训练OPenNLP误差

Question

我正在尝试使用OpenNLP来训练一个名称实体模型，但是得到这个错误不知道缺少什么。我是这个OPENNLP的新手，任何一个请帮助，如果需要的话可以提供Train.txt文件。

lineStream = opennlp.tools.util.PlainTextByLineStream@b52598
Indexing events using cutoff of 0

Computing event counts...  done. 514 events
Indexing...  done.
Sorting and merging events... done. Reduced 514 events to 492.
Done indexing.
Incorporating indexed data for training...  
done.
Number of Event Tokens: 492
    Number of Outcomes: 1
  Number of Predicates: 3741
...done.
Computing model parameters ...
Performing 1 iterations.
1:  ... loglikelihood=0.0   1.0
Exception in thread "main" java.lang.IllegalArgumentException: Model not compatible with     name finder!
at opennlp.tools.namefind.TokenNameFinderModel.<init>(TokenNameFinderModel.java:81)
at opennlp.tools.namefind.TokenNameFinderModel.<init>(TokenNameFinderModel.java:106)
at opennlp.tools.namefind.NameFinderME.train(NameFinderME.java:374)
at opennlp.tools.namefind.NameFinderME.train(NameFinderME.java:432)
at opennlp.tools.namefind.NameFinderME.train(NameFinderME.java:443)
at Train2.main(Train2.java:36)
Java Result: 1
BUILD SUCCESSFUL (total time: 2 seconds)

我的代码是

    File fileTrainer=new File("/home/ashfaq/Documents/Train.txt");
    File output=new File("/home/ashfaq/Documents/trainedModel.bin");
    ObjectStream<String> lineStream = new PlainTextByLineStream(new    FileInputStream(fileTrainer), "UTF-8");
    ObjectStream<NameSample> sampleStream = new NameSampleDataStream(lineStream);
    System.out.println("lineStream = " + lineStream);
    TokenNameFinderModel model = NameFinderME.train("en", "location", sampleStream, Collections.<String, Object>emptyMap(), 1, 0);

    BufferedOutputStream modelOut = null;
    try {
        modelOut = new BufferedOutputStream(new FileOutputStream(output));
        model.serialize(modelOut);
    } finally {
        if (modelOut != null)
            modelOut.close();
    }

Answer 1

我知道这在很久以前就被问到了，我也面临着一个类似的问题，分类设置了一个合适的截止点来解决我的问题。因此，如果你给一个截止值为1，它可能会有帮助(免责声明：-我还没有测试它)

如果你想保留一个默认的截止值(也就是5)，那么你必须至少训练5次才能让它识别。

Answer 2

这通常是因为在培训数据中的标记之后没有空格。例如,

<START:person>bob<END> 
will fail but 
<START:person> bob <END> 
will succeed.

如果这不能解决问题，则发布大量培训数据。另外，确保培训文件中的每个句子都在一行上。换句话说，所有句子都不应该包含\n，必须以\n结尾