Replication package: Would an AI reference recommender have found what information scientists cited? A temporally split benchmark across 45 years of Journal of Information Science

Version 1.0.2 corrects the README citation to the concept DOI (10.5281/zenodo.23907010), so the package always points to the latest version. No analysis code or result changed. Replication package for a study that tests whether AI reference recommenders recover the references that information scientists actually cited. The benchmark is built from every machine-readable reference list in Journal of Information Science (1979–2026), retrieved from the public OpenAlex API on 2026-11-03. Contents: benchmark construction and evaluation code; configuration and seeds; per-query metrics, bootstrap intervals, paired tests, ablations, robustness checks, error analysis and six figures; data manifest; README with reproduction instructions. Scope: 2,315 citing articles, 54,073 candidate works and 71,473 positive citation pairs; 545 test articles from 2021 onward. Systems: BM25, TF-IDF, BGE-small, a bibliometric heuristic and reciprocal rank fusion. Licence: code MIT; documentation, derived aggregates and figures CC BY 4.0; source metadata CC0 1.0 (OpenAlex). Raw abstract text is not redistributed.