Chinese Grammatical Error Correction, Annotation, and Evaluation

This book presents a research grounded and practice-oriented guide to Chinese grammatical error correction, with the goal of making the full pipeline understandable, reproducible, and extensible. Chinese grammatical error correction is now mature enough to have widely used datasets, established evaluation metrics, and a fast-moving system landscape, yet the field still lacks a single, coherent reference that connects these components. The need is particularly acute for Chinese, where segmentation, span definition, and reordering phenomena make grammatical error annotation decisions inseparable from evaluation outcomes and system design choices. By presenting the full workflow from data and annotation to metrics and modeling, the book consolidates a synthesis that is currently distributed across shared-task reports, general evaluation-metric documentation, and system papers. The authors do not treat Chinese grammatical error correction as an isolated task, but rather as a meeting point of linguistic analysis, annotation design, evaluation methodology, and system building. The book also offers practical grammatical error annotation guidance, including a clear contrast between manual schemes and automated pipelines, and an explicit treatment of how tokenization and segmentation reshape the notion of an edit. The book is ideal for readers who need both conceptual clarity and concrete procedures: graduate students entering the field, researchers who wish to compare approaches fairly, and engineers who require reliable computational pipelines.