in reply to Re: longest common substring
in thread longest common substring

For more than 2 strings, I'd also suggest Tree::Suffix (although I've never used this module. It is just in theory the best algorithm).

Replies are listed 'Best First'.
Re^3: longest common substring
by Limbic~Region (Chancellor) on Jan 02, 2008 at 19:33 UTC
    lima1,
    The original thread I referenced was for longest common subsequence. I still have yet to find an implementation for finding the longest common subsequences for > 2 strings other than the one I posted. I adapted it for common substrings to show it was possible. I would definitely use and recommend Tree::Suffix now that I know about it. I mentioned my home grown implementation in this thread in case the XS version of the module flaviusm was using was also broken. Thanks!

    Cheers - L~R

[reply]
      I think the best exact solutions for the longest common subsequence problem are the dynamic programming ones. But this is only reasonable for 2-4 strings (0(n^2) to O(n^4)). So people use heuristics. A trivial one is to calculate a simple (tree) clustering with something like UPGMA. For example: 
      $a='AAAB';
$b='AABB';
$c='ABBB';
$d='BBBB';

     +---$a 
+----+
|    +---$b 
|
+
|
|    +---$c 
+----+
     +---$d

      [download]
      Then you calculate the subsequences of the leafs (a,b) and (c,d) and in the next step (according some special rules) the subsequences of the subsequences.
[reply]
[d/l]

In Section Seekers of Perl Wisdom