<![CDATA[This study employs Artificial Intelligence (AI) to examine public sentiment and emotion surrounding Indonesia’s dormant bank account regulation issued by the Financial Transaction Reports and Analysis Center (PPATK). Drawing on 3,028 YouTube comments, the study addresses a gap in Indonesian public policy research, where social media analysis has largely relied on basic sentiment polarity without incorporating psychology-based mood-state models. We develop an integrated AI-driven analytical framework combining Latent Dirichlet Allocation (LDA) topic modelling (k = 13 clusters), lexicon-based sentiment scoring visualized through a heatmap, and an adapted Profile of Mood States (POMS) multiclass emotion classification scheme for Indonesian-language discourse. Rather than merely combining techniques, the framework operationalizes a layered analytical structure linking thematic clustering, polarity intensity, and differentiated mood-state profiling within a unified workflow. Statistical testing confirms that the observed emotional distribution significantly deviates from a uniform pattern (χ² = 15140.00, dof = 5, p < 0.001). The findings indicate that Depression (n = 2121) and Confusion (n = 603) dominate the discourse, suggesting that public responses are characterized more by hopelessness and uncertainty than overt hostility. Conceptually, this study advances policy discourse analysis by integrating psychology-based mood-state interpretation into digital public opinion research, enabling a more granular understanding of how regulatory decisions resonate emotionally within developing country contexts. Operationally, the results demonstrate how emotion-based analytics can inform stages of the policy cycle, particularly agenda-setting and communication evaluation, by identifying dominant emotional signals that may indicate risks to institutional trust. These findings provide structured empirical insight into the emotional dimensions of financial regulation debates while acknowledging the need for continued methodological refinement.]]>
