Fusion language model generation method and device, word error correction method and electronic equipment

Abstract

The embodiment of the invention provides a fusion language model generation method and device, a word error correction method and electronic equipment. The fusion language model generation method comprises the steps of determining weights of at least two pre-trained language models; obtaining a fused prior probability according to the weight and the prior probability of the sample error correction candidate words in the spelling error correction training set in each language model; adjusting the weight of each language model according to the fused prior probability, the transition probability of sample spelling suggested words with the sample error correction candidate words as the sample spelling wrong words and the label information of the sample error correction candidate words; and generating a fusion language model based on a pre-trained language model and the adjusted weight of each language model. According to the method, the pre-trained language models are fused according to different weights to generate the fused language model, error correction is carried out on spelling wrong words by utilizing the fused language model, and reasonable words can be selected from the error correction candidate set to serve as final spelling suggested words.

Description

Technical field[0001]Embodiments of the present application relate to Internet technology, and more particularly to a fusion language model generation method and apparatus, a word error correction method, and an electronic device.Background technique[0002]Often, because people are negligent or unskilled with English words, English words spell errors in English articles or document writing; in the Internet age, the editing of English words via the keyboard and other hardware interactive equipment, It is also very easy to cause spelling errors for words.[0003]At present, the error correction of the words of the spelling error is mainly selected from the correct word from the error correction candidate, including: selecting the optimal word from the error correction candidate collection as spelling error correction, through a certain algorithm. Thus the purpose of automatic error correction. The error correction candidate is often included in several spelling error corrections, how to ...

AI Technical Summary

Problems solved by technology

[0004] In view of this, one of the technical problems solved by the embodiments of the present application is to provide a method and device for generating a fusion language model, a word error correction method and electronic equipment, so as to overcome the error correction of misspelled words in the prior art. , only considering the word frequency information to select spelling error correction suggestions from the error correction candidate set, resulting in inaccurate error correction and poor error correction effect, so as to make full use of the context information of misspelled words and integrate multiple language models to realize the correction of misspelled words Perform error correction to provide reasonable error correction suggestions as much as possible, so as to achieve better error correction effect

Images