Neural Sequence Modelling for Essay Scoring

Automated Essay Scoring (AES) is the use of specialised computer software to assign scores for essays written in an academic environment. Its growing interest has been motivated by several factors including rising costs of education, need for grading standards, and major technological breakthroughs. Despite the positive results in literature, there still remain many critical challenges that need to be addressed to ensure the wide-spread adoption of AES systems. These challenges can be divided into three main categories: meaningfulness, transparency, and robustness.

This investigation aims to address these challenges while also attempting to improve the human-machine inter-rater agreement. Motivated by the recent success of neural networks, we conduct a systematic investigation of deep representation learning; initially using a basic recurrent neural network (RNN) but extending to Long-short term memory cells and deep bi-directional architectures as well. In order to evaluate the AES system, an adapted visualisation technique was implemented. The visualisation identifies portions of the text that are discriminative of writing quality.

Overall it was found that deep bi-directional model, DBLSTM, are more e↵ective in capturing features discriminative of writing quality than shallower uni-directional models. Although the results did not surpass existing stateof- the-art, our methodology lays the foundations for a potentially rewarding avenue for future AES systems.