Results of the Common Source Experiment with Swiss-Prot test data.

Each line corresponds to the best result (classifier) obtained for each source as we used the training and test data from 1 up to 100 features after SVD processing and the classification techniques Naïve Bayes, J48 and KNN with . The last two columns refer to the number of features or attributes (in the occurrence matrix that resulted in the best classification model) and to the number of classes in each classifier. The TPR corresponds to the recall and was omitted.