Sentiment analysis and opinion mining on educational data: A survey
Project Overview
This document examines the role of generative AI and sentiment analysis in education, highlighting how advanced AI and natural language processing (NLP) techniques can significantly improve the analysis of student feedback to enhance teaching and learning practices. It details various levels of sentiment analysis, including document, sentence, entity, and aspect analysis, and discusses sentiment annotation techniques that can inform educational decision-making and evaluation. The findings suggest that effective sentiment analysis can lead to more tailored pedagogical strategies, ultimately benefiting both educators and learners. However, the document also addresses challenges such as handling negation, detecting opinion spam, and managing multi-polarity in feedback, which can complicate the analysis process. It identifies potential future research directions to further refine these methodologies and improve their application in educational contexts. Overall, the insights provided underscore the transformative potential of generative AI in optimizing educational experiences through enhanced feedback analysis.
Key Applications
Sentiment analysis tools for educational feedback
Context: Used in educational institutions to analyze student feedback surveys and enhance pedagogical practices.
Implementation: Development of tools using machine learning, deep learning, and transformers for sentiment classification.
Outcomes: Improved understanding of student opinions, enhanced teaching practices, and informed decision-making.
Challenges: Challenges in sentiment annotation, multi-polarity of feedback, and the need for domain-specific tools.
Implementation Barriers
Technical Challenge
Difficulty in handling negation words in sentiment analysis which may lead to misinterpretation of sentiments.
Proposed Solutions: Developing models that can recognize and account for negation in text.
Data Quality Issue
Opinion spam detection is critical as students may provide fake feedback.
Proposed Solutions: Automated systems for spam detection and using deep learning techniques to identify spam comments.
Complexity in Feedback
Multi-polarity in feedback where students express mixed sentiments towards different aspects of a course.
Proposed Solutions: Using dictionary-based sentiment knowledge and advanced NLP techniques to accurately capture mixed sentiments.
Project Team
Thanveer Shaik
Researcher
Xiaohui Tao
Researcher
Christopher Dann
Researcher
Haoran Xie
Researcher
Yan Li
Researcher
Linda Galligan
Researcher
Contact Information
For information about the paper, please contact the authors.
Authors: Thanveer Shaik, Xiaohui Tao, Christopher Dann, Haoran Xie, Yan Li, Linda Galligan
Source Publication: View Original PaperLink opens in a new window
Project Contact: Dr. Jianhua Yang
LLM Model Version: gpt-4o-mini-2024-07-18
Analysis Provider: Openai