Text categorization question

[icojo ()]

Hello,
I am new to Weka.
I am trying to build a classifier on a corpus of texts that are only positive examples. I am currently using StringToWord (with tf-idf transf.) filter with a SnowBall english stemmer feeding data to  OneClass classifier, but i can't get some relevant results.
My corpus is big but still it contains some noise also.
I've tried libsvm one-class classifier but that one is even less usefull for me.
Did anybody use this classifier for text categorization purpose ?
Any help is welcome :)