Explainable Ai Frameworks For Decision Support In Data Governance To Enhance User Trust And Compliance

Abstract

This study sets out to understand the contribution of Explainable AI (XAI) techniques to data governance decision support with respect to organisational trust, accountability, transparency and compliance outcomes. To address this objective, a systematic literature review was performed on scientific publications and relevant policy documents. The retrieved articles were analysed with relevance coding to XAI methods, data governance applications and trust-related implications. A thematic analysis was then applied to group results into categories that allow for contrasting and comparing the different XAI approaches and their impact on data governance.

The study revealed that four major XAI approaches exist (feature attribution, rule extraction, counterfactual explanations, and interpretable-by-design models), and they all contribute differently to governance workflows. The feature attribution methods, which include SHAP (SHapley Additive exPlanations) or LIME (Local Interpretable Model-agnostic Explanations), contribute to the transparency and accuracy of trust through the recognition of decision pathways, especially in relation to privacy-related risks. Rule extraction improves auditability and accountability by generating human-readable logic in finance and health governance to reinforce compliance. The counterfactual method allows fairness audits to be improved through bias mitigation, but there is still uncertainty concerning the conformation to discrimination law. On the other hand, the Interpretable models directly integrated transparency into the design of system architecture for public administration. User studies, trust scales, usability indicators, and audit outcomes were all evaluated, and they all showed improved stakeholder confidence and compliance readiness. IBM‘s AI Fairness 360, and Microsoft‘s Responsible AI dashboard which shows improved audit breaches and better organizational decision-making, are real-life cases, that back these findings. 

In general, trust, accountability and compliance are attributes strengthened by XAI, which has been proven to improve data governance. Companies must take the double approach: standardise formats and train stakeholders. Policy makers should, while making a regulatory framework, include XAI.