From Mechanism Discovery to Proposal Closure: Graph-Grounded Hierarchical Search for Scientific Ideation

Large language models can generate research ideas that appear novel, but novelty alone does not determine whether an idea is worth following up. In practice, researchers also need ideas that are feasible, experimentally testable, and consistent with the evidence they build on. We therefore assess early-stage proposal quality through four dimensions: novelty, feasibility, experimental design, and evidence consistency. We formulate scientific ideation as graph-grounded hierarchical search over a heterogeneous evidence graph. Constructed at the method level, the graph encodes core methods, components, limitations, future work, baselines, datasets, and their relations. On top of this representation, we propose a two-stage framework: the first stage explores candidate mechanism directions over evidence-backed topic and problem frames, and the second stage refines promising candidates through evidence-grounded closure, binding ideas to concrete baselines, datasets, and experimental protocols. In controlled evaluation on problem frames drawn from computer science literature, we compare methods under matched problem frames and our four-dimensional protocol. The proposed framework produces stronger ideation-stage proposals than direct LLM generation and text-retrieval baselines, with the clearest gains in feasibility, experimental design, and evidence consistency while novelty remains competitive. Ablations further suggest that these gains depend on the combination of structured evidence, hierarchical search, and proposal closure.