Welcome to the Monastery | |
PerlMonks |
comment on |
( [id://3333]=superdoc: print w/replies, xml ) | Need Help?? |
No: 1 + 50C1 + 50C2 = 1 + 50 + 1225 = 1276, nearly 4 times the 326 different regexps required for a length-25 string. Sorry, I was lapse in my language. Instead of "...length of the strings...", I should have said sequences. That is, if you double the length of the sequences being searched, from 1000 to 2000 chars (average), but search for the same 25-char needle, then the time taken is slightly less that double. I agree if the length of the needles double, the number of regexes required close to quadruples. As I understand the problem, using /./ for the fuzziness is the right thing. However, whilst using a regex with 2 wildcards will match anythng that the same regex with only one wildcard would match, the problem with discarding the latter is that you will no longer be able to record how fuzzy the match was. Other than perhaps capturing the wildcards and then using substr to determine which of the wildcards would have matched had it not been wild. Overall, the slowdown through capturing + the additional work determining the fuzz factor, it is probably quicker to retain the 1-mis and 2-mis regexes? In reply to Re^3: Fuzzy Searching: Optimizing Algorithm Selection
by BrowserUk
|
|