/*
* Licensed to the Apache Software Foundation (ASF) under one or more
* contributor license agreements. See the NOTICE file distributed with
* this work for additional information regarding copyright ownership.
* The ASF licenses this file to You under the Apache License, Version 2.0
* (the "License"); you may not use this file except in compliance with
* the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package opennlp.tools.namefind;
import java.util.ArrayList;
import java.util.Iterator;
import java.util.List;
import java.util.Map;
import opennlp.tools.ml.model.Event;
import opennlp.tools.util.ObjectStream;
import opennlp.tools.util.SequenceCodec;
import opennlp.tools.util.Span;
import opennlp.tools.util.featuregen.AdditionalContextFeatureGenerator;
import opennlp.tools.util.featuregen.WindowFeatureGenerator;
/**
* Class for creating an event stream out of data files for training an name
* finder.
*/
public class NameFinderEventStream extends opennlp.tools.util.AbstractEventStream<NameSample> {
private NameContextGenerator contextGenerator;
private AdditionalContextFeatureGenerator additionalContextFeatureGenerator = new AdditionalContextFeatureGenerator();
private String type;
private SequenceCodec<String> codec;
/**
* Creates a new name finder event stream using the specified data stream and context generator.
* @param dataStream The data stream of events.
* @param type null or overrides the type parameter in the provided samples
* @param contextGenerator The context generator used to generate features for the event stream.
*/
public NameFinderEventStream(ObjectStream<NameSample> dataStream, String type, NameContextGenerator contextGenerator, SequenceCodec codec) {
super(dataStream);
this.codec = codec;
if (codec == null) {
this.codec = new BioCodec();
}
this.contextGenerator = contextGenerator;
this.contextGenerator.addFeatureGenerator(new WindowFeatureGenerator(additionalContextFeatureGenerator, 8, 8));
if (type != null)
this.type = type;
else
this.type = "default";
}
public NameFinderEventStream(ObjectStream<NameSample> dataStream) {
this(dataStream, null, new DefaultNameContextGenerator(), null);
}
/**
* Generates the name tag outcomes (start, continue, other) for each token in a sentence
* with the specified length using the specified name spans.
* @param names Token spans for each of the names.
* @param type null or overrides the type parameter in the provided samples
* @param length The length of the sentence.
* @return An array of start, continue, other outcomes based on the specified names and sentence length.
*
* @deprecated use the BioCodec implementation of the SequenceValidator instead!
*/
@Deprecated
public static String[] generateOutcomes(Span[] names, String type, int length) {
String[] outcomes = new String[length];
for (int i = 0; i < outcomes.length; i++) {
outcomes[i] = NameFinderME.OTHER;
}
for (Span name : names) {
if (name.getType() == null) {
outcomes[name.getStart()] = type + "-" + NameFinderME.START;
}
else {
outcomes[name.getStart()] = name.getType() + "-" + NameFinderME.START;
}
// now iterate from begin + 1 till end
for (int i = name.getStart() + 1; i < name.getEnd(); i++) {
if (name.getType() == null) {
outcomes[i] = type + "-" + NameFinderME.CONTINUE;
}
else {
outcomes[i] = name.getType() + "-" + NameFinderME.CONTINUE;
}
}
}
return outcomes;
}
public static List<Event> generateEvents(String[] sentence, String[] outcomes, NameContextGenerator cg) {
List<Event> events = new ArrayList<Event>(outcomes.length);
for (int i = 0; i < outcomes.length; i++) {
events.add(new Event(outcomes[i], cg.getContext(i, sentence, outcomes,null)));
}
cg.updateAdaptiveData(sentence, outcomes);
return events;
}
@Override
protected Iterator<Event> createEvents(NameSample sample) {
if (sample.isClearAdaptiveDataSet()) {
contextGenerator.clearAdaptiveData();
}
String outcomes[] = codec.encode(sample.getNames(), sample.getSentence().length);
// String outcomes[] = generateOutcomes(sample.getNames(), type, sample.getSentence().length);
additionalContextFeatureGenerator.setCurrentContext(sample.getAdditionalContext());
String[] tokens = new String[sample.getSentence().length];
for (int i = 0; i < sample.getSentence().length; i++) {
tokens[i] = sample.getSentence()[i];
}
return generateEvents(tokens, outcomes, contextGenerator).iterator();
}
/**
* Generated previous decision features for each token based on contents of the specified map.
* @param tokens The token for which the context is generated.
* @param prevMap A mapping of tokens to their previous decisions.
* @return An additional context array with features for each token.
*/
public static String[][] additionalContext(String[] tokens, Map<String, String> prevMap) {
String[][] ac = new String[tokens.length][1];
for (int ti=0;ti<tokens.length;ti++) {
String pt = prevMap.get(tokens[ti]);
ac[ti][0]="pd="+pt;
}
return ac;
}
}