Review analysis and summarization

Abstract

This Project focuses on our ongoing work to perform Aspect-Based Sentiment Analysis (ABSA) and summarization on academic reviews and feedback.The goal is to extract insights by analyzing sentiments associated with specific aspects of course feedback and providing concise summaries for better usability. Particularly in this report, we focus on collecting and annotating course feedback to better understand sentiments regarding various aspects of a course. The annotation process involves tagging feedback with aspect terms, categories, sentiments, and opinion terms, facilitated by a personalized annotation tool. This tool streamlines the process with an intuitive interface and a detailed taxonomy for organizing categories. To ensure consistency and quality, similarity testing was conducted using metrics like the F1 score and confusion matrices, helping refine the annotation guidelines and improve inter-annotator agreement (IAA).