GraphMind, a model that jointly processes micro- and macro-level structures for novelty prediction and rationale generation, is proposed that significantly outperforms baseline LLMs in both novelty score prediction and rationale generation.
Italo Luis da Silva, Hanqi Yan, Lin Gui et al.· Proceedings of the 32nd ACM...· 1 citation
Scientific papers may relate by problem, method, result, or contribution, but document-level retrievers collapse these into a single similarity score without saying why they are related. Citation- and similarity-based retrieval alone also confines search to the neighbourhood of what is already known, whereas generative retrieval generates document identifiers directly, enabling the exploratory retrieval that scientific discovery depends on. We connect papers in a graph whose edges are typed by these four facets, derived from facet items and citation signals, and distil it into a generative retriever whose identifiers are the papers'own facet text. Two graph properties do not survive naive distillation. First, because every training pair is an edge, naive enumeration indexes just 84% of the corpus. Coverage-aware distillation makes every paper learnable through a reverse-neighbour fallback, a minimum-coverage threshold, and edge-importance weighting. Second, constrained decoding guarantees that every generated identifier is a valid paper, but not that the graph connects it to the query. Graph-weighted reciprocal rank fusion scales each candidate's rank term by its query-candidate edge weight, dropping unsupported ones. On LitWeave, our constructed corpus of 11,359 NLP papers, Graft recovers 91% of its graph teacher's Recall@20 with no nearest-neighbour index or encoder at inference, and outperforms the graph teacher on query papers outside the corpus. It reproduces the graph's own facet labels at 0.922 precision, so every returned paper arrives labelled with the facet that surfaced it rather than an opaque score.
Italo Luis da Silva, Hanqi Yan, Yujing Wang et al.· 0 citations
Assessing scientific novelty is inherently complex, requiring evaluation of both a paper's internal structure and its contribution within the broader research landscape. Existing large language model (LLM) approaches often rely on surface-level similarity or citation retrieval, overlooking the integration between a paper's content and its contextual grounding in related literature. To address this gap, we introduce SciNova, a benchmark containing 3,063 papers from both ICLR and NeurIPS, with full content, bibliographies, and peer review scores for novelty prediction. Building on this benchmark, we propose GraphMind, a model that jointly processes micro- and macro-level structures for novelty prediction and rationale generation. It represents each paper as a hierarchical graph that captures its claims, methods, and experiments (micro-level), while its related paper graphs encode both cited and semantically similar works (macro-level). Experiments show that by incorporating this bi-level graph, GraphMind significantly outperforms baseline LLMs in both novelty score prediction and rationale generation.
Italo Luis da Silva, Hanqi Yan, Lin Gui et al.· Proceedings of the 32nd ACM...· 1 citation
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
