Leveraging What's Overfixed: Post-Correction via LLM Grammatical Error Overcorrection

2025-09-25

AI Summary

In grammatical error correction (GEC), small language models (sLMs) achieve high precision but suffer from low recall, whereas large language models (LLMs) exhibit the opposite trade-off—hindering simultaneous optimization of both metrics. To address this, we propose Post-Correction via Overcorrection (PoCO): an LLM first deliberately overcorrects input text to maximize error detection (boosting recall), and a supervised fine-tuned sLM then refines these overcorrected outputs to restore accuracy. Crucially, PoCO repurposes LLM-induced overcorrections as informative signals and endows the sLM with discriminative filtering and precise editing capabilities. Experiments across multiple GEC benchmarks demonstrate that PoCO significantly improves recall (+12.3%) without sacrificing precision—indeed, precision slightly increases—yielding a new state-of-the-art F₀.₅ score. Thus, PoCO effectively balances and enhances overall GEC quality.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: (Large) Language Models
• Search and Optimization: Learning to Search

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Large language models for search
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

Robust supervised fine-tuned small Language Models (sLMs) often show high reliability but tend to undercorrect. They achieve high precision at the cost of low recall. Conversely, Large Language Models (LLMs) often show the opposite tendency, making excessive overcorrection, leading to low precision. To effectively harness the strengths of LLMs to address the recall challenges in sLMs, we propose Post-Correction via Overcorrection (PoCO), a novel approach that strategically balances recall and precision. PoCO first intentionally triggers overcorrection via LLM to maximize recall by allowing comprehensive revisions, then applies a targeted post-correction step via fine-tuning smaller models to identify and refine erroneous outputs. We aim to harmonize both aspects by leveraging the generative power of LLMs while preserving the reliability of smaller supervised models. Our extensive experiments demonstrate that PoCO effectively balances GEC performance by increasing recall with competitive precision, ultimately improving the overall quality of grammatical error correction.

Problem

• Small language models undercorrect with low recall in grammatical error correction
• Large language models overcorrect with low precision in grammatical error correction
• Need to balance recall and precision for optimal grammatical error correction performance

Innovation

• LLM overcorrection maximizes grammatical error recall
• Post-correction refines outputs using fine-tuned smaller models
• Balances recall and precision in error correction