Method and device for expanding training sample

Abstract

The embodiment of the invention provides a method and a device for expanding training samples. The method comprises: acquiring a to-be-expanded initial training sample group, the to-be-expanded initial training sample group comprising a first number of training samples, the training samples comprising original machine statements and original user statements in a historical dialogue group and category labels corresponding to the dialogue group, and each training sample in the initial training sample group having a first category label; obtaining a second number of training samples from the initial training sample group; for each dialogue group in the second number of training samples, inputting a first machine statement related to the original machine statement in each dialogue group into apre-trained dialogue generation model of a first class label, and generating a first user statement corresponding to the first machine statement in each dialogue group; and adding the first machine statement and the corresponding first user statement in each dialogue group into the initial training sample group as extended training samples to obtain an extended training sample group. Data equalization of training samples can be realized.

Description

technical field [0001] One or more embodiments of this specification relate to the field of computers, and in particular to methods and devices for expanding training samples. Background technique [0002] When the robot answers the user's question, after the dialogue between the robot and the user, classify the dialogue groups including machine sentences and user sentences, and determine the user's request according to the classification results. The above classification may include determining a standard question sentence corresponding to the dialogue group, so that the robot provides an answer corresponding to the standard question sentence. Wherein, the standard questions are also referred to as standard questions, which are sorted out questions that some users may ask. Each question has a question ID. [0003] In the prior art, the classification model is usually trained by using historical conversation groups as training samples, and then the current conversation gro...

AI Technical Summary

Problems solved by technology

Data imbalance will cause the classification model to be biased to determine the dialogue group as the category of training samples with a large amount of data, so how to solve the problem of data imbalance has become the core issue of improving classification accuracy

Images