Natural language correction method and system, equipment and storage medium

Abstract

The invention provides a natural language correction method and system, equipment and a storage medium. The method comprises the steps that a pre-correction text word vector sequence serves as model input, an after-correction text word vector sequence serves as model output, and a sequence-to-sequence model for correction is trained; a correct text for training is adopted to serve as a training set of a language model, and the probability of united and continuous occurrence of all r characters is subjected to statistical analysis, wherein r is preset statistical fineness; a to-be-corrected text word vector sequence is input into the sequence-to-sequence model for correction, and multiple candidate statements are output; and the candidate statements are input into the language model, the probability of occurrence of each candidate statement is calculated, and the candidate statement with the highest probability of occurrence is used as an output correction statement. By the adoption ofthe scheme, a traditional natural language processing method is replaced to realize natural language correction of Chinese, maintenance cost is lowered, and accuracy is substantially improved.

Description

technical field [0001] The present invention relates to the technical field of artificial intelligence, in particular to a natural language correction method, system, equipment and storage medium. Background technique [0002] In the field of natural language processing, it is often necessary to process text information, because of different sources of text, there may be errors such as typos and grammatical errors. Natural language correction technology is widely used to correct text errors, correcting wrong sentences, phrases or words into correct and processable text information, among which the application of correcting wrong sentences is the most common. [0003] At present, the most extensive Chinese natural language correction system uses traditional natural language processing (NLP) technology to solve the problem of sentence correction from two aspects: 1. Pre-set the wrong word library containing all wrong words to match sentences The wrong words in the sentence, c...

AI Technical Summary

Problems solved by technology

[0004] However, due to the diversity of language expressions, the accuracy of matching wrong words through the wrong thesaurus is relatively low, and some grammatical and punctuation errors cannot be corrected

At the same time, this method requires a lot of manpower to collect and sort out the wrong lexicon. Therefore, in the actual application of enterprises, the natural language correction system based on traditional natural language processing cannot meet the needs of online applications in terms of accuracy and cost.

Images