Grammatical Error Correction

The CoNLL dataset is a widely used resource in the field of natural language processing (NLP). The term “CoNLL” stands for Conference on Natural Language Learning. It originates from a series of shared tasks organized at the Conferences of Natural Language Learning.

The Cambridge Learner Corpus First Certificate in English (CLC FCE) dataset consists of short texts, written by learners of English as an additional language in response to exam prompts eliciting free-text answers and assessing mastery of the upper-intermediate proficiency level. The texts have been manually error-annotated using a taxonomy of 77 error types. The full dataset consists of 323,192 sentences. The publicly released subset of the dataset, named FCE-public, consists of 33,673 sentences split into test and training sets of 2,720 and 30,953 sentences, respectively. Source: Compositional Sequence Labeling Models for Error Detection in Learner Writing

JFLEG is for developing and evaluating grammatical error correction (GEC). Unlike other corpora, it represents a broad range of language proficiency levels and uses holistic fluency edits to not only correct grammatical errors but also make the original text more native sounding. Source: JFLEG: A Fluency Corpus and Benchmark for Grammatical Error Correction

CoNLL-2014 will continue the CoNLL tradition of having a high profile shared task in natural language processing. This year's shared task will be grammatical error correction, a continuation of the CoNLL shared task in 2013. A participating system in this shared task is given short English texts written by non-native speakers of English. The system detects the grammatical errors present in the input texts, and returns the corrected essays. The shared task in 2014 will require a participating system to correct all errors present in an essay (i.e., not restricted to just five error types in 2013). Also, the evaluation metric will be changed to F0.5, weighting precision twice as much as recall. The grammatical error correction task is impactful since it is estimated that hundreds of millions of people in the world are learning English and they benefit directly from an automated grammar checker. However, for many error types, current grammatical error correction methods do not achieve a high performance and thus more research is needed. Source: CoNLL-2014 Shared Task: Grammatical Error Correction Image source: Tou Ng et al.

WI-LOCNESS is part of the Building Educational Applications 2019 Shared Task for Grammatical Error Correction. It consists of two datasets: - LOCNESS: is a corpus consisting of essays written by native English students. - Cambridge English Write & Improve (W&I): Write & Improve (Yannakoudakis et al., 2018) is an online web platform that assists non-native English students with their writing. Specifically, students from around the world submit letters, stories, articles and essays in response to various prompts, and the W&I system provides instant feedback. Since W&I went live in 2014, W&I annotators have manually annotated some of these submissions and assigned them a CEFR level. Source: The BEA-2019 Shared Task on Grammatical Error Correction Image source: WI-LOCNESS

MuCGEC is a multi-reference multi-source evaluation dataset for Chinese Grammatical Error Correction (CGEC), consisting of 7,063 sentences collected from three different Chinese-as-a-Second-Language (CSL) learner sources. Each sentence has been corrected by three annotators, and their corrections are meticulously reviewed by an expert, resulting in 2.3 references per sentence.

AKCES-GEC is a new dataset on grammatical error correction for Czech. Source: Grammatical Error Correction in Low-Resource Scenarios

UA-GEC: Grammatical Error Correction and Fluency Corpus for the Ukrainian Language

Grammatical error correction dataset for text from Wikipedia.

Are you the kind of person who makes a lot of typos when writing code? Or are you the one who fixes them by making "fix typo" commits? Either way, thank you—you contributed to the state-of-the-art in the NLP field. GitHub Typo Corpus is a large-scale dataset of misspellings and grammatical errors along with their corrections harvested from GitHub. It contains more than 350k edits and 65M characters in more than 15 languages, making it the largest dataset of misspellings to date. Source: GitHub

Grammatical error correction dataset for text from Yahoo! Answers

Lang-8 Preprocessed Dataset (for GED): - Dataset: Lang-8, a publicly available dataset containing user-generated content, primarily from second-language learners, focused on writing errors. - Task: Grammatical Error Detection (GED). - Size: 200,000 sentences, with each sentence labeled as '0' for incorrect and '1' for its corrected version. - Preprocessing: The dataset has been cleaned and transformed to improve model performance, including removing noise, handling inconsistent annotations, and preparing it for training. - Usage: Ideal for training and evaluating models for GED and GEC.

• a fine-grained corpus to detect, identify and correct the chinese grammatical errors. * collected mainly from multi-choice questions in public school Chinese examinations * with multiple references * Online Evaluation Site for test set: https://codalab.lisn.upsaclay.fr/competitions/8020

Kor-Lang8 is a Korean grammatical error correction (GEC) dataset extracted from the NAIST Lang-8 Learner Corpora by the language label. It contains more than 109K sentence pairs. Source: Towards standardizing Korean Grammatical Error Correction: Datasets and Annotation Image Source: https://arxiv.org/pdf/2210.14389v2.pdf

Kor-Learner is a Korean grammatical error correction (GEC) dataset made from the NIKL learner corpus containing essays written by Korean learners and their grammatical error correction annotations by their tutors in an morpheme-level XML file format. It contains more than 28K sentence pairs. Source: Towards standardizing Korean Grammatical Error Correction: Datasets and Annotation Image Source: https://arxiv.org/pdf/2210.14389v2.pdf

Kor-Learner is a Korean grammatical error correction (GEC) dataset collected grammatically from two sources, and the correct sentences were read using Google Text-to-Speech(TTS) system. The general public was tasked with dictating grammatically correct sentences and transcribe them. It contains more than 17K sentence pairs. Source: Towards standardizing Korean Grammatical Error Correction: Datasets and Annotation Image Source: https://arxiv.org/pdf/2210.14389v2.pdf

NaSGEC is a new dataset to facilitate research on Chinese grammatical error correction (CGEC) for native speaker texts from multiple domains. Previous CGEC research primarily focuses on correcting texts from a single domain, especially learner essays. Source: NaSGEC: a Multi-Domain Chinese Grammatical Error Correction Dataset from Native Speaker Texts Image Source: NaSGEC: a Multi-Domain Chinese Grammatical Error Correction Dataset from Native Speaker Texts

We present a new annotated corpus of written learner English, derived from essays submitted to the learning platform Write & Improve (W&I). Users of W&I are presented with automated scoring and feedback on grammatical errors, and are encouraged to act on their error feedback, submitting multiple versions of their essays for any given prompt. We build the corpus on this interplay between users and prompts, collecting sets of essays submitted by users for a selected list of 50 popular prompts. The prompts include 20 aimed at beginner learners of English, 20 aimed at intermediate learners, and 10 at advanced learners. This distribution reflects the greater use of W&I by beginner and intermediate learners of English. We ensured that the prompts were not likely to elicit personal information and covered a broad range of tasks and topics. This list of prompts enabled us to identify 5050 essay sets written by 766 users, forming the basis for the Write & Improve Corpus, which is being made available for non-commercial use on the ELiT website. We describe the steps we took to ensure the corpus contains appropriate texts, does not include personal information, and will come with annotations relating to Common European Framework of Reference (CEFR) level and grammatical errors. All essays were submitted between 2020 and 2022 by registered users of W&I who have supplied their first language (L1) in an optional questionnaire. In total, there are more than 23K essays containing more than 3.5 million word tokens. The final versions of each essay set amount to 762K word tokens. There are 22 different L1s in the corpus, with the most common being Spanish, Portuguese, Japanese, Arabic and Vietnamese. We present some descriptive statistics for the corpus, and consider some research use cases.