Re^2: A Better Word Morph Builder

in reply to Re: A Better Word Morph Builder
in thread A Better Word Morph Builder

I have benchmarked your method. If i am not mistaken, is's another example of bidirectional search, isn't it?

Benchmark results:

             Rate  find_path       solo find_path2  transform
find_path  5.91/s         --       -62%       -75%       -97%
solo       15.7/s       166%         --       -35%       -92%
find_path2 24.1/s       308%        53%         --       -87%
transform   188/s      3084%      1097%       681%         --
[download]

Note: Your solution seems to be suitable only for basic English charset:

  for my $letter ( 'a'..'z' ) {
[download]

And what about the Unicode? ;)

Comment on Re^2: A Better Word Morph Builder Select or Download Code

Replies are listed 'Best First'.
Re^3: A Better Word Morph Builder by Solo (Deacon) on Jun 29, 2006 at 20:43 UTC
Thanks for the benchmark! You're right, it is bi-directional. And you caught me, `'a'..'z'` was a precomputed optimization based on the set of characters found in my wordlist. ;) Does 2of12 include additional characters? --Solo -- You said you wanted to be around when I made a mistake; well, this could be it, sweetheart.	[reply] [d/l] [select]
Re^4: A Better Word Morph Builder by Ieronim (Friar) on Jun 29, 2006 at 21:57 UTC
2of12 includes only `[a-z'0-9-]`. But as i am Russian, i wanted the algorithm to work with any language ;) And it does, but takes a HUGE amount of time to process my 1Mb russian wordlist :(	[reply] [d/l]

In Section Seekers of Perl Wisdom