Model training method and device for generating similar texts and computer equipment

Abstract

The invention relates to a model training method and device for generating similar texts and computer equipment. The method comprises the steps that a preset similar sentence pair training set is used for conducting text similar relation training on a first word vector representation model, a second word vector representation model is obtained, and the second word vector representation model is used for recognizing the similar relation between texts; inputting each text in the single-intention text data training set into a second word vector representation model to obtain a word vector corresponding to each text; based on the single-intention text data training set and the word vector corresponding to each text, constructing a similar text sentence pair data set in the same service field as the single-intention text data training set; and training the initial similar text generation model according to the similar text sentence pair data set to obtain a target similar text generation model. The target similar text generation model obtained by using the model training method provided by the invention can automatically generate a plurality of similar texts, so that the generation efficiency of the similar texts is greatly improved.

Description

technical field [0001] The present application relates to the technical field of artificial intelligence, in particular to a similar text generation method, device and computer equipment. Background technique [0002] With the development of artificial intelligence technology, computers can identify corresponding intent tags for different input texts. The accuracy of intent label recognition will be directly related to the user experience of artificial intelligence products. In actual production operations, intent recognition models are usually designed and trained using supervised learning methods. Therefore, the quantity and quality of sample data sets play a crucial role in the recognition effect of the model. [0003] In order to achieve a better model recognition effect, at present, the traditional method of manually labeling similar problems is often used to obtain training samples, and enterprises often have to invest more manpower and material resources in the caref...

AI Technical Summary

Problems solved by technology

[0003] In order to achieve a better model recognition effect, at present, the traditional method of manually labeling similar problems is often used to obtain training samples, and enterprises often have to invest more manpower and material resources in the careful collection and construction of sample data sets. This method of sample accumulation leads to low collection efficiency of sample data sets

Images