporal date = new Temporal(jCas, 75, 84); // September date.addToIndexes(); processJCas(); assertEquals(2, JCasUtil.select(jCas, Relation.class).size()); Relation r1 = JCasUtil.selectByIndex(jCas, Relation.class, 0); assertEquals("reckons", r1.getValue()); assertEquals(person, r1.getSource()); assertEquals("uk.gov.dstl.baleen.types.semantic.Entity", r1.getTarget().getTypeName()); assertEquals("the current account deficit", r1.getTarget().getCoveredText()); Relation r2 = JCasUtil.selectByIndex(jCas, Relation.class, 1); assertEquals("will narrow to", r2.getValue()); assertEquals("uk.gov.dstl.baleen.types.semantic.Entity", r2.getSource().getTypeName()); assertEquals("the current account deficit", r2.getSource().getCoveredText()); assertEquals(money, r2.getTarget()); }
@Test public void test3() throws UIMAException { createPOS(); Person person = new Person(jCas, 0, 2); // He person.addToIndexes(); Money money = new Money(jCas, 54, 71); // only $1.8 billion money.addToIndexes(); Temporal date = new Temporal(jCas, 75, 84); // September date.addToIndexes(); processJCas(); assertEquals(2, JCasUtil.select(jCas, Relation.class).size()); Relation r1 = JCasUtil.selectByIndex(jCas, Relation.class, 0); assertEquals("reckons", r1.getValue()); assertEquals(person, r1.getSource()); assertEquals("uk.gov.dstl.baleen.types.semantic.Entity", r1.getTarget().getTypeName()); assertEquals("the current account deficit", r1.getTarget().getCoveredText()); Relation r2 = JCasUtil.selectByIndex(jCas, Relation.class, 1); assertEquals("will narrow to", r2.getValue()); assertEquals("uk.gov.dstl.baleen.types.semantic.Entity", r2.getSource().getTypeName()); assertEquals("the current account deficit", r2.getSource().getCoveredText()); assertEquals(money, r2.getTarget()); }
assertEquals("visited", r1.getValue()); assertEquals(0, r1.getSentenceDistance()); assertEquals(1, r1.getWordDistance()); assertEquals("visited London", r2.getValue()); // Not great, but is the shortest dependency path assertEquals(0, r2.getSentenceDistance()); assertEquals(3, r2.getWordDistance());