Levenshtein Distance Algorithm: Fastest Implementation in C#

Here is a cleaned-up performance test for several different implementations of levenshtein I have blogged about recently. This test was emailed to me by Ahmed Ghoneim, who has also kindly agreed to its publication on my blog. I am very grateful to him for his excellent contribution. I have slightly altered his file to do away with the unnecessary local variables in my C2C# port of the GNULevenshtein method.

I would like to hear from you which methods perform best on your machine. Please drop a comment ^_^!

C#LevenshteinAlgorithmPerformanceTest.cs
code only

PackagePackages
code, data and sample binary in zip and self-executable zip formats

Please note that the GNULevenshtein method was found to be buggy! Here is the new replacement method.

Comments

Anonymous said…
I tried GNULevenshtein version with the words 'fot' and 'foot'. The resulting distance is 0. I thought it was 1.

Thanks
December 12, 2007 at 8:53 PM
Cetin Sert said…
That's a bug. I will try to fix it somehow. Thanks for your comment ^_^
December 12, 2007 at 10:42 PM
Cetin Sert said…
I have now tested the original version in C++ and this bug seems to be in the original code and not something I have introduced in my C# port. The section of the code that cuts equal start and end sequences is responsible for this behaviour. Removing this section resolves the issue but I will try to see how I can keep that part intact yet correct the results.
December 12, 2007 at 11:31 PM
Anonymous said…
Awesome. I am investigating alternative algorithms having a suffix trie dictionary. I already use it for exact matching and I am wondering if it can be used also for approximate matching. Do yo have some good reference for this?
Thanks
December 13, 2007 at 1:35 AM
Cetin Sert said…
'x' and 'foxot' return 5 instead of 4 and that's after I have fixed the bug in the stripping of equal start and end sequences. Which means there is something fundementally wrong with the reference implementation. I will be looking for a better implementation to replace the GNULevenshtein method altogether. Thanks for bringing this up! I recommend using UnsafeVectorLevenshtein.

By the way I don't know anything about suffix trie dictionaries so I can't help you there. I would like to add you to my email contacts and visit your blog if you have one. Please drop a comment.
December 13, 2007 at 2:00 AM
Anonymous said…
Sure thing: managed_dot_theory_at_live_dot_com

bye
December 13, 2007 at 3:17 AM
Cetin Sert said…
Thanks by the way I ported another equivalently optimized but highly obfuscated edit distance implementation which also lets one specify the cost of substitution.

http://www.koders.com/c/fidF5020AACF174C23AF41E57A8DF4080BF1FDC4849.aspx?s=levenshtein

I am going to update my blog ASAP after I have run several tests. Thank you very much for helping me correct my code.
December 13, 2007 at 3:22 AM
Anonymous said…
There is an algorithm for Levenshtein distance that uses diagonals and lazy evaluation, and runs in O(|A|(1 + DAB)) time (where |A| is the length of the first string, and DAB is the Levenshtein distance). Where the distance is small (strings are similar), this algorithm should be pretty fast. See http://www.csse.monash.edu.au/~lloyd/tildeStrings/Alignment/92.IPL.html

I have implemented it in Java as an exercise (feel free to improve it). http://www.its.caltech.edu/~xuanluo/Levenshtein.java
It should be pretty easy to port it into C# or something like that. I would be interested in how it does if you test it.
January 7, 2008 at 8:32 AM
Cetin Sert said…
^_^ Open-source is a great culture! Thanks for sharing your code... I'll have a go at it as soon as I can regain my consciousness: I'm literally blown up by my first exposure to F# and Haskell. o_O I can't believe that I missed out on Haskell for so long.
January 7, 2008 at 8:59 AM
Anonymous said…
Performance test (just run it from your ZIP archive):

Match: MULTIPLICATION Time: 509 Distance: 3 GNULevenshtein
Match: MULTIPLICATION Time: 479 Distance: 3 UnsafeVectorLevenshtein
Match: MULTIPLICATION Time: 657 Distance: 3 VectorLevenshtein
Match: MULTIPLICATION Time: 948 Distance: 3 MatrixLevenshtein
Match: MULTIPLICATION Time: 852 Distance: 3 JaggedLevenshtein

PC: Intel Core Quad 6600
OS: Windows Server 2008 x64
.NET: v3.5 SP1

Thank you for the perfect code!
Anton.
http://kyta.spb.ru
October 16, 2008 at 4:56 PM
Anonymous said…
Performance measurements (just unpacked zip):

Match: MULTIPLICATION Time: 1033 Distance: 3 GNULevenshtein
Match: MULTIPLICATION Time: 1191 Distance: 3 UnsafeVectorLevenshtein
Match: MULTIPLICATION Time: 1239 Distance: 3 VectorLevenshtein
Match: MULTIPLICATION Time: 3554 Distance: 3 MatrixLevenshtein
Match: MULTIPLICATION Time: 2962 Distance: 3 JaggedLevenshtein

PC: Athlon x2 4850e
OS: Ubuntu 8.04 amd64
.NET: Mono 1.9

Thank you!
Anton.
October 17, 2008 at 3:54 AM
Anonymous said…
The package download link is broken
November 12, 2008 at 4:32 PM
Cetin Sert said…
@anonymous: the link has been fixed! Thank you for reporting *^o^* !!
November 12, 2008 at 5:24 PM
Post a Comment

Popular posts from this blog

WordSmith Tools 5.0, Tenka Text in China

Mike Scott has finally released a public beta of WordSmith Tools 5.0 . I want to quote him on a few points: System Requirements WordSmith Tools 5 is for Windows 2000 or later. It will be happiest on a fairly modern PC (e.g bought in the last 4 years) with plenty of memory and hard disk space. Or an Intel Mac running Windows . Now that's an interesting world view we have here. What's an Intel Mac anyway? It's a PC produced by Apple. So this sentence actually reads: System Requirements: a PC running Windows or a PC (produced by Apple) running Windows. Technically deep Mr. Scott seems to be. -__-" Here is my favourite one: What's new in version 5 WordSmith is organic software ! Version 4 was a complete new re-write. Since it was launched in 2004, numerous re-compilations of WS4 were issued (about one a week on average), sometimes with a very small bug-fix but other times incorporating changes users had suggested. At the same time the Help was updated. Version 5.0 was
Read more

Mono 1.2.5 binaries for Solaris 10/x86

I think I've finally managed to botch it all up. I created a JRE-style zipped package, which you can download here 1,2 at your own risk ;). Whoa, it's taken me 4 days just to get so far with mono/solaris. ^o^ I wanna sleep like the relakkuma now hehe... These binaries are compiled for 32-bit/x86 processors. They also run on 64-bit/x86-64 processors. They are not compiled for running winforms applications. To extract the archive use: "gtar -zxvf"
Read more