A highly accurate Chinese spell checking method, system, and medium.

Abstract

This invention discloses a highly accurate Chinese spelling checking method, system, and medium. The method includes the following steps: acquiring a corpus to be used, configuring a replacement table to be used, and generating near-phonetic error samples and near-shape error samples to be used based on the corpus; training a decision model to be used based on the corpus, and training a checking model to be used based on the near-phonetic and near-shape error samples; and performing text checking and error correction operations on the acquired text to be checked based on the replacement table, the trained decision model, and the checking model. This invention can efficiently generate high-quality and highly realistic training samples, and can detect errors in Chinese text by independently setting the trained error detection model and post-processing decision model, and perform high-quality error correction judgment after error detection, further reducing the model's error correction rate and improving the correction quality, thus opening up a new method for Chinese spelling error detection.

Images