Analisis Konseptual Tax Sentiment Intelligence untuk Meningkatkan Kepatuhan Pajak di Indonesia

Abstract

Purpose: This study conceptually examines the application of Tax Sentiment Intelligence based on artificial intelligence as an innovation to improve tax compliance in Indonesia while exploring the relationship between trust in tax authorities, public perception in digital spaces, and taxpayer compliance.
Method: A systematic literature review for the period 2020–2025 was conducted using Google Scholar, OpenAlex, and CrossRef via Publish or Perish, and article selection followed the PRISMA protocol, resulting in 11 relevant studies.
Results: The bibliographic analysis identified two main clusters with weak interconnections: the methodological cluster of sentiment analysis techniques and the applied cluster of technology in public administration and taxation, with the framework and system nodes serving as connectors that underscore the urgency of an integrative framework. The literature synthesis indicates that negative public sentiment on social media correlates with weakened tax morale and declining voluntary compliance, while positive perceptions of tax authorities significantly promote compliance. Trust in tax institutions acts as the primary mediator between public perception and compliance behavior, but it can be undermined by perceptions of corruption or negative narratives circulating in the digital space. The integration of AI and NLP enables real-time sentiment monitoring, detection of shifts in collective sentiment, and early warning functions for the Directorate General of Taxes
Implications: The study recommends developing a Tax Sentiment Dashboard that combines diverse data sources, validation and cleansing mechanisms, locally adaptive algorithms, and interpretation modules that link sentiment indices with compliance indicators.
Novelty: This research proposes an initial conceptual framework for applying AI-based sentiment analysis in the Indonesian taxation context.