EssayGAN: Essay Data Augmentation Based on Generative Adversarial Networks for Automated Essay Scoring

Round 1

Reviewer 1 Report

In this paper, a new method for data augmentation of automatic essay scoring dataset is proposed. The proposed method works using different generators for each score. This structure makes the model complicated. The authors should state their proposed method’s motivation and benefits in comparison to other data augmentation methods. Specifically, comparing the proposed method with other GAN-based data augmentation methods can give the readers an insight into the benefits of this method. Overall the paper is well-written. I have some comments to improve the quality of the paper that are listed below:

-          It is recommended to evaluate the outcome of EssayGAN for the AES classification task to show its performance to score essays automatically.

-          The authors should compare their proposed method with other methods used ASAP dataset, especially the ones using GAN.

-          The programming language, libraries and computer resources used to run the experiments should be mentioned.

-          Computational complexity of the proposed method should be discussed in comparison with other methods.

-          More recent references can be used (2020-2022)

-          Conclusion section should be extended.

Author Response

First of all, we appreciated your considerate comments on the paper. Based on your comments, we have tried to revise the paper as accurately as possible.

The modifications of the manuscript are marked in blue so you can easily recognize the revisions.

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The manuscript should be restructured according to instructions to the authors. It should content follow chapters: Introduction, Materials & Methods, Results and Discussion, Conclusions. The results should be critically commented on and compared with already published studies. The claims should be supported by statistical analysis with a given significance level.

Author Response

First of all, we appreciated your considerate comments on the paper. Based on your comments, we have tried to revise the paper as accurately as possible.

The modifications of the manuscript are marked in blue so you can easily recognize the revisions.

Please see the attachment.

Author Response File: Author Response.pdf