Search papers, labs, and topics across Lattice.
This paper introduces PaperPilot, a multi-turn literature search agent that utilizes workflow induction to enhance the scientific literature search process. By constructing an executable directed acyclic graph (DAG) of paper-search operators and refining it through user feedback, PaperPilot significantly improves search performance metrics, including Hit@5 and MRR, while minimizing execution errors. The results underscore the effectiveness of explicit and editable search workflows in aligning search agents with the nuanced and evolving intents of users in scientific research.
PaperPilot transforms scientific literature search by enabling users to iteratively refine their search strategies through an interactive workflow, achieving a remarkable reduction in execution errors.
Scientific literature search often requires more than retrieving papers from a single query: users'intents are underspecified, preference-dependent, and evolve through interaction. Existing search agents typically rely on fixed pipelines or implicit language-only reasoning, making their search strategies difficult to control, inspect, and refine. We introduce PaperPilot, a multi-turn literature search agent that frames scientific search as workflow induction. Given an anchor paper and a user query, PaperPilot constructs an executable DAG of paper-search operators, including keyword search, citation expansion, filtering, scoring, reranking, and evidence extraction. User feedback is then used to refine both the query and the workflow itself. We train PaperPilot with supervised workflow imitation and preference optimization over controlled workflow corruptions. Experiments show that PaperPilot-9B improves over the base Qwen3.5-9B toolset agent under multi-turn interaction, increasing Hit@5 from 58.0 to 77.0, MRR from 47.5 to 59.4, and nDCG@10 from 26.8 to 32.5, while reducing workflow execution errors from 9.5% to 0%. These results show that explicit, editable search workflows provide an effective and controllable interface for aligning literature search agents with complex scientific intent.
Stanford HAI