war/src/main/java/pnews/NER.java

   1 package pnews;
   2
   3 import java.io.IOException;
   4 import java.util.ArrayList;
   5 import java.util.List;
   6 import java.util.logging.Logger;
   7
   8 import edu.stanford.nlp.ie.crf.CRFClassifier;
   9 import edu.stanford.nlp.util.CoreMap;
  10 import edu.stanford.nlp.util.Triple;
  11 import pnews.servlet.Config;
  12
  13 /** https://stanfordnlp.github.io/CoreNLP/api.html */
  14 public class NER {
  15         private static final String CLASS_NAME = NER.class.getName();
  16         private static final Logger LOG = Logger.getLogger(CLASS_NAME);
  17         private static final CRFClassifier<CoreMap> classifier = CRFClassifier.getDefaultClassifier();
  18
  19         public static List<String> classify(String str, List<String> entities, Config config) throws ClassCastException, ClassNotFoundException, IOException {
  20
  21                 List<Triple<String, Integer, Integer>> triples;
  22                 String w;
  23                 final String FUNCTION_NAME = "classify";
  24
  25                 LOG.entering(CLASS_NAME, FUNCTION_NAME, str);
  26
  27                 OpenNLP.classify(str, entities, config);
  28
  29                 synchronized (classifier) {
  30                         triples = classifier.classifyToCharacterOffsets(str);
  31                 }
  32
  33                 for (Triple<String, Integer, Integer> t: triples) {
  34                         w = str.substring(t.second, t.third);
  35                         if (!config.isBlacklistedEntity(w) && !entities.contains(w))
  36                                 entities.add(w);
  37                 }
  38
  39                 LOG.exiting(CLASS_NAME, FUNCTION_NAME, entities);
  40
  41                 return entities;
  42         }
  43
  44         public static void main(String[] args) throws Exception {
  45                 List<String> lst;
  46
  47                 lst = classify("I live in Washington and New York in United States.", new ArrayList<>(), new Config());
  48                 for (String str: lst)
  49                         System.out.println(str);
  50         }
  51 }