13 Nov 2004
WIDM 04: Lee et al. Co-training Web Block Classification
5
Which approach to use
•A obvious approach is to build a supervised classifier
–Train on labeled examples (f1,f2,…,fi,…,fn, C)
–Test by distilling features (f1,f2,…,fi,…,fn) = ?
•
•Training data costly, need to use unlabeled data
•The feature sets are largely orthogonal
• = Try co-training!