Text generation method and device

Abstract

The invention provides a text generation method and device, and relates to the technical field of natural languages. The invention provides a text generation method and device. The method comprises the steps of determining a first segmented word in a first text corpus; under the condition that the first segmented word is matched with the preset feature, replacing the first segmented word with a feature mark corresponding to a preset feature to obtain a second text corpus, obtaining a first word vector corresponding to a first segmented word in the first text corpus, the first segmented word and the first feature vector corresponding to the feature mark in the second text corpus; inputting the first word vector and the first feature vector into the text generation model at the moment, obtaining the output target word vectors of the target semantic structure, and then obtaining the target text corpus according to the target word vectors. The target text corpus obtained by the embodimentof the invention comprises the required and complete target semantic structure, and the first text corpus does not limit the obtaining mode, so that the problems of high similarity and poor extensioneffect between the extended target text corpuses are avoided.

Description

technical field [0001] The present invention relates to the technical field of natural language, in particular to a text generation method and device. Background technique [0002] In order to meet the growing needs of named entity recognition, speech recognition, speech synthesis, machine translation, etc., text recognition models are needed for text recognition. [0003] At present, the corresponding text recognition model is usually trained by collecting text corpus in different scenarios, different fields, and different language families. However, in order to ensure that the text recognition model is fully trained and improve the accuracy of the recognition results, a large amount of text corpus that meets the model training requirements is usually required. However, limited by the acquisition conditions, sometimes the quantity and quality of the text corpus cannot be guaranteed, making the text corpus sparse, resulting in insufficient training of the training text reco...

AI Technical Summary

Problems solved by technology

However, limited by the collection conditions, sometimes the quantity and quality of the text corpus cannot be guaranteed, making the text corpus sparse, resulting in insufficient training of the training text recognition model

[0004] In order to expand the text corpus, the existing technology usually adopts the method of adding noise. On the basis of the original text corpus, a new text corpus is generated by means of synonym replacement, random word insertion, random word deletion, random word exchange, etc., but the synonym replacement may As a result, the new text corpus has a high similarity with the original text corpus, and the expansion effect is poor, and random word insertion, deletion, exchange, etc. in the text corpus may destroy the semantic structure of the text corpus, and the destruction of the semantic structure of the text corpus may affect the performance of the text recognition model. The efficiency of training and the accuracy of recognition results

Images