Deep Grading: Incorporating Argument Strength into Neural Essay Evaluation

Author First name, Last name, Institution

Sarah Kohail, Zayed University
Omar Alfandi, Zayed University

Document Type

Conference Proceeding

Source of Publication

Proceedings of the 2024 16th International Conference on Education Technology and Computers

Publication Date

1-21-2025

Abstract

The automation of essay grading poses a significant challenge, especially when evaluating the aspects of argument strength in essays designed for discussion, debating and argumentation. Addressing this challenge, we developed a methodology for calculating argument strength, effectively reducing the impact of essay length variations. Leveraging this argument strength scoring methodology, we built a deep neural network model aimed at learning enhanced embedding representations for essays, capable of capturing the intricate nuances of argumentation, which encompass information beyond textual and structural characteristics. Our results demonstrate competitive performance, surpassing traditional and deep learning embedding baselines.

ISBN

9798400717819

Publisher

ACM

First Page

275

Last Page

279

Disciplines

Computer Sciences

Keywords

Argument strength, Neural network, Essay grading, Argumentation, Deep learning

Indexed in Scopus

no

Open Access

no

Share

COinS