Automated Domain Specific Essay Scoring

Automated essay scoring (AES) task involves using computer technology to grade written assessment and assigning a score based on its perceived quality. AES has been among the most significant Natural Language Processing (NLP) applications especially due to its educational and commercial value. Accordingly, AES is a well-studied topic with previous studies focusing on improving the performance via state-of-the-art deep learning and NLP methods to more accurately score the essays. The majority of these studies rely on publicly available corpora and limited datasets. However, domain-specific essay scoring tasks (e.g., law and accounting exams) would benefit from customized NLP methods for improved performance. In this research, we aim to develop domain-specific NLP methods for improved AES. Specifically, we will perform this analysis in 3 dimensions: domain-specific models, interpretable deep learning and weak supervision. At the end of the project, via our investigations in these three dimensions, a modern AES will be established

Faculty Supervisor:

Mucahit Cevik

Student:

Partner:

Blees AI

Discipline:

Engineering

Sector:

Information and cultural industries; Professional, scientific and technical services

University:

Toronto Metropolitan University

Program: