Re: Request for Comment: Callable CRM114 Classifiers (libcrm114)
Bill Yerazunis <[email protected]>
| Newsgroups | gmane.mail.spam.crm114 |
|---|---|
| Message-ID | <[email protected]> |
From: Thomas Spahni <[email protected]> Dear Bill, dear listmembers, The above thread seemingly died some weeks ago. This leads me to the question whether the dust has settled by now. Is there any information which direction the libcrm114-project is heading to and what's its current status? Best regards Thomas Thomas: It's moving forward; I had to do a redirect because of a segfault report in the FSCM classifier. The good news is that I think I've fixed it (I'm bashing it now) and in the process it got significantly faster and uses less memory/disk for the statistics files. F'rinstance, it runs FSCM tenfold-validate on the 4147-file SA corpus in just 9 minutes with 38 total errors. :-) That's over 75 msgs/second (13 milliseconds a message!) I still need to tweak the new pR values (the normalization is low now; the optimal thick threshold seems to be around 0.05 pR's and it should be remapped to be around 5 to 10 to match most other classifiers. I have another dataset to run it against and then I will be uploading it (thanks for the help, Paolo). SO- to answer your question, yes, libcrm114 is still alive and growing, but I need time to mull over the suggestions. A lot of it has to do with object-oriented versus data-and-procedure oriented coding styles. - Bill Yerazunis ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/