FIELD: physics.
SUBSTANCE: in the method for training the information classifier, phrase samples containing the target keyword are extracted from information for selection. Binary tags are assigned to the phrase samples to obtain a training set of samples based on the attribution of each of the phrase samples to the target class. Each phrase sample is split into words in the training set of samples to get a set of words. The given characteristic set of a set of words containing the characteristic words is selected. The classifier is constructed, based on characteristic words and trained on the basis of the results of assigning binary marks in the training set of samples.
EFFECT: improving the accuracy of the information recognition results.
14 cl, 9 dwg
