#stringmetric 
- Requirements: Scala 2.10
- Documentation: Scaladoc
- Issues: Enhancements, Questions, Bugs
- Versioning: Semantic Versioning v2.0
- Dice / Sorensen (Similarity metric)
- Double Metaphone (Queued phonetic metric and algorithm)
- Hamming (Similarity metric)
- Jaccard (Similarity metric)
- Jaro (Similarity metric)
- Jaro-Winkler (Similarity metric)
- Levenshtein (Similarity metric)
- Metaphone (Phonetic metric and algorithm)
- Monge-Elkan (Queued similarity metric)
- Match Rating Approach (Queued phonetic metric and algorithm)
- Needleman-Wunch (Queued similarity metric)
- N-Gram (Similarity metric)
- NYSIIS (Phonetic metric and algorithm)
- Overlap (Similarity metric)
- Ratcliff-Obershelp (Similarity metric)
- Refined NYSIIS (Phonetic metric and algorithm)
- Refined Soundex (Phonetic metric and algorithm)
- Tanimoto (Queued similarity metric)
- Tversky (Queued similarity metric)
- Smith-Waterman (Queued similarity metric)
- Soundex (Phonetic metric and algorithm)
- Weighted Levenshtein (Similarity metric)
SBT:
libraryDependencies = "com.rockymadden.stringmetric" %% "stringmetric-core" % "0.27.4"Gradle:
compile 'com.rockymadden.stringmetric:stringmetric-core_2.10:0.27.4'Maven:
<dependency>
<groupId>com.rockymadden.stringmetric</groupId>
<artifactId>stringmetric-core_2.10</artifactId>
<version>0.27.4</version>
</dependency>Useful for approximate string matching and measurement of string distance. Most metrics calculate the similarity of two strings as a double with a value between 0 and 1. A value of 0 being completely different and a value of 1 being completely similar.
Dice / Sorensen Metric:
DiceSorensenMetric(1).compare("night", "nacht") // 0.6
DiceSorensenMetric(1).compare("context", "contact") // 0.7142857142857143Note you must specify the size of the n-gram you wish to use.
Hamming Metric:
HammingMetric.compare("toned", "roses") // 3
HammingMetric.compare("1011101", "1001001") // 2Note the exception of integers, rather than doubles, being returned.
Jaccard Metric:
JaccardMetric(1).compare("night", "nacht") // 0.3
JaccardMetric(1).compare("context", "contact") // 0.35714285714285715Note you must specify the size of the n-gram you wish to use.
Jaro Metric:
JaroMetric.compare("dwayne", "duane") // 0.8222222222222223
JaroMetric.compare("jones", "johnson") // 0.7904761904761904
JaroMetric.compare("fvie", "ten") // 0.0Jaro-Winkler Metric:
JaroWinklerMetric.compare("dwayne", "duane") // 0.8400000000000001
JaroWinklerMetric.compare("jones", "johnson") // 0.8323809523809523
JaroWinklerMetric.compare("fvie", "ten") // 0.0Levenshtein Metric:
LevenshteinMetric.compare("sitting", "kitten") // 3
LevenshteinMetric.compare("cake", "drake") // 2Note the exception of integers, rather than doubles, being returned.
N-Gram Metric:
NGramMetric(1).compare("night", "nacht") // 0.6
NGramMetric(2).compare("night", "nacht") // 0.25
NGramMetric(2).compare("context", "contact") // 0.5Note you must specify the size of the n-gram you wish to use.
Overlap Metric:
OverlapMetric(1).compare("night", "nacht") // 0.6
OverlapMetric(1).compare("context", "contact") // 0.7142857142857143Note you must specify the size of the n-gram you wish to use.
Ratcliff/Obershelp Metric:
RatcliffObershelpMetric.compare("aleksander", "alexandre") // 0.7368421052631579
RatcliffObershelpMetric.compare("pennsylvania", "pencilvaneya") // 0.6666666666666666Weighted Levenshtein Metric:
WeightedLevenshteinMetric(10, 0.1, 1).compare("book", "back") // 2
WeightedLevenshteinMetric(10, 0.1, 1).compare("hosp", "hospital") // 0.4
WeightedLevenshteinMetric(10, 0.1, 1).compare("hospital", "hosp") // 40Note you must specify the weight of each operation. Delete, insert, and then substitute. Note that while a double is returned, it can be outside the range of 0 to 1, based upon the weights used.
