Authorship Verification as a One-Class Classification Problem
Moshe Koppel - Bar-Ilan University
Jonathan Schler - Bar-Ilan University
In the authorship verification problem, we are given examples of the writingof a single author and are asked to determine if given long texts were or werenot written by this author. We present a new learning-based method foradducing the “depth of difference” between two example sets and offer evidencethat this method solves the authorship verification problem with very highaccuracy. The underlying idea is to test the rate of degradation of theaccuracy of learned models as the best features are iteratively dropped fromthe learning process.