Tag
This paper proposes a context-augmented training framework for multi-hop question-answering, showing that combining context graphs with knowledge graphs and using reinforcement learning improves performance in biomedical domains.