Predicting New Concept--Object Associations in Astronomy by Mining the Literature

We construct a concept--object knowledge graph from the entire astro-ph corpus up to July 2025, using an automated pipeline to extract named astrophysical objects from OCR-processed papers, resolve them to SIMBAD identifiers, and link them to scientific concepts annotated in the source corpus. We then ask whether the historical structure of this graph can forecast new concept--object associations before they appear in print. Because the underlying concepts are derived from clustering, semantically related concepts inevitably overlap; we address this by applying an inference-time concept-similarity smoothing step uniformly to all methods. Evaluating across four temporal cutoffs on a physically meaningful subset of concepts, we find that an implicit-feedback matrix factorization model (Alternating Least Squares; ALS) with concept smoothing outperforms the strongest neighborhood baseline (KNN based on text-embedding concept similarity) by 16.8\% on NDCG@100 (0.144 vs.\ 0.123) and 19.8\% on Recall@100 (0.175 vs.\ 0.146), and exceeds the best recency heuristic by 96\% and 88\%, respectively. These results indicate that historical literature encodes predictive structure not captured by either global heuristics or local neighborhood voting, suggesting a path toward tools that could help triage follow-up targets for costly telescope resources.