/*
* LensKit, an open source recommender systems toolkit.
* Copyright 2010-2014 LensKit Contributors. See CONTRIBUTORS.md.
* Work on LensKit has been funded by the National Science Foundation under
* grants IIS 05-34939, 08-08692, 08-12148, and 10-17697.
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU Lesser General Public License as
* published by the Free Software Foundation; either version 2.1 of the
* License, or (at your option) any later version.
*
* This program is distributed in the hope that it will be useful, but WITHOUT
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
* FOR A PARTICULAR PURPOSE. See the GNU General Public License for more
* details.
*
* You should have received a copy of the GNU General Public License along with
* this program; if not, write to the Free Software Foundation, Inc., 51
* Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
*/
package org.grouplens.lenskit.knn.item.model;
import com.google.common.base.Stopwatch;
import it.unimi.dsi.fastutil.longs.*;
import org.grouplens.lenskit.core.Transient;
import org.grouplens.lenskit.knn.item.ItemSimilarity;
import org.grouplens.lenskit.knn.item.ItemSimilarityThreshold;
import org.grouplens.lenskit.knn.item.ModelSize;
import org.grouplens.lenskit.scored.ScoredId;
import org.grouplens.lenskit.transform.threshold.Threshold;
import org.grouplens.lenskit.util.ScoredItemAccumulator;
import org.grouplens.lenskit.util.TopNScoredItemAccumulator;
import org.grouplens.lenskit.util.UnlimitedScoredItemAccumulator;
import org.grouplens.lenskit.vectors.SparseVector;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import javax.annotation.concurrent.NotThreadSafe;
import javax.inject.Inject;
import javax.inject.Provider;
import java.util.List;
import java.util.concurrent.TimeUnit;
/**
* Build an item-item CF model from rating data.
* This builder takes a very simple approach. It does not allow for vector
* normalization and truncates on the fly.
*
* @author <a href="http://www.grouplens.org">GroupLens Research</a>
*/
@NotThreadSafe
public class ItemItemModelBuilder implements Provider<ItemItemModel> {
private static final Logger logger = LoggerFactory.getLogger(ItemItemModelBuilder.class);
private final ItemSimilarity itemSimilarity;
private final ItemItemBuildContext buildContext;
private final Threshold threshold;
private final NeighborIterationStrategy neighborStrategy;
private final int modelSize;
@Inject
public ItemItemModelBuilder(@Transient ItemSimilarity similarity,
@Transient ItemItemBuildContext context,
@Transient @ItemSimilarityThreshold Threshold thresh,
@Transient NeighborIterationStrategy nbrStrat,
@ModelSize int size) {
itemSimilarity = similarity;
buildContext = context;
threshold = thresh;
neighborStrategy = nbrStrat;
modelSize = size;
}
@Override
public SimilarityMatrixModel get() {
logger.debug("building item-item model");
logger.debug("using similarity function {}", itemSimilarity);
logger.debug("similarity function is {}",
itemSimilarity.isSparse() ? "sparse" : "non-sparse");
logger.debug("similarity function is {}",
itemSimilarity.isSymmetric() ? "symmetric" : "non-symmetric");
LongSortedSet allItems = buildContext.getItems();
Long2ObjectMap<ScoredItemAccumulator> rows = makeAccumulators(allItems);
final int nitems = allItems.size();
LongIterator outer = allItems.iterator();
Stopwatch timer = Stopwatch.createStarted();
int ndone = 0;
while (outer.hasNext()) {
ndone += 1;
final long itemId1 = outer.nextLong();
if (logger.isTraceEnabled()) {
logger.trace("computing similarities for item {} ({} of {})",
itemId1, ndone, nitems);
}
SparseVector vec1 = buildContext.itemVector(itemId1);
LongIterator itemIter = neighborStrategy.neighborIterator(buildContext, itemId1,
itemSimilarity.isSymmetric());
ScoredItemAccumulator row = rows.get(itemId1);
while (itemIter.hasNext()) {
long itemId2 = itemIter.nextLong();
if (itemId1 != itemId2) {
SparseVector vec2 = buildContext.itemVector(itemId2);
double sim = itemSimilarity.similarity(itemId1, vec1, itemId2, vec2);
if (threshold.retain(sim)) {
row.put(itemId2, sim);
if (itemSimilarity.isSymmetric()) {
rows.get(itemId2).put(itemId1, sim);
}
}
}
}
if (logger.isDebugEnabled() && ndone % 100 == 0) {
logger.debug("computed {} of {} model rows ({}s/row)",
ndone, nitems,
String.format("%.3f", timer.elapsed(TimeUnit.MILLISECONDS) * 0.001 / ndone));
}
}
timer.stop();
logger.info("built model for {} items in {}", ndone, timer);
return new SimilarityMatrixModel(finishRows(rows));
}
private Long2ObjectMap<ScoredItemAccumulator> makeAccumulators(LongSet items) {
Long2ObjectMap<ScoredItemAccumulator> rows = new Long2ObjectOpenHashMap<ScoredItemAccumulator>(items.size());
LongIterator iter = items.iterator();
while (iter.hasNext()) {
long item = iter.nextLong();
ScoredItemAccumulator accum;
if (modelSize == 0) {
accum = new UnlimitedScoredItemAccumulator();
} else {
accum = new TopNScoredItemAccumulator(modelSize);
}
rows.put(item, accum);
}
return rows;
}
private Long2ObjectMap<List<ScoredId>> finishRows(Long2ObjectMap<ScoredItemAccumulator> rows) {
Long2ObjectMap<List<ScoredId>> results = new Long2ObjectOpenHashMap<List<ScoredId>>(rows.size());
for (Long2ObjectMap.Entry<ScoredItemAccumulator> e: rows.long2ObjectEntrySet()) {
results.put(e.getLongKey(), e.getValue().finish());
}
return results;
}
}