war/src/main/java/pnews/OpenNLP.java

   1 package pnews;
   2
   3 import java.io.IOException;
   4 import java.io.InputStream;
   5 import java.util.List;
   6 import java.util.logging.Logger;
   7
   8 import opennlp.tools.namefind.NameFinderME;
   9 import opennlp.tools.namefind.TokenNameFinderModel;
  10 import opennlp.tools.tokenize.TokenizerME;
  11 import opennlp.tools.tokenize.TokenizerModel;
  12 import opennlp.tools.util.Span;
  13
  14 /** http://www.devglan.com/artificial-intelligence/opennlp-named-entity-recognition-example **/
  15 public class OpenNLP {
  16         private static final String CLASS_NAME = OpenNLP.class.getName();
  17         private static final Logger LOG = Logger.getLogger(CLASS_NAME);
  18         private static TokenNameFinderModel organizationModel;
  19         private static TokenNameFinderModel personModel;
  20         private static TokenNameFinderModel locationModel;
  21         private static TokenNameFinderModel timeModel;
  22         private static TokenizerModel tokenModel;
  23
  24         public static List<String> classify(String str, List<String> entities) throws IOException {
  25                 classify(str, getOrganizationModel(), entities);
  26
  27                 classify(str, getPersonModel(), entities);
  28                 classify(str, getLocationModel(), entities);
  29
  30                 classify(str, getTimeModel(), entities);
  31
  32                 return entities;
  33         }
  34
  35         private static TokenNameFinderModel getOrganizationModel() throws IOException {
  36                 synchronized (OpenNLP.class) {
  37                         if (organizationModel == null) {
  38                                 InputStream inputStream = OpenNLP.class.getResourceAsStream("/en-ner-organization.bin");
  39                                 organizationModel = new TokenNameFinderModel(inputStream);
  40                         }
  41                 }
  42
  43                 return organizationModel;
  44         }
  45
  46         private static TokenNameFinderModel getPersonModel() throws IOException {
  47                 synchronized (OpenNLP.class) {
  48                         if (personModel == null) {
  49                                 InputStream inputStream = OpenNLP.class.getResourceAsStream("/en-ner-person.bin");
  50                                 personModel = new TokenNameFinderModel(inputStream);
  51                         }
  52                 }
  53
  54                 return personModel;
  55         }
  56
  57         private static TokenNameFinderModel getLocationModel() throws IOException {
  58                 synchronized (OpenNLP.class) {
  59                         if (locationModel == null) {
  60                                 InputStream inputStream = OpenNLP.class.getResourceAsStream("/en-ner-location.bin");
  61                                 locationModel = new TokenNameFinderModel(inputStream);
  62                         }
  63                 }
  64
  65                 return locationModel;
  66         }
  67
  68         private static TokenNameFinderModel getTimeModel() throws IOException {
  69                 synchronized (OpenNLP.class) {
  70                         if (timeModel == null) {
  71                                 InputStream inputStream = OpenNLP.class.getResourceAsStream("/en-ner-time.bin");
  72                                 timeModel = new TokenNameFinderModel(inputStream);
  73                         }
  74                 }
  75
  76                 return timeModel;
  77         }
  78
  79         private static List<String> classify(String str, TokenNameFinderModel model, List<String> entities) throws IOException {
  80                 String entity;
  81
  82                 NameFinderME nameFinder = new NameFinderME(model);
  83                 String[] tokens = tokenize(str);
  84                 Span nameSpans[] = nameFinder.find(tokens);
  85
  86                 for(Span s: nameSpans)  {
  87                         if (s.getProb() < 0.60)
  88                                 continue ;
  89
  90                         entity = null;
  91                         for (int i = s.getStart(); i < s.getEnd(); i++)
  92                                 if (entity == null)
  93                                         entity = tokens[i];
  94                                 else
  95                                         entity += " " + tokens[i];
  96
  97                         LOG.finest(entity + " " + s.getProb() + " " + s.toString());
  98                         if (!entities.contains(entity))
  99                                 entities.add(entity);
 100                 }
 101
 102                 return entities;
 103         }
 104
 105         public static String[] tokenize(String sentence) throws IOException {
 106                 synchronized (OpenNLP.class) {
 107                         if (tokenModel == null) {
 108                                 InputStream inputStreamTokenizer = OpenNLP.class.getResourceAsStream("/en-token.bin");
 109                                 tokenModel = new TokenizerModel(inputStreamTokenizer);
 110                         }
 111                 }
 112                 TokenizerME tokenizer = new TokenizerME(tokenModel);
 113                 return tokenizer.tokenize(sentence);
 114         }
 115 }