public KnnSimsCommand(
File sourceFile,
File destinationFile,
Charset charset,
SingleEnumerating indexDeligate,
int k)
throws IOException {
super(sourceFile, destinationFile, charset, EnumeratingDelegates.toPair(indexDeligate));
super.setComparator(Comparators.fallback(classComparator, nearnessComparator));
setK(k);
}
public ObjectMergeTask() {
setComparator(Comparators.<T>naturalOrderIfPossible());
}
public ObjectMergeTask(ObjectSource<T> sourceA, ObjectSource<T> sourceB, ObjectSink<T> sink) {
setSourceA(sourceA);
setSourceB(sourceB);
setSink(sink);
setComparator(Comparators.<T>naturalOrderIfPossible());
}
@Override
public Comparator<Weighted<TokenPair>> getComparator() {
return isReverse() ? Comparators.reverse(getCombinedComparator()) : getCombinedComparator();
}
public Comparator<Weighted<TokenPair>> getCombinedComparator() {
return Comparators.fallback(getClassComparator(), getNearnessComparator());
}
/** * Task that read in a file and produces the k-nearest-neighbors for each base entry. Assumes the * file is composed of entry, entry, weight triples that are delimited by tabs. * * @author Hamish I A Morgan <[email protected]> */ @Parameters(commandDescription = "Perform k-nearest-neighbours on a similarity file.") public final class KnnSimsCommand extends SortEventsCommand { private static final Log LOG = LogFactory.getLog(KnnSimsCommand.class); @Parameter( names = {"-k"}, description = "The maximum number of neighbours to produce per word.") private int k = ExternalKnnSimsCommand.DEFAULT_K; private Comparator<Weighted<TokenPair>> classComparator = Weighted.recordOrder(TokenPair.firstIndexOrder()); private Comparator<Weighted<TokenPair>> nearnessComparator = Comparators.reverse(Weighted.<TokenPair>weightOrder()); public KnnSimsCommand( File sourceFile, File destinationFile, Charset charset, SingleEnumerating indexDeligate, int k) throws IOException { super(sourceFile, destinationFile, charset, EnumeratingDelegates.toPair(indexDeligate)); super.setComparator(Comparators.fallback(classComparator, nearnessComparator)); setK(k); } public KnnSimsCommand() { setK(100); } public Comparator<Weighted<TokenPair>> getCombinedComparator() { return Comparators.fallback(getClassComparator(), getNearnessComparator()); } @Override public Comparator<Weighted<TokenPair>> getComparator() { return isReverse() ? Comparators.reverse(getCombinedComparator()) : getCombinedComparator(); } @Override @Deprecated public void setComparator(Comparator<Weighted<TokenPair>> comparator) { throw new UnsupportedOperationException( "Class and nearness comparators should be set instead."); } public Comparator<Weighted<TokenPair>> getClassComparator() { return classComparator; } public void setClassComparator(Comparator<Weighted<TokenPair>> classComparator) { this.classComparator = classComparator; } public Comparator<Weighted<TokenPair>> getNearnessComparator() { return nearnessComparator; } public void setNearnessComparator(Comparator<Weighted<TokenPair>> nearnessComparator) { this.nearnessComparator = nearnessComparator; } public final int getK() { return k; } public final void setK(int k) { if (k < 1) throw new IllegalArgumentException("k < 1"); this.k = k; } private boolean first = false; @Override protected ObjectSource<Weighted<TokenPair>> openSource(File file) throws FileNotFoundException, IOException { final ObjectSource<Weighted<TokenPair>> src = first ? BybloIO.openSimsSource(file, getCharset(), getIndexDelegate()) : BybloIO.openNeighboursSource(file, getCharset(), getIndexDelegate()); first = true; return src; } @Override protected ObjectSink<Weighted<TokenPair>> openSink(File file) throws FileNotFoundException, IOException { return new KFirstReducingObjectSink<Weighted<TokenPair>>( new WeightSumReducerObjectSink<TokenPair>( BybloIO.openNeighboursSink(file, getCharset(), getIndexDelegate())), classComparator, k); } @Override protected Objects.ToStringHelper toStringHelper() { return super.toStringHelper().add("k", k); } }
