/**
* Copyright (c) 2009/09-2012/08, Regents of the University of Colorado
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions are met:
*
* 1. Redistributions of source code must retain the above copyright notice, this
* list of conditions and the following disclaimer.
* 2. 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.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
* WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
* DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
* ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
* (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
* LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
* ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
* SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
/**
* Copyright 2012/09-2013/04, 2013/11-Present, University of Massachusetts Amherst
* Copyright 2013/05-2013/10, IPSoft Inc.
*
* Licensed 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 com.clearnlp.classification.train;
import java.io.PrintStream;
import java.util.Arrays;
import java.util.Collection;
import java.util.List;
import java.util.Map;
import com.carrotsearch.hppc.ObjectIntOpenHashMap;
import com.carrotsearch.hppc.cursors.ObjectCursor;
import com.clearnlp.classification.instance.StringInstance;
import com.clearnlp.classification.model.StringModel;
import com.clearnlp.classification.vector.SparseFeatureVector;
import com.clearnlp.classification.vector.StringFeatureVector;
import com.google.common.collect.Lists;
import com.google.common.collect.Maps;
/**
* Train space containing string vectors.
* @since 1.0.0
* @author Jinho D. Choi ({@code jdchoi77@gmail.com})
*/
public class StringTrainSpace extends AbstractTrainSpace
{
/** Casted from {@likn AbstractTrainSpace#m_model}. */
private StringModel s_model;
/** The label count cutoff (exclusive). */
private int l_cutoff;
/** The feature count cutoff (exclusive). */
private int f_cutoff;
/** The list of all training instances. */
private List<StringInstance> s_instances;
/** The map between labels and their counts. */
private ObjectIntOpenHashMap<String> m_labels;
/** The map between features and their counts. */
private Map<String,ObjectIntOpenHashMap<String>> m_features;
/**
* Constructs a train space containing string vectors.
* @param hasWeight {@code true} if features are assigned with different weights.
* @param labelCutoff the label count cutoff (exclusive).
* @param featureCutoff the feature count cutoff (exclusive).
*/
public StringTrainSpace(boolean hasWeight, int labelCutoff, int featureCutoff)
{
super(new StringModel(), hasWeight);
s_model = (StringModel)m_model;
l_cutoff = labelCutoff;
f_cutoff = featureCutoff;
s_instances = Lists.newArrayList();
m_labels = new ObjectIntOpenHashMap<String>();
m_features = Maps.newHashMap();
}
public void printInstances(PrintStream fout)
{
int i, size = s_instances.size();
String[] instances = new String[size];
StringInstance p;
for (i=0; i<size; i++)
{
p = s_instances.get(i);
instances[i] = p.getLabel() + DELIM_COL + p.getFeatureVector().toString();
}
Arrays.sort(instances);
for (String instance : instances)
fout.println(instance);
}
/** Adds a training instance to this space. */
public void addInstance(StringInstance instance)
{
addLexica(instance);
s_instances.add(instance);
}
public void addInstances(Collection<StringInstance> instances)
{
for (StringInstance instance : instances)
addInstance(instance);
}
/**
* Adds a training instance to this space.
* @param line {@code <label>}{@link AbstractTrainSpace#DELIM_COL}{@link StringFeatureVector#toString()}.
