seminar 1/23 note 2

Experiment part
IO-bridge : consider only siblings of the target page within a human labeled dataset
bridge : if two or more pages have the same class, while not committing what that class would be, we call these documents bridges.
IO-bridge : a page B points to both document b1 and b2, and the way from b1 to b2 is to traverse the edge (B,b1) against its direction and then (B,b2). So B is IO-bridge for b1 and b2 because of the inlink to B (b1->B) and followed by an outlink (B->b2). http://delivery.acm.org/10.1145/280000/276332/p307-chakrabarti.pdf?key1=276332&key2=0117839611&coll=GUIDE&dl=GUIDE&CFID=9534706&CFTOKEN=44327035

seminar 1/23 note 1

Experiment part
K+C from Calado : link-based and content-based combination for web document classification
kNN for content-based , co-citation for link-based.
co-citation : a Web page author will insert links to pages related to his own pages. Apply co-
citation to Web documents by treating links as citations.
http://delivery.acm.org/10.1145/960000/956938/p394-calado.pdf?key1=956938&key2=5009639611&coll=GUIDE&dl=GUIDE&CFID=11980192&CFTOKEN=64981017