Stylistic and lexical co-training for webpage block classification

13 Nov 2004

WIDM 04: Lee et al. Co-training Web Block Classification

Evaluations

•Adapted co-training:

–Sample balancing: preserve ratio of noisily labeled examples, poor performance without it

–Replace unlabeled data at each round

•Use BoosTexter: handles word features easily

•Five fold cross validation

•

•General performance?

•Specific performance on:

–Fine-grained classification?

–XHTML / DIV pages?

–Others’ tasks?