Theses in Progress

Using Transformers for Automatic Short Answer Grading (ASAG)

November 20, 2019 – ,

Are we capable of automating teachers?

One of the most important jobs of teachers is assessing the learner’s knowledge. Typically, this is done by asking questions – verbally or in written form. The answers of the learner are then analysed with regard to the understanding of the topic they demonstrate. In the case of short answer grading, questions are posed to allow free-text answers and are typically answerable in a few words or sentences. Important criteria for the scoring of short answers are the correctness, completeness and relevance of the answers in the context of the questions.
The idea of automating the grading process of short answers has fascinated researches for many years. With the increasing success of deep learning approaches in many natural language processing (NLP) tasks, it stands to reason that they may also be useful for tackling this well-established problem.

Thus, the goal of this thesis is to investigate the usage of transformer models for automatic short answer grading.

Keywords: Deep Learning, NLP, Transformers, Automatic Short Answer Grading

Research Area(s): Knowledge & Educational Technologies

Tutor: Filighera, Christoph Rensing

Student: Yantao Shi

Theses in Progress