Skip to main content

Table 4 Classifier performances on sub-domains. The entries of the table are AUC values of Naïve Bayes classifier performance results. The AUC was derived by training the Naïve Bayes classifier on a dataset (row) and use the model learned to classify a second dataset (column). The first row of numbers is the AUC from 10-fold cross-validation. The second row of numbers is the AUC values when the classifiers are trained using abstracts from other categories

From: Automating document classification for the Immune Epitope Database

 

Classification performance on sub-domain (AUC)

Train Set

AC

Allergen

ER

Other

Sub-domain (cross-validation)

0.847

0.856

0.858

0.861

All other sub-domains

0.797

0.817

0.852

0.784