Overview of the NLPCC 2023 Shared Task: Chinese Spelling Check.

This paper provides an overview of the Chinese Spelling Check shared task 8, held at NLPCC 2023. The task aims to correct spelling errors in Chinese sentences, including homophonic errors, visually similar errors, and other types of errors found in Chinese news articles. We present the task’s description, previous work related to Chinese Spelling Check, statistics on the provided dataset, evaluation results, and a summary of the submitted approaches. The dataset consists of 1k instances for development and 11k instances for evaluation, collected from publicly available news articles and books. The errors in the dataset were manually identified and labeled. A total of 47 teams registered for the task, with 12 teams submitting their results. The evaluation metric used is the F1 score, with the highest achieved score being 0.5888. The submitted approaches employ various techniques to enhance performance, each focusing on different aspects of the problem. For more information, please visit our official website.