Students’ comments classification and opinion mining on instructor’s performance assessment

Abstract

Assessing instructor’s performance has become increasingly necessary because of the ever-changing demands for quality education in the modern day society. This paper describes the design and development of an assessment tool that analyzes textual student’s comments written in Taglishuano automatically. It also attempts to prove the usefulness of a simple Convolutional Neural Network (CNN) as a classifier for limited comments corpus and compare its performance to the baseline models such as Multinomial Naïve Bayes and Support Vector Machines. The result shows that baseline models work well on top of students’ comments corpus. Also, CNN indicates a promising task for opinion mining without using pre-trained vectors. Adding more instances and fine tuning the datasets and with little modification of CNN architecture and hyperparameters better results can be achieved. The proposed system which shows how bar chart and word clouds visualization techniques in addition to the quantitative analysis can aid in getting unique insight into an instructor’s performance.