274586869bcf68b6034d17fcb5c38e51fb728343
[pnews.git] / 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
12 /** https://stanfordnlp.github.io/CoreNLP/api.html */
13 public class NER {
14         private static final String CLASS_NAME = NER.class.getName();
15         private static final Logger LOG = Logger.getLogger(CLASS_NAME);
16         private static final CRFClassifier<CoreMap> classifier = CRFClassifier.getDefaultClassifier();
17         
18         public static List<String> classify(String str, List<String> entities) throws ClassCastException, ClassNotFoundException, IOException {
19                 
20                 List<Triple<String, Integer, Integer>> triples;
21                 String w;
22                 final String FUNCTION_NAME = "classify";                       
23                 
24                 LOG.entering(CLASS_NAME, FUNCTION_NAME, str);
25
26                 OpenNLP.classify(str, entities);
27                                 
28                 synchronized (classifier) {
29                         triples = classifier.classifyToCharacterOffsets(str);
30                 }
31                  
32                 for (Triple<String, Integer, Integer> t: triples) {
33                         w = str.substring(t.second, t.third);
34                         if (!entities.contains(w))
35                                 entities.add(w);
36                 }
37                 
38                 entities.remove("CNET");
39                 entities.remove("Read More");
40                 entities.remove("New");
41                 entities.remove("App");
42                 entities.remove("Digital Trends");
43                 
44                 LOG.exiting(CLASS_NAME, FUNCTION_NAME, entities);
45                 
46                 return entities;
47         }
48         
49         public static void main(String[] args) throws Exception {
50                 List<String> lst;
51                 
52                 lst = classify("I live in Washington and New York in United States.", new ArrayList<>());
53                 for (String str: lst)
54                         System.out.println(str);
55         }
56 }