Cognitive Policy Engine: AI-driven Policy-to-Rule Transformation for Social Benefit Eligibility Systems

This paper introduces a Cognitive Policy Engine (CPE) that utilizes advanced AI technologies to automatically read and analyze social program policies (e.g., SNAP, Medicaid) and generate structured eligibility rules. These rules are used for the automated evaluation of applicants and officer-assisted decision support. By leveraging cloud-based NLP services, the CPE performs entity extraction and dependency parsing to convert policy text into machine-readable JSON format. This enables the seamless evaluation of applicant eligibility and provides human readable, explainable outputs for program officers. A proof-of-concept demonstrates the feasibility of this system on multiple policies, generating structured rules and delivering actionable insights that improve efficiency and transparency in decision-making. The proposed framework demonstrates how AI-driven policy-to-rule transformation can be operationalized responsibly for large-scale public benefit administration with human-in-the-loop governance.