*/
public void addInstance(String line)
{
addInstance(toInstance(line, b_weight));
}
public void appendSpace(StringTrainSpace space)
{
appendSpaceLabels(space);
appendSpaceFeatures(space);
appendSpaceInstances(space);
}
private void appendSpaceLabels(StringTrainSpace space)
{
ObjectIntOpenHashMap<String> mLabels = space.m_labels;
String label;
for (ObjectCursor<String> cur : mLabels.keys())
{
label = cur.value;
m_labels.put(label, m_labels.get(label) + mLabels.get(label));
}
}
private void appendSpaceFeatures(StringTrainSpace space)
{
Map<String,ObjectIntOpenHashMap<String>> mFeatures = space.m_features;
ObjectIntOpenHashMap<String> tMap, sMap;
String value;
for (String type : mFeatures.keySet())
{
sMap = mFeatures.get(type);
if (m_features.containsKey(type))
{
tMap = m_features.get(type);
for (ObjectCursor<String> cur : sMap.keys())
{
value = cur.value;
tMap.put(value, tMap.get(value) + sMap.get(value));
}
}
else
m_features.put(type, sMap);
}
}
private void appendSpaceInstances(StringTrainSpace space)
{
s_instances.addAll(space.s_instances);
}
public void clear()
{
s_instances.clear();
m_labels .clear();
m_features .clear();
}
private void addLexica(StringInstance instance)
{
addLexicaLabel(instance.getLabel());
addLexicaFeatures(instance.getFeatureVector());
}
private void addLexicaLabel(String label)
{
m_labels.put(label, m_labels.get(label)+1);
}
private void addLexicaFeatures(StringFeatureVector vector)
{
ObjectIntOpenHashMap<String> map;
int i, size = vector.size();
String type, value;
for (i=0; i<size; i++)
{
type = vector.getType(i);
value = vector.getValue(i);
if (m_features.containsKey(type))
{
map = m_features.get(type);
map.put(value, map.get(value)+1);
}
else
{
map = new ObjectIntOpenHashMap<String>();
map.put(value, 1);
m_features.put(type, map);
}
}
}
@Override
public void build(boolean clearInstances)
{
LOG.info("Building:\n");
initModelMaps();
StringInstance instance;
int y, i, size = s_instances.size();
SparseFeatureVector x;
for (i=0; i<size; i++)
{
instance = s_instances.get(i);
if ((y = s_model.getLabelIndex(instance.getLabel())) < 0)
continue;
x = s_model.toSparseFeatureVector(instance.getFeatureVector());
a_ys.add(y);
a_xs.add(x.getIndices());
if (b_weight) a_vs.add(x.getWeights());
}
a_ys.trimToSize();
a_xs.trimToSize();
if (b_weight) a_vs.trimToSize();
LOG.info("- # of labels : "+s_model.getLabelSize()+"\n");
LOG.info("- # of features : "+s_model.getFeatureSize()+"\n");
LOG.info("- # of instances: "+a_ys.size()+"\n");
if (clearInstances) s_instances.clear();
}
@Override
public void build()
{
build(true);
}
/** Called by {@link StringTrainSpace#build()}. */
private void initModelMaps()
{
// initialize label map
String label;
for (ObjectCursor<String> cur : m_labels.keys())
{
label = cur.value;
if (m_labels.get(label) > l_cutoff)
s_model.addLabel(label);
}
s_model.initLabelArray();
// initialize feature map
ObjectIntOpenHashMap<String> map;
String value;
for (String type : m_features.keySet())
{
map = m_features.get(type);
for (ObjectCursor<String> cur : map.keys())
{
value = cur.value;
if (map.get(value) > f_cutoff)
s_model.addFeature(type, value);
}
}
/* for (String label : UTHppc.getSortedKeys(m_labels))
{
if (m_labels.get(label) > l_cutoff)
s_model.addLabel(label);
}
s_model.initLabelArray();
// initialize feature map
List<String> types = new ArrayList<String>(m_features.keySet());
ObjectIntOpenHashMap<String> map;
Collections.sort(types);
for (String type : types)
{
map = m_features.get(type);
for (String value : UTHppc.getSortedKeys(map))
{
if (map.get(value) > f_cutoff)
s_model.addFeature(type, value);
}
}*/
}
/** Pair of label and feature vector. */
static public StringInstance toInstance(String line, boolean hasWeight)
{
String[] tmp = line.split(DELIM_COL);
String label = tmp[0];
StringFeatureVector vector = new StringFeatureVector(hasWeight);
int i, size = tmp.length;
for (i=1; i<size; i++)
vector.addFeature(tmp[i]);
return new StringInstance(label, vector);
}
}