Knowledge Discovery Based on Sentiment Analysis of Public Perceptions About Generative AI on X

Abstract

Public discourse surrounding Generative Artificial Intelligence (GenAI) reflects diverse attitudes ranging from optimism to ethical concern, particularly as these technologies become increasingly discussed in educational contexts. This study examines public perceptions of GenAI on the social media platform X using a knowledge discovery approach that integrates multiple topic modeling techniques and Aspect-Based Sentiment Analysis (ABSA). A total of 111,675 English-language tweets collected between June 23, 2024, and June 23, 2025, were analyzed using five topic modeling methods BERTopic, Top2Vec, LDA, LSA, and NMF to identify dominant discussion themes and evaluate topic coherence. Sentiment toward specific GenAI aspects was subsequently examined using ABSA to capture fine-grained public attitudes. The results indicate that topics related to ethics and creativity are predominantly associated with negative sentiment, while innovation and cloud-related discussions show higher levels of positive sentiment. Education-related topics are largely characterized by neutral sentiment, suggesting exploratory and informational discourse. These findings highlight the importance of addressing ethical awareness, trust, and AI literacy in informatics education. By combining multi-model topic analysis with aspect-level sentiment interpretation, this study provides methodological insights and empirical evidence to support responsible GenAI integration in educational contexts.

Ähnliche Arbeiten