Developing a User-Centric Data Labelling Tool for Sentiment Analysis : A Comparative Study of Explainable AI and Non-Explainable AI

Abstract

Sentiment analysis, especially within platforms abundant in user-contributed content such as social media, has gained paramount importance. However, given the intricate nuances of human emotions, the 'human-in-the-loop' (HITL) methodology emerges as crucial, combining human intuition with machine-driven precision to achieve greater sentiment labelling outcomes. This research contrasts the transparent nature of Explainable AI (XAI) against the more unknowable non-explainable AI systems. Initial findings highlight a marked user inclination towards XAI.