This three-year, $597,369 project grant from the National Science Foundation's Division of Information and Intelligent Systems, under the Computer and Information Science and Engineering program (CFDA 47.070), will support the development of computational tools to automatically understand and reason with tax law documents at The Johns Hopkins University. The project aims to advance natural language processing and automated reasoning capabilities for the legal domain, with a focus on mapping tax statutes and case law into machine-interpretable rules to identify potential tax avoidance strategies and their unintended enablement. Researchers will build benchmark datasets, extend recent work on converting text to structured representations that support inference, and develop legal ontologies and models to analyze U.S. case law. Progress is expected in areas such as semantic parsing, information extraction, schema induction, textual inference, and domain-specialized language model pre-training. Outcomes may include new perspectives on creating and applying legal language using advances in AI and computational statutory reasoning.