Useful for indexing by word pronunciation and performing sounds-like comparisons. All metrics return a boolean value indicating if the two strings sound the same, per the algorithm used. All metrics have an algorithm counterpart which provide the means to perform indexing by word pronunciation.
Metaphone Metric:
MetaphoneMetric.compare("merci", "mercy") // true
MetaphoneMetric.compare("dumb", "gum") // falseMetaphone Algorithm:
MetaphoneAlgorithm.compute("dumb") // tm
MetaphoneAlgorithm.compute("knuth") // n0NYSIIS Metric:
NysiisMetric.compare("ham", "hum") // true
NysiisMetric.compare("dumb", "gum") // falseNYSIIS Algorithm:
NysiisAlgorithm.compute("macintosh") // mcant
NysiisAlgorithm.compute("knuth") // nnatRefined NYSIIS Metric:
RefinedNysiisMetric.compare("ham", "hum") // true
RefinedNysiisMetric.compare("dumb", "gum") // falseRefined NYSIIS Algorithm:
RefinedNysiisAlgorithm.compute("macintosh") // mcantas
RefinedNysiisAlgorithm.compute("westerlund") // wastarladRefined Soundex Metric:
RefinedSoundexMetric.compare("robert", "rupert") // true
RefinedSoundexMetric.compare("robert", "rubin") // falseRefined Soundex Algorithm:
RefinedSoundexAlgorithm.compute("hairs") // h093
RefinedSoundexAlgorithm.compute("lambert") // l7081096Soundex Metric:
SoundexMetric.compare("robert", "rupert") // true
SoundexMetric.compare("robert", "rubin") // falseSoundex Algorithm:
SoundexAlgorithm.compute("rupert") // r163
SoundexAlgorithm.compute("lukasiewicz") // l222StringAlgorithm:
StringAlgorithm.computeWithMetaphone("abcdef")
StringAlgorithm.computeWithNysiis("abcdef")StringMetric:
StringMetric.compareWithJaccard(1)("abcdef", "abcxyz")
StringMetric.compareWithJaroWinkler("abcdef", "abcxyz")It is possible to decorate algorithms and metrics with additional functionality, which you can mix and match. Decorations include:
-
withMemoization: Computations and comparisons are cached. Future calls made with identical arguments will be looked up, rather than computed.
-
withTransform: Transform arguments prior to computation/comparison. A handful of pre-built transforms are located in the transform module.
Non-decorated:
MetaphoneAlgorithm.compute("abcdef")
MetaphoneMetric.compare("abcdef", "abcxyz")Using memoization:
(MetaphoneAlgorithm withMemoization).compute("abcdef")Using a transform so that we only examine alphabetical characters:
(MetaphoneAlgorithm withTransform filterAlpha).compute("abcdef")
(MetaphoneMetric withTransform filterAlpha).compare("abcdef", "abcxyz")Using a functionally composed transform so that we only examine alphabetical characters, but the case will not matter:
val composedTransform = (filterAlpha andThen ignoreAlphaCase)
(MetaphoneAlgorithm withTransform composedTransform).compute("abcdef")
(MetaphoneMetric withTransform composedTransform).compare("abcdef", "abcxyz")Making your own transform:
val myTransform: StringTransform = (ca) => ca.filter(_ == 'x')
(MetaphoneAlgorithm withTransform myTransform).compute("abcdef")
(MetaphoneMetric withTransform myTransform).compare("abcdef", "abcxyz")Using memoization and a transform:
((MetaphoneAlgorithm withMemoization) withTransform filterAlpha).compute("abcdef")$ git clone https://github.com/rockymadden/stringmetric.git
$ cd stringmetric
$ sbt clean package
$ ./project/build.sh
$ ./target/cli/jarometric abc xyzGet help:
$ metaphonemetric --help
Compares two strings to determine if they are phonetically similarly, per the Metaphone algorithm.
Syntax:
metaphonemetric [Options] string1 string2...
Options:
-h, --help
Outputs description, syntax, and options.Get comparison value with metrics:
$ jarowinklermetric dog dawg
0.75Get representation value with phonetic algorithms:
$ metaphonealgorithm dog
tkThe MIT License (MIT)
Copyright (c) 2013 Rocky Madden (http://rockymadden.com/)
Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in
all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
THE SOFTWARE.