Sequential testing in classifier evaluation yields biased estimates of effectiveness.

William Webber, Mossaab Bagdouri, David D. Lewis, Douglas W. Oard

Browse the full SIGIR paper archive.