Sentiment analysis of student feedback: A comparative study employing lexicon and machine learning techniques

Abstract

One of the most pressing concerns in contemporary education examines the integration of big data and artificial intelligence methodologies to enhance the educational learning outcome. Towards that purpose, it is imperative to leverage the unstructured data originating from student feedback, particularly in the form of comments to open-ended questions aiming to extract emotions and opinions conveyed within their messages. Our research goal is to ascertain the most efficient approach to tackle this difficult task by conducting a comparison of sentiment analysis methods, including Machine Learning (ML) and Lexicon based models. Both lexicon based and machine learning approaches were implemented using an open source data mining platform while also utilizing student comments submitted at the end of academic semesters. Our study reveals a promising approach that effectively addresses the issue at hand, particularly within the domain of educational data. Additionally, it emphasizes the key aspects that led to the selection of this approach effectively highlighting the weaknesses and strengths inherent in each method