A Comprehensive Approach to Automated Essay Scoring through Weak Supervision, AutoML, and Interpretability

Scoring essays is important for learning and business. With AI and machine learning, we can save time and resources by automating the process. This project looks at automatic essay grading from three angles. First, we generate more data using automatic methods called data programming. This helps improve the machine learning models. Then, we refine the data using NLP techniques. Second, we will use Auto Machine learning techniques to easily determine the parameter for training. Third, we make the model more understandable. Deep learning models are often hard to understand, so we’ll use methods to explain the decisions made by the algorithm. We’ll focus on local and post-hoc explainability models. For data programming, we’ll use weak supervision. For interpretability, we’ll use methods like SHAP and LIME.

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: