banner.tagging

Class CRFTagger

java.lang.Object

banner.tagging.CRFTagger

All Implemented Interfaces:

Tagger

Direct Known Subclasses:

NBestCRFTagger

public class CRFTagger
extends Object
implements Tagger

Field Summary

Fields

Modifier and Type	Field and Description
protected cc.mallet.fst.CRF	model

Constructor Summary

Constructors

Modifier	Constructor and Description
protected	CRFTagger(cc.mallet.fst.CRF model, FeatureSet featureSet, int order)

Method Summary

Modifier and Type	Method and Description
void	describe(String fileName)
Set<String>	getFeatureNames()
List<List<String>>	getFeatureRepresentation(Sentence sentence)
protected cc.mallet.types.Instance	getInstance(Sentence sentence)
Map<String,Double>	getMaxWeights()
Map<String,Double>	getMinWeights()
int	getOrder()
protected static List<String>	getTagList(cc.mallet.types.Sequence<Object> tags)
static CRFTagger	load(InputStream f, dragon.nlp.tool.Lemmatiser lemmatiser, dragon.nlp.tool.Tagger posTagger, Tagger preTagger) Loads a CRFTagger from the specified file.
void	tag(Sentence sentence) Add Mentions to the Sentence.
static CRFTagger	train(Set<Sentence> sentences, int order, TagFormat format, FeatureSet featureSet) Trains and returns a CRFTagger on the specified Sentence s.
void	write(File f) Serializes and writes this CRFTagger to the specified file

Methods inherited from class java.lang.Object

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

Field Detail

model

protected cc.mallet.fst.CRF model

Constructor Detail

CRFTagger

protected CRFTagger(cc.mallet.fst.CRF model,
                    FeatureSet featureSet,
                    int order)

Method Detail

load

public static CRFTagger load(InputStream f,
                             dragon.nlp.tool.Lemmatiser lemmatiser,
                             dragon.nlp.tool.Tagger posTagger,
                             Tagger preTagger)
                      throws IOException

Loads a CRFTagger from the specified file. As the lemmatiser and part-of-speech tagger both require data, these cannot be written to disk and must be passed in new.

Parameters:

f - The file to load the CRFTagger from, as written by the write() method.

lemmatiser - The Lemmatiser to use

posTagger - The part-of-speech Tagger to use

Returns:

A new instance of the CRFTagger contained in the specified file

Throws:

IOException

train

public static CRFTagger train(Set<Sentence> sentences,
                              int order,
                              TagFormat format,
                              FeatureSet featureSet)

Trains and returns a CRFTagger on the specified Sentence s. This method may take hours or even days to complete. When training, you will likely need to increase the amount of memory used by the Java virtual machine (try adding "-Xms1024m" to the command line).

Parameters:

sentences - The Sentences to train the tagger on

order - The CRF order to use

format - The TagFormat to use

Returns:

A trained CRFTagger; ready to tag unseen sentences or be output to disk

write

public void write(File f)

Serializes and writes this CRFTagger to the specified file

Parameters:

f - The file to write this CRFTagger to

tag

public void tag(Sentence sentence)

Description copied from interface: Tagger

Add Mentions to the Sentence. The Sentence must have been tokenized previously.

Specified by:

tag in interface Tagger

Parameters:

sentence - The sentence to which Mentions should be added

getInstance

protected cc.mallet.types.Instance getInstance(Sentence sentence)

getTagList

protected static List<String> getTagList(cc.mallet.types.Sequence<Object> tags)

getOrder

public int getOrder()

Returns:

The CRF order used by this tagger. Order 1 means that the last state is used and order 2 means that the last 2 states are used.

getFeatureNames

public Set<String> getFeatureNames()

getFeatureRepresentation

public List<List<String>> getFeatureRepresentation(Sentence sentence)

describe

public void describe(String fileName)
              throws IOException

Throws:

IOException

getMaxWeights

public Map<String,Double> getMaxWeights()

getMinWeights

public Map<String,Double> getMinWeights()