no.priv.garshol.duke.comparators
Class JaroWinklerTokenized

java.lang.Object
  no.priv.garshol.duke.comparators.JaroWinklerTokenized

All Implemented Interfaces:

Comparator

public class JaroWinklerTokenized
extends Object
implements Comparator

A tokenized approach to string similarity, based on Jaccard equivalence and the Jaro-Winkler metric. FIXME: Do we actually need this, or is DiceCoefficientComparator better? I guess Dice probably is better. However, the code for not allowing same token to be matched twice is unique to this comparator. Should we reuse in Dice, or just support more methods than just Dice?

Constructor Summary

JaroWinklerTokenized()

Method Summary

double	compare(String s1, String s2)
boolean	isTokenized() Returns true if the comparator breaks string values up into tokens when comparing.

Methods inherited from class java.lang.Object

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

Constructor Detail

JaroWinklerTokenized

public JaroWinklerTokenized()

Method Detail

isTokenized

public boolean isTokenized()

Description copied from interface: Comparator

Returns true if the comparator breaks string values up into tokens when comparing. Necessary because this impacts indexing of values.

Specified by:

isTokenized in interface Comparator

compare

public double compare(String s1,
                      String s2)

Specified by:

compare in interface Comparator