com.swabunga.spell.engine
Class EditDistance

java.lang.Object
  com.swabunga.spell.engine.EditDistance

public class EditDistance

extends java.lang.Object

This class is based on Levenshtein Distance algorithms, and it calculates how similar two words are. If the words are identitical, then the distance is 0. The more that the words have in common, the lower the distance value. The distance value is based on how many operations it takes to get from one word to the other. Possible operations are swapping characters, adding a character, deleting a character, and substituting a character. The resulting distance is the sum of these operations weighted by their cost, which can be set in the Configuration object. When there are multiple ways to convert one word into the other, the lowest cost distance is returned.
Another way to think about this: what are the cheapest operations that would have to be done on the "original" word to end up with the "similar" word? Each operation has a cost, and these are added up to get the distance.

See Also:

Configuration.COST_REMOVE_CHAR, Configuration.COST_INSERT_CHAR, Configuration.COST_SUBST_CHARS, Configuration.COST_SWAP_CHARS

Field Summary

static Configuration

config JMH Again, there is no need to have a global class matrix variable in this class.

Constructor Summary

EditDistance()

Method Summary

static int	getDistance(java.lang.String word, java.lang.String similar)
static void	main(java.lang.String[] args)

Methods inherited from class java.lang.Object

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

Field Detail

config

public static Configuration config

JMH Again, there is no need to have a global class matrix variable in this class. I have removed it and made the getDistance static final DMV: I refactored this method to make it more efficient, more readable, and simpler. I also fixed a bug with how the distance was being calculated. You could get wrong distances if you compared ("abc" to "ab") depending on what you had setup your COST_REMOVE_CHAR and EDIT_INSERTION_COST values to - that is now fixed. WRS: I added a distance for case comparison, so a misspelling of "i" would be closer to "I" than to "a".

Constructor Detail

EditDistance

public EditDistance()

Method Detail

getDistance

public static final int getDistance(java.lang.String word,
                                    java.lang.String similar)

main

public static void main(java.lang.String[] args)
                 throws java.lang.Exception

Throws:

java.lang.Exception