LANGUAGE INDEPENDENT KERNEL METHODS FOR CLASSIFYING TEXTS WITH DISPUTED PATERNITY
Liviu P. Dinu, Marius Popescu · 2009
The main goal of this paper is to investigate the behaviour of string kernels in author identifica- tion for texts with disputed paternity. We tested the method on the well known case of authorship of dis- puted Federalist papers and on an example from Romanian literature: did Radu Albala found the continua- tion of Mateiu Caragiale's novel Sub pecetea tainei, or did he write himself the respective continuation? A second objective of the paper is to see if the performance achieved is inherent to the kernel type, that is, if the performance remains the same if the kernel is used in conjunction with different kernel methods. We tested it with Support Vector Machines (SVM) and Kernel Fisher Discriminant (KFD).