TRAINING SAMPLE GENERATION METHOD AND DEVICE FOR TEXT CLASSIFICATION, AND COMPUTER APPARATUS

摘要

A training sample generation method and device for text classification, and a computer apparatus, applied to the technical field of classification models. The method comprises: acquiring a first classification result of a pre-trained first text classification model for a feature vector of actual text (S110); inputting the feature vector of the actual text to a pre-trained second text classification model so as to acquire a second classification result, wherein the second text classification model is used to test whether the first classification result of the first classification model is correct (S120); determining, by means of comparison, whether the first classification result is identical to the second classification result (S130); and if so, determining that the first classification result is correct, and storing the feature vector of the actual text and the first classification result as a training sample in a database, wherein the training sample is used for training a text classification model in a model training stage (S140).