Re: [Classifier4j-devel] pseudocode

[Nick L. <ni...@ma...>]

Sorry - I've been away for a few days.

To train Classifier4J, do something like this:

http://sourceforge.net/mailarchive/forum.php?thread_id=3D5110155&forum_id=
=3D34026

That will y the text passed to the teachMatch & teachNonMatch methods as=20
it goes.

To classify after training use the classify(String) and/or=20
isMatch(String) methods.

Nick


satmeet wrote:

> hi ,
> I looked at the Code again and have gone through, the archives of the=20
> mailing list (first 5 months). I am writing down a pseudocode for=20
> Bayesian that I maybe you could help me visualize in terms of=20
> CLASSIFIER4J.
> I am uploading this to>>  www.satmeet.com/bayesian.html
> its optimized for IE (sorry ,I was in a hurry)
>
> I would like to know, what is the sequence for making tokens from=20
> text, then using them for Training and Classification . If you could=20
> just tell me the flow of Classifier4J according to this given pattern=20
> , I know you are busy people but I will be very greatful if you could=20
> help me .
>
> In pseudo-code training is ,Given: an email message, X, and a label Ci=20
> =CE=B5 {CN,CS},
> 1. break X into its tokens, hx1, . . . , xki
> 2. for each token, xj
> (a) Increment the counter for token xj for class Ci
> (b) Increment the count of total tokens in class Ci
> 3. Increment the total number of email messages for class Ci     =20
>
>
> And Classification can be written as :
> Given: an UNLABELED email message, X
> 1. PN :=3D Pr[CN]
> 2. PS :=3D Pr[CS]
> 3. break X into its tokens, hx1, . . . , xki
> 4. for each token, xj
> (a) PN :=3D PN =E2=80=A2 Pr[xj |CN]
> (b) PS :=3D PS =E2=80=A2 Pr[xj |CS]
> 5. if PN >PS then return NORMAL
> 6. else return SPAM
>
>
> Where Pr[CN] =3D#NORMAL emails     =3D     #NORMAL emails
>                  --------------           ---------------
>                  total # emails           # SPAM + # NORMAL
>
> Pr[CS] =3D # SPAM emails           =3D   #SPAM emails
>               ----------                ------------
>              total # emails             #SPAM + # NORMAL
>
> Pr[xj |Ci] =3D # of tokens of type xj seen in class Ci
>                 -------------------------------------------
>                      total # of tokens seen in class Ci
>
> Thanking you
>
> Satmeet
>
>
>
>