Guard Completion for Autoformalizing Argumentative Material Inferences

The proposed method converts non-monotonic natural language arguments into monotonic formal inferences via explicitly constructed guard sets. This prevents the risk of systems adding arbitrary premises to make any claim provable during the autoformalization process.