Decision surfaces of the VTT1 (top) and VTT5 (bottom) classifiers with SP (left) and bigram (right) features, for the documents in test data. The decision surface and x(d) and y(d) are computed according to Eq. (7) for VTT1 (top) and Eq. (8) VTT5 (bottom), for every document d in test set. The plots for VTT1 surfaces display many documents d with the same values of y(d), plotted in horizontal rows, while VTT5 displays a smoother ranking of documents. This happens because VTT1 uses information from a single NER tool (ABNER protein mentions), while VTT5 uses information from five such tools; thus, while in the VTT1 plot many documents have the same value of ABNER protein mentions, in the VTT5 plot the various NER measurements lead to a finer distinction between documents.