Text sequence generation method and system

Abstract

The embodiment of the invention provides a text sequence generation method and system. A adversarial network model comprises a generator and a discriminator; the generator comprises a feature feedbacknetwork module and is used for calling high-order features from the discriminator so as to generate a guidance feature vector; the generator generates a next word of the pre-output text sequence based on the guidance feature vector; the discriminator updates the high-order features according to the newly constructed pre-output text sequence and the initial text sequence; the above steps are iteratively executed until the total number of words of the newly constructed pre-generated text sequence reaches a set threshold, the total number of words is input into a discriminator for discrimination, and an update gradient is calculated according to a classification discrimination result; parameters of the generator are updated according to the updating gradient; and the above steps are iteratively executed until the generator and the discriminator converge, and a target text sequence is output. According to the embodiment of the invention, the feature conversion feedback module is added, and the feature vector extracted by the discriminator is transmitted to the generator as a guide signal, so that the generation quality of the generator is improved.

Description

technical field [0001] The present invention relates to the field of computer technology, in particular to a text sequence generation method and system thereof. Background technique [0002] In recent years, with the continuous development of artificial intelligence, natural language processing has also received extensive attention. Among them, text generation, as one of the medium and long-term research topics in the field of natural language processing, has been applied to various technologies, such as machine translation, text summarization, automatic question answering, intelligent customer service, etc. Text generation aims to predict and generate grammatically and semantically correct word sequences in continuous specific contexts, and is an integral part of interactive artificial intelligence systems. [0003] At present, many scholars have made extraordinary achievements in text generation. When implementing text generation tasks, we often use feedforward neural ne...

AI Technical Summary

Problems solved by technology

[0008] In view of this, the embodiment of the present invention provides a text sequence generation method and its system to solve the problem in the prior art that the feedback signal of the discriminator in the generation confrontation network model is weak, causing the generator to receive scalar signals and cannot effectively guide The network is updated in the correct direction, resulting in poor convergence

Images