Re: Case sensitivity in CLASSIFY and LEARN

"Ger Hobbelt" <[email protected]>
Newsgroups gmane.mail.spam.crm114
Message-ID <[email protected]>
On Tue, Sep 30, 2008 at 1:31 AM, James Lee <csejl-/[email protected]> wrote:
> I'm wondering if the CLASSIFY and LEARN statements are case sensitive.  If so, does that mean I have to convert all texts into the same case before feeding them into the CLASSIFY and LEARN statements to achieve higher accuracy
[...]

Yup, CLASSIFY and LEARN are caseSENsitive (technically, they treat
each byte value as a different one and perform no
(codepage/xyz/...-)conversions whatsoever).

That means, in case you want case-sensitive (or other modifying)
behaviour for these two fellas, you'll need to preprocess your data
using, e.g, 'translate /a-z/ /A-Z/' (see CRM QUICKREF and manual for
more info).


> The <nocase> flag only applies to the word pattern part of the statement, correct?

Perfect score! ;-)

That regex is used to 'pick' the characters (or should I say: bytes)
from the input, which are bundled together to form another 'word'
which will be converted into a [set of] feature[s] for training
(learn) or classification (classify).


> In my daemonization code (using FIFO), if I feed in a buffer that's too big I get the following error and the entire program exits:
>
> /usr/bin/crm: *ERROR*
[...]
> I used the -w flag to increase

Good. Warned for that issue. This is the 'proper' fix (better: the
only real viable one for CRM114 ATM).


> prevent the entire program from exiting in case something like this happens?  For example, is it possible to ignore this error, re-initialize and carry on (to the next iteration of a loop)?

Yes, you may wish to check out the 'trap' script command: you can
'catch' any ERROR with that -- FATAL errors (reported as '*FATAL*')
cannot be catched though.

Again, see QUICKREF and the manual (I believe it's available on the
wiki, right? Didn't check; I am lazy) for some examples of use.

(Of course, this bypasses the second issue when you perform 'error
recovery': your client/server protocol might need an extra to signal a
particular CRM114 operation went belly up this way. ;-) )


-- 
Met vriendelijke groeten / Best regards,

Ger Hobbelt

--------------------------------------------------
web: http://www.hobbelt.com/
 http://www.hebbut.net/
mail: [email protected]
mobile: +31-6-11 120 978
--------------------------------------------------

-------------------------------------------------------------------------
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=/
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.