com.hankcs.hanlp.model.crf

Class CRFTagger

java.lang.Object

com.hankcs.hanlp.model.crf.CRFTagger

Direct Known Subclasses:

CRFNERecognizer, CRFPOSTagger, CRFSegmenter

public abstract class CRFTagger
extends Object

Author:

hankcs

Field Summary

Fields

Modifier and Type	Field and Description
protected LogLinearModel	model

Constructor Summary

Constructors

Constructor and Description
CRFTagger()
CRFTagger(String modelPath)

Method Summary

Modifier and Type	Method and Description
protected abstract void	convertCorpus(Sentence sentence, BufferedWriter bw)
void	convertCorpus(String pkuPath, String tsvPath)
void	dumpTemplate(String templatePath) 导出特征模板
protected abstract String	getDefaultFeatureTemplate()
String	getTemplate() 获取特征模板
void	train(String trainCorpusPath, String modelPath)
void	train(String trainFile, String modelFile, int maxitr, int freq, double eta, double C, int threadNum, int shrinkingSize, Encoder.Algorithm algorithm)
void	train(String templFile, String trainFile, String modelFile, int maxitr, int freq, double eta, double C, int threadNum, int shrinkingSize, Encoder.Algorithm algorithm) 训练

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

model

protected LogLinearModel model

Constructor Detail

CRFTagger

public CRFTagger()

CRFTagger

public CRFTagger(String modelPath)
          throws IOException

Throws:

IOException

Method Detail

train

public void train(String templFile,
                  String trainFile,
                  String modelFile,
                  int maxitr,
                  int freq,
                  double eta,
                  double C,
                  int threadNum,
                  int shrinkingSize,
                  Encoder.Algorithm algorithm)
           throws IOException

训练

Parameters:

templFile - 模板文件

trainFile - 训练文件

modelFile - 模型文件

maxitr - 最大迭代次数

freq - 特征最低频次

eta - 收敛阈值

C - cost-factor

threadNum - 线程数

shrinkingSize -

algorithm - 训练算法

Throws:

IOException

train

public void train(String trainCorpusPath,
                  String modelPath)
           throws IOException

Throws:

IOException

train

public void train(String trainFile,
                  String modelFile,
                  int maxitr,
                  int freq,
                  double eta,
                  double C,
                  int threadNum,
                  int shrinkingSize,
                  Encoder.Algorithm algorithm)
           throws IOException

Throws:

IOException

convertCorpus

protected abstract void convertCorpus(Sentence sentence,
                                      BufferedWriter bw)
                               throws IOException

Throws:

IOException

getDefaultFeatureTemplate

protected abstract String getDefaultFeatureTemplate()

convertCorpus

public void convertCorpus(String pkuPath,
                          String tsvPath)
                   throws IOException

Throws:

IOException

dumpTemplate

public void dumpTemplate(String templatePath)
                  throws IOException

导出特征模板

Parameters:

templatePath -

Throws:

IOException

getTemplate

public String getTemplate()

获取特征模板

Returns: