Research on Key Technologies of Chinese Spelling Check Based on Machine Learning

Chinese spelling error correction technology is an important technology that teachers of Chinese as a foreign language should master. The research on Chinese spelling error correction technology can not only be applied to Pinyin input method, document editing tools, search engines, chat robots, voice assistants, etc., but also help Chinese learners at home and abroad to learn Chinese, improve learning efficiency and reduce learning pressure. However, the existing Chinese spelling error correction technology has some problems, such as insufficient context information of word vector, and the generated candidate words are limited to the confusion set. This is an urgent need for new Chinese spelling error correction technology to automatically detect and correct data to improve the error correction performance of text., In this paper, we propose a Spell model based on machine learning. After obtaining the final candidate words, we extract the feature information of the candidate words and their sentences, and use XGBoost classifier to filter out the correct words. The experiment shows that the performance of Spell model is significantly improved compared with the traditional model.