Tag
MaSCoD is a multi-agent framework that pre-organizes structural hypotheses before direct-edge judgment to improve retention of potentially relevant causal relations in candidate graph generation, evaluated on datasets using GPT-5.4 and GPT-4o.