Conversation error correction method and device and electronic equipment

Abstract

The invention provides a conversation error correction method and device and electronic equipment, and the method comprises the steps: carrying out the hidden layer weighting of a sentence through a word granularity language model, and obtaining a hidden layer weighting combination vector of the sentence; then, carrying out grammar diagnosis on the hidden layer weighted combination vector of the sentence; obtaining a grammar diagnosis result of the sentence; according to the embodiment of the invention, the word granularity language model is obtained by training a large amount of unlabeled data; the purpose of the invention is to enhance the grammatical diagnosis correctness of a grammatical diagnosis model with less annotation data, and the learning training word granularity language model can effectively improve the annotation effect and generalization ability of the grammatical diagnosis model, and has feasibility and usability in practice.

Description

technical field [0001] The invention relates to the technical field of dialogue error correction, in particular to a dialogue error correction method, device and electronic equipment. Background technique [0002] With the development of the Internet, virtual robots based on artificial intelligence technology are more and more widely used in the field of enterprise user services. Robot customer service does not need to rest, it can respond to users' questions more quickly and standardizedly, communicate with users in the form of voice dialogue or text chat, and liberate human customer service from a large number of repetitive questions and answers. [0003] In the dialogue scenario between the user and the customer service robot, for example, the sentence entered by the user may contain grammatical errors, for example, when typing too much, which is redundant, because the downstream dialogue and translation models basically use relatively standardized corpus Training, these...

AI Technical Summary

Problems solved by technology

[0003] In the dialogue scenario between the user and the customer service robot, for example, the sentence entered by the user may contain grammatical errors, for example, when typing too much, which is redundant, because the downstream dialogue and translation models basically use relatively standardized corpus Training, these grammatical errors will have a great adverse effect on these translation models, such as repeated redundant input, which will cause the translation model to translate repeatedly, greatly affecting the readability of the translated text

[0004] In order to improve the dialogue error correction performance between users and customer service robots, the existing technology uses a grammatical error diagnosis model to detect these errors in advance, and then reconstructs the sentence according to the type of error judged by the model. However, because the existing grammatical error diagnosis model only relies on Manually labeled training data, and the number of these manually labeled training data is relatively small, which greatly reduces the error correction performance and generalization ability of the existing grammatical error diagnosis model

Images