001/** 
002 * Copyright (c) 2010, Regents of the University of Colorado 
003 * All rights reserved.
004 * 
005 * Redistribution and use in source and binary forms, with or without
006 * modification, are permitted provided that the following conditions are met:
007 * 
008 * Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer. 
009 * Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution. 
010 * Neither the name of the University of Colorado at Boulder nor the names of its contributors may be used to endorse or promote products derived from this software without specific prior written permission. 
011 * 
012 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
013 * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
014 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
015 * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
016 * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
017 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
018 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
019 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
020 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
021 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
022 * POSSIBILITY OF SUCH DAMAGE. 
023 */
024package org.cleartk.timeml.event;
025
026import java.util.ArrayList;
027import java.util.List;
028
029import org.apache.uima.UimaContext;
030import org.apache.uima.analysis_engine.AnalysisEngineProcessException;
031import org.apache.uima.jcas.JCas;
032import org.apache.uima.resource.ResourceInitializationException;
033import org.cleartk.ml.CleartkAnnotator;
034import org.cleartk.ml.Feature;
035import org.cleartk.ml.Instance;
036import org.cleartk.ml.feature.extractor.CleartkExtractor;
037import org.cleartk.ml.feature.extractor.FeatureExtractor1;
038import org.cleartk.timeml.type.Event;
039import org.cleartk.token.type.Sentence;
040import org.cleartk.token.type.Token;
041import org.apache.uima.fit.util.JCasUtil;
042
043import com.google.common.collect.Lists;
044
045/**
046 * <br>
047 * Copyright (c) 2010, Regents of the University of Colorado <br>
048 * All rights reserved.
049 * 
050 * Base class for annotators of TimeML EVENT attributes.
051 * 
052 * @author Steven Bethard
053 */
054public abstract class EventAttributeAnnotator<OUTCOME_TYPE> extends CleartkAnnotator<OUTCOME_TYPE> {
055
056  /**
057   * The list of feature extractors that will be applied directly to the Event annotation.
058   * 
059   * Subclasses should override {@link #initialize(org.apache.uima.UimaContext)} to fill this list.
060   */
061  protected List<FeatureExtractor1<Event>> eventFeatureExtractors;
062
063  /**
064   * The list of feature extractors that will be applied to the Event annotation, with a Sentence
065   * window.
066   * 
067   * Subclasses should override {@link #initialize(org.apache.uima.UimaContext)} to fill this list.
068   */
069  protected List<CleartkExtractor<Event, Token>> contextExtractors;
070
071  /**
072   * The attribute value that should be considered as a default, e.g. "NONE". When the attribute
073   * value is null in the CAS, this value will be used instead. Additionally, when the classifier
074   * produces this value, no attribute will be added.
075   * 
076   * @return The default attribute value.
077   */
078  protected abstract OUTCOME_TYPE getDefaultValue();
079
080  /**
081   * Get the attribute value from the Event annotation. Typically this will be by calling something
082   * like {@link Event#getTense()}.
083   * 
084   * If this method returns null, {@link #getDefaultValue()} will be called to produce an
085   * appropriate attribute value.
086   * 
087   * @param event
088   *          The Event annotation whose attribute is to be retrieved.
089   * @return The selected attribute value.
090   */
091  protected abstract OUTCOME_TYPE getAttribute(Event event);
092
093  /**
094   * Set the attribute value on the Event annotation. Typically this will be by calling something
095   * like {@link Event#setTense(String)}.
096   * 
097   * This method will not be called if the value is equal to {@link #getDefaultValue()}.
098   * 
099   * @param event
100   *          The Event annotation whose attribute is to be set.
101   * @param value
102   *          The new attribute value.
103   */
104  protected abstract void setAttribute(Event event, OUTCOME_TYPE value);
105
106  @Override
107  public void initialize(UimaContext context) throws ResourceInitializationException {
108    super.initialize(context);
109    this.eventFeatureExtractors = Lists.newArrayList();
110    this.contextExtractors = Lists.newArrayList();
111  }
112
113  @Override
114  public void process(JCas jCas) throws AnalysisEngineProcessException {
115    for (Sentence sentence : JCasUtil.select(jCas, Sentence.class)) {
116      for (Event event : JCasUtil.selectCovered(jCas, Event.class, sentence)) {
117
118        // assemble features
119        List<Feature> features = new ArrayList<Feature>();
120        for (FeatureExtractor1<Event> extractor : this.eventFeatureExtractors) {
121          features.addAll(extractor.extract(jCas, event));
122        }
123        for (CleartkExtractor<Event, Token> extractor : this.contextExtractors) {
124          features.addAll(extractor.extractWithin(jCas, event, sentence));
125        }
126
127        // if training, determine the attribute value and write the
128        // instance
129        if (this.isTraining()) {
130          OUTCOME_TYPE attribute = this.getAttribute(event);
131          if (attribute == null) {
132            attribute = this.getDefaultValue();
133          }
134          Instance<OUTCOME_TYPE> instance = new Instance<OUTCOME_TYPE>();
135          instance.addAll(features);
136          instance.setOutcome(attribute);
137          this.dataWriter.write(instance);
138        }
139
140        // if predicting, propose an attribute value and modify the
141        // event annotation
142        else {
143          OUTCOME_TYPE label = this.classifier.classify(features);
144          this.setAttribute(event, label);
145        }
146      }
147    }
148  }
149}