Korean Grammatical Error Correction Based on Transformer with Copying Mechanisms and Grammatical Noise Implantation Methods

Abstract

Grammatical Error Correction (GEC) is the task of detecting and correcting various grammatical errors in texts. Many previous approaches to the GEC have used various mechanisms including rules, statistics, and their combinations. Recently, the performance of the GEC in English has been drastically enhanced due to the vigorous applications of deep neural networks and pretrained language models. Following the promising results of the English GEC tasks, we apply the Transformer with Copying Mechanism into the Korean GEC task by introducing novel and effective noising methods for constructing Korean GEC datasets. Our comparative experiments showed that the proposed system outperforms two commercial grammar check and other NMT-based models.

Performance Evaluation and Improvement of Deep Echo State Network Models in English Writing Assistance and Grammar Error Correctionn

“…Throughout the development of grammatical error correction, grammatical error correction can be divided into three categories, namely, rule-based grammatical error correction, classifier-based error correction, and machine translation-based error correction [7]. Literature [8] uses rule-based error correction, but the cost is very high and complex uncertainty; literature [9] through a large amount of text data, the use of recurrent neural networks from which to learn deep features such as grammar and context; literature [10] uses a bidirectional recurrent neural network to represent the context of the target word, based on the features to predict the target this from the candidate sequences, and at the same time use the attention algorithm to capture the words in the sentence between the dependencies, which improves the accuracy of error correction; literature [11] carries out training and word selection for different grammatical error types using a bidirectional long and short time domain memory network, and the learning results are significantly due to other models; literature [12] proposes an error correction method by extracting task-specific sentence features, and assembling machine translation algorithms based on largescale networked corpus language models; and literature [13] proposes a discriminative reordering model that using sentence phrase syntactic features to deal with local contextual error effects; Literature [14] applies neural machine translation methods to the task of grammatical error correction, and for the occurrence of unregistered words in the sentence, through unsupervised alignment and word-level translation models. There are many current grammar error correction design methods, but there are fewer such studies on how to go about judging the effectiveness of error correction methods in English writing tutoring, and the performance evaluation methods are all qualitative and lack quantitative analysis methods.…”

Towards Harnessing the Most of ChatGPT for Korean Grammatical Error Correction

“…Deep learning has recently emerged as a potent tool for grammatical correction, notably overcoming the challenges of prior approaches. Methods have been introduced specifically for Korean, enabling error correction model training without parallel data [1,7,8]. These primarily rely on noise injection processes that yield pseudo-parallel corpora from unlabeled mono corpora [1,9].…”

Adversarial Grammatical Error Generation: Application to Persian Language

“…In [26], Kasewa et al used neural machine translation to learn the distribution of language learner errors and utilized it as a data augmentation technique to induce similar errors into the grammatically-correct text. In [27], Lee et al introduced a Korean GC model based on transformers with a coping mechanism. They also used a systematic process for generating grammatical noise which reflects general linguistic mistakes.…”