ws.palladian.extraction.entity.tagger

Class OpenNlpNer

java.lang.Object

ws.palladian.extraction.entity.NamedEntityRecognizer

ws.palladian.extraction.entity.TrainableNamedEntityRecognizer

ws.palladian.extraction.entity.tagger.OpenNlpNer

All Implemented Interfaces:

ClassifyingTagger, Tagger

public class OpenNlpNer
extends TrainableNamedEntityRecognizer
implements ClassifyingTagger

This class wraps the OpenNLP Named Entity Recognizer which uses a maximum entropy approach.

The following models exist already for this recognizer:

• Date
• Location
• Money
• Organization
• Percentage
• Person
• Time

Author:

David Urbansky, Philipp Katz

See Also:

Apache OpenNLP Developer Documentation: Name Finder, OpenNLP Tools Models

Constructor Summary

Constructors

Constructor and Description
OpenNlpNer(opennlp.tools.tokenize.Tokenizer tokenizer, opennlp.tools.sentdetect.SentenceDetector sentenceDetector) Create a new OpenNlpNer.

Method Summary

Modifier and Type	Method and Description
List<ClassifiedAnnotation>	getAnnotations(String inputText)
String	getModelFileEnding()
String	getName()
boolean	loadModel(String configModelFilePath) Load the models for the entity recognizer.
boolean	oneModelPerConcept()
boolean	setsModelFileEndingAutomatically()
boolean	train(String trainingFilePath, String modelFilePath)

Methods inherited from class ws.palladian.extraction.entity.TrainableNamedEntityRecognizer

getModelFileEndingIfNotSetAutomatically, train, train

Methods inherited from class ws.palladian.extraction.entity.NamedEntityRecognizer

evaluate, evaluate, evaluate, evaluate, getTaggingFormat, setTaggingFormat, tag

Methods inherited from class java.lang.Object

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

Constructor Detail

OpenNlpNer

public OpenNlpNer(opennlp.tools.tokenize.Tokenizer tokenizer,
                  opennlp.tools.sentdetect.SentenceDetector sentenceDetector)

Create a new OpenNlpNer. The NER requires a tokenizer and a sentence detector in order to work properly.

Parameters:

tokenizer - The tokenizer to use, not null.

sentenceDetector - The sentence detector to use, not null.

Method Detail

loadModel

public boolean loadModel(String configModelFilePath)

Load the models for the entity recognizer. All files in the specified directory with the file name extension ".bin" are considered as OpenNLP TokenNameFinderModels.

Specified by:

loadModel in class TrainableNamedEntityRecognizer

Parameters:

configModelFilePath - The path to the folder where the models lie.

getAnnotations

public List<ClassifiedAnnotation> getAnnotations(String inputText)

Specified by:

getAnnotations in interface ClassifyingTagger

Specified by:

getAnnotations in interface Tagger

Specified by:

getAnnotations in class NamedEntityRecognizer

getModelFileEnding

public String getModelFileEnding()

Specified by:

getModelFileEnding in class TrainableNamedEntityRecognizer

setsModelFileEndingAutomatically

public boolean setsModelFileEndingAutomatically()

Specified by:

setsModelFileEndingAutomatically in class TrainableNamedEntityRecognizer

oneModelPerConcept

public boolean oneModelPerConcept()

Overrides:

oneModelPerConcept in class TrainableNamedEntityRecognizer

train

public boolean train(String trainingFilePath,
                     String modelFilePath)

Specified by:

train in class TrainableNamedEntityRecognizer

getName

public String getName()

Specified by:

getName in class NamedEntityRecognizer