edu.ucla.sspace.matrix

Class RowComparator

java.lang.Object

edu.ucla.sspace.matrix.RowComparator

public class RowComparator
extends Object

A utility class for finding the k most-similar words to a provided row in a Matrix. The comparisons required for generating the list maybe be run in parallel by configuring an instance of this class to use multiple threads.

All instances of this class are thread-safe.

Author:

David Jurgens

Constructor Summary

Constructors

Constructor and Description
RowComparator() Creates this WordComparator with as many threads as processors.
RowComparator(int numProcs) Creates this WordComparator with the specified number of threads.

Method Summary

Methods

Modifier and Type	Method and Description
SortedMultiMap<Double,Integer>	getMostSimilar(Matrix m, int row, int kNearestRows, Similarity.SimType similarityType) Compares the specified row to all other rows, returning the k-nearest rows according to the similarity metric.
SortedMultiMap<Double,Integer>	getMostSimilar(Matrix m, int row, int kNearestRows, SimilarityFunction simFunction) Compares the specified row to all other rows, returning the k-nearest rows according to the similarity metric.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

RowComparator

public RowComparator()

Creates this WordComparator with as many threads as processors.

RowComparator

public RowComparator(int numProcs)

Creates this WordComparator with the specified number of threads.

Method Detail

getMostSimilar

public SortedMultiMap<Double,Integer> getMostSimilar(Matrix m,
                                            int row,
                                            int kNearestRows,
                                            Similarity.SimType similarityType)

Compares the specified row to all other rows, returning the k-nearest rows according to the similarity metric.

Parameters:

m - The Matrix containing data points to be compared

row - The current row in m to be compared against all other rows

kNearestRows - The number of most similar rows to retain

similarityType - The similarity method to use when comparing rows

Returns:

a mapping from the similarity to the kNearestRows most similar rows

getMostSimilar

public SortedMultiMap<Double,Integer> getMostSimilar(Matrix m,
                                            int row,
                                            int kNearestRows,
                                            SimilarityFunction simFunction)

Compares the specified row to all other rows, returning the k-nearest rows according to the similarity metric.

Parameters:

m - The Matrix containing data points to be compared

row - The current row in m to be compared against all other rows

kNearestRows - The number of most similar rows to retain

simFunction - The SimilarityFunction to use when comparing rows

Returns:

a mapping from the similarity to the k most similar rows