Data for "Generalization and the Rise of System-level Creativity in Science": paper-level Foundation / Extension / Generalization indices (OpenAlex and Semantic Scholar)

Paper-level data for Hongbo Fang and James Evans, Generalization and the Rise of System-level Creativity in Science, arXiv:2510.03240. Only a filtered subset of papers is included: papers with at least 5 citations within 5 years of publication (citation_count_y5 ≥ 5) and at least 1 reference (reference_count ≥ 1), published 1945–2019. This leaves 23,448,431 of 191,457,232 OpenAlex papers and 28,325,016 of 218,455,234 Semantic Scholar papers. Papers outside these criteria are not in the files. Paper-level Foundation / Extension / Generalization (F / E / G) indices, computed with a 5-year citation window, for two bibliographic corpora: OpenAlex (23,448,431 papers) and the Semantic Scholar Academic Graph (28,325,016 papers). Each citing paper is classified by comparing how many of the focal paper's references it also cites with how many of the focal paper's other citing papers it also cites: Extension (more upstream), Foundation (more downstream), or Generalization (neither). The indices are the shares of each kind among the focal paper's 5-year citations. Columns: id, title, pub_year, reference_count, citation_count_y5, e_y5, f_y5, g_y5. See README.md for details, provenance and source-data versions (OpenAlex snapshot 2025-01-29; S2AG release 2025-12-09). Code: https://github.com/fhbzc/Generalization-replication Source data: OpenAlex (CC0); Semantic Scholar Academic Graph (ODC-BY 1.0; Kinney et al., 2023, The Semantic Scholar Open Data Platform).