List of usage examples for edu.stanford.nlp.semgraph SemanticGraph getChildrenWithRelns
public Set<IndexedWord> getChildrenWithRelns(IndexedWord vertex, Collection<GrammaticalRelation> relns)
From source file:count_dep.Count_dep.java
public LinkedList<Event> GetEvents(SemanticGraph dependencies, CoreMap sentence) { LinkedList<Event> res = new LinkedList<>(); LinkedList<IndexedWord> roots = new LinkedList<>(); List<CoreLabel> words = sentence.get(TokensAnnotation.class); List<GrammaticalRelation> senserel = new LinkedList<>(); senserel.add(GrammaticalRelation.valueOf("nsubj")); senserel.add(GrammaticalRelation.valueOf("dobj")); for (CoreLabel word : words) { if (word.tag().length() >= 2 && ("VB".equals(word.tag().substring(0, 2)) || "NN".equals(word.tag().substring(0, 2)))) { IndexedWord iword = new IndexedWord(word); roots.add(iword);/*from w ww .ja va 2s.co m*/ } } for (IndexedWord word : roots) { Event e = new Event(); e.trigger = word.word(); try { Set<IndexedWord> children = dependencies.getChildren(word); children.stream().forEach((iw) -> { e.arguments.add(new EventArgument(iw.word(), "")); }); if (dependencies.inDegree(word) > 0) { IndexedWord parent = dependencies.getParent(word); if (parent.tag().length() >= 2 && "VB".equals(parent.tag().substring(0, 2))) { Set<IndexedWord> children1 = dependencies.getChildrenWithRelns(parent, senserel); children1.remove(word); children1.stream().forEach((iw) -> { e.arguments.add(new EventArgument(iw.word(), "")); }); } else { e.arguments.add(new EventArgument(dependencies.getParent(word).word(), "")); } } } catch (java.lang.IllegalArgumentException error) { continue; } res.add(e); } return res; }
From source file:featureExtractor.NLPFeatures.java
static void processLine(String text, int lineId) throws IOException { bw_root.write(Integer.toString(lineId)); bw_subj.write(Integer.toString(lineId)); bw_underRoot.write(Integer.toString(lineId)); bw_nerType.write(Integer.toString(lineId)); //text = "A gigantic Hong Kong set was constructed in downtown Detroit. The set was so big that the Detroit People Mover track ended up becoming part of the set and shooting had to be adjusted to allow the track to move through the set. ";//"One of three new television series scheduled for release in 2014 based on DC Comics characters. The others being Constantine (2014) and The Flash (2014). "; HashMap<String, Integer> nerCount = new HashMap<>(); int superlativePOS = 0; try {/*from www . j a v a2 s . co m*/ Annotation document = new Annotation(text); pipeline.annotate(document); List<CoreMap> sentences = document.get(CoreAnnotations.SentencesAnnotation.class); for (CoreMap sentence : sentences) { SemanticGraph dependencies = sentence .get(SemanticGraphCoreAnnotations.CollapsedDependenciesAnnotation.class); // getting root words for (IndexedWord rword : dependencies.getRoots()) { //System.out.println(rword.lemma()); //System.out.println(rword.ner()); if (rword.ner().equals("O")) bw_root.write("\t" + rword.ner() + ":" + rword.lemma()); //else if(rword.ner().equals("PERSON")) else bw_root.write("\t" + rword.ner() + ":" + rword.originalText()); /* else bw_root.write(" entity_" + rword.ner()); */ // under root for (IndexedWord child : dependencies.getChildren(rword)) { //System.out.println("here: " + child.originalText()); /* if(child.ner().equals("PERSON")) bw_underRoot.write(" " + child.originalText()); else*/ if (!child.ner().equals("O")) bw_underRoot.write("\t" + child.ner() + ":" + child.originalText()); } // nsubj | nsubpass words GrammaticalRelation[] subjects = { EnglishGrammaticalRelations.NOMINAL_SUBJECT, EnglishGrammaticalRelations.NOMINAL_PASSIVE_SUBJECT }; for (IndexedWord current : dependencies.descendants(rword)) for (IndexedWord nsubWord : dependencies.getChildrenWithRelns(current, Arrays.asList(subjects))) { //System.out.println("wow: " + nsubWord.originalText()); if (!nsubWord.ner().equals("O")) bw_subj.write("\t" + nsubWord.ner() + ":" + nsubWord.originalText()); else { //System.out.println(nsubWord.lemma()); bw_subj.write("\t" + nsubWord.ner() + ":" + nsubWord.lemma()); } /* else bw_subj.write(" entity_"+nsubWord.ner()); */ } } // NER Types frequency for (CoreLabel token : sentence.get(CoreAnnotations.TokensAnnotation.class)) { String pos = token.get(CoreAnnotations.PartOfSpeechAnnotation.class); String ne = token.get(CoreAnnotations.NamedEntityTagAnnotation.class); if (pos.equals("JJS") || pos.equals("RBS")) superlativePOS++; nerCount.putIfAbsent(ne, 0); nerCount.put(ne, nerCount.get(ne) + 1); } //System.out.println("dependency graph:\n" + dependencies); } } catch (Exception e) { System.out.println("IGNORED:"); } bw_nerType.write("\t" + Integer.toString(superlativePOS)); for (String ne : ners) { if (nerCount.containsKey(ne)) bw_nerType.write("\t" + nerCount.get(ne).toString()); else bw_nerType.write("\t0"); } bw_root.write("\n"); bw_underRoot.write("\n"); bw_nerType.write("\n"); bw_subj.write("\n"); if (lineId % 25 == 0) { bw_root.flush(); bw_underRoot.flush(); bw_nerType.flush(); bw_subj.flush(); } }
From source file:me.aatma.languagetologic.graph.nodes.KBNLAntiDStateNodeCloud.java
public static boolean isKBNLAntiDStateNode(SemanticGraph dependencies, IndexedWord localRoot) { boolean isAntiD = false; List<GrammaticalRelation> preps = new ArrayList<GrammaticalRelation>(); // This will force AntiD to have a subject preps.add(NLPTools.getGR("nsubj", null)); preps.add(NLPTools.getGR("csubj", null)); // and some preposition. preps.add(NLPTools.getGR("prep", "of")); preps.add(NLPTools.getGR("prep", "from")); preps.add(NLPTools.getGR("prep", "to")); preps.add(NLPTools.getGR("prep", "on")); preps.add(NLPTools.getGR("prep", "in")); preps.add(NLPTools.getGR("prep", "along")); preps.add(NLPTools.getGR("prep", "than")); preps.add(NLPConstants.prep_with);//from w w w . jav a 2 s. co m Set<IndexedWord> children = dependencies.getChildrenWithRelns(localRoot, preps); isAntiD = (children.size() >= 2); log.info("This node is " + (isAntiD ? "potentially a" : "not a") + " AntiDStateNode."); return isAntiD; }