Pinyin-based dialogue system text error correction system and method

Abstract

The invention discloses a pinyin-based dialogue system text error correction system and a pinyin-based dialogue system text error correction method. According to the embodiment of the invention, afterabnormal information is detected in a dialogue system text based on pinyin, the abnormal information is matched with a set information list by adopting a set text fuzzy matching strategy (TFM), and then matched information is obtained; and similarity calculation is performed on the abnormal information and the obtained matched information through adoption of the set Pinyin-based text similarity algorithm (PTS) to obtain information with the highest similarity, and the abnormal information is corrected. According to the embodiments of the invention, the error correction accuracy is improved onthe basis of reducing the error correction cost.

Description

technical field [0001] The invention relates to language processing technology in the field of computers, in particular to a text error correction system and method of a dialogue system based on pinyin. Background technique [0002] Correcting speech errors based on speech recognition results is an important task in the speech understanding process. Due to the limitation of the accuracy of speech recognition, errors often occur in the results of speech recognition, which will hinder the follow-up work of speech understanding and increase the difficulty of speech understanding. [0003] When correcting the dialogue text based on pinyin, it is divided into two steps, the first step is error detection, and the second step is error correction. Among them, the commonly used methods for error detection include maximum entropy, n-gram speech model, etc., and error correction uses perplexity sets or language models to selectively replace detected errors. [0004] With the developm...

AI Technical Summary

Problems solved by technology

[0005] However, using the deep learning model to correct the text of the dialogue system based on Pinyin also requires a large amount of data for training, which takes a long time and is not easy to implement, which increases the cost of error correction

In addition, the trained deep learning model performs error correction on the text of the dialogue system based on Pinyin in a certain field during training. When the deep learning module is transferred and applied to the text of the dialogue system based on Pinyin in another field, then The accuracy of error correction will be greatly reduced, that is, the migration of the error correction of the text of the dialogue system based on Pinyin using the deep learning model is poor

Images