Tag
This paper introduces RM-EVAL, a reward model trained on human preference data for reference-free meta-evaluation of grammatical error correction, and shows how it can improve GEC systems via reward-guided text generation.
This paper presents an optimized prompt-based method for minimal-edit grammatical error correction that achieves state-of-the-art results by using taxonomy-based instructions, batching, and LLM-assisted prompt optimization.
This paper introduces STAGEET, a stage-wise typed edit-tagging framework for grammatical error correction that improves interpretability and achieves state-of-the-art results on Arabic benchmarks.
This paper introduces Grammatical Error Representation (GER), a novel method for retrieving in-context demonstrations based on error patterns rather than semantic similarity, significantly improving multilingual grammatical error correction performance in LLMs with in-context learning.
Proposes CoCoGEC, a counterfactual generation framework that alters error-irrelevant contexts in GEC training data to improve model robustness, achieving significant F0.5 gains on perturbed benchmarks.
Introduces ArabiGEE, the first comprehensive Arabic grammatical error explanation taxonomy with a hierarchical structure spanning orthographic, morphological, syntactic, and lexical dimensions, comprising 27 error types, 140 correction types, and 324 explanations.
This paper refines word-based grammatical error annotation for L2 Korean by addressing problems in existing resources, including surface target realization and single-reference evaluation, and demonstrates improvements using KoBART-based correction.