An Event-Based Framework for Facilitating Real-Time Sentiment Analysis in Educational Contexts

Abstract

Sentiment analysis has been a hot topic nowadays that has been broadly applied in various disciplines such as media and finance, but its application to the education domain is limited to generating insights by applying existing methods to a selected corpus at the individual record level. Many educational data like student forum posts and ongoing course evaluation responses can be categorised as event data. However, insufficient attention is paid to the temporal and influential features of event data in these educational corpora. This paper proposes a novel event-based framework for addressing the complexity of the sentiment analysis process in the context of education. The framework features an event data model for educational sentiment analysis and an architecture that harnesses both sentiment analysis algorithms and the complex event processing technology, aiming to achieve timely warning and action on defined complex events. To validate the framework, a prototype is implemented and applied to detecting student emergency occurrences from university student forum posts.