Data and code for "The fear of becoming obsolete from automation to AI, 1965–2026: citation structure, measurement fragmentation, and a four-axis specification"

This deposit contains the data, derived matrices, screening records and analysis code needed toreproduce the bibliometric analyses and the item-level measurement audit reported in the article"The fear of becoming obsolete from automation to AI, 1965-2026: citation structure, measurementfragmentation, and a four-axis specification". The review follows SPAR-4-SLR and PRISMA 2020. Search date: June 2026 (Scopus). The corpus wasfrozen at the search date; every reported analysis was computed on that frozen export rather thanon live queries. CONTENTS01_protocol Review protocol specified before data collection (SPAR-4-SLR stage plan, inclusion and exclusion criteria, analysis plan).02_corpus Corpus A (428 records) and Corpus B (340 records), 19 licence-safe bibliographic metadata fields each.03_screening_and_reliability Screening decisions for all 784 retrieved records with exclusion codes E1-E8; both double-coding reliability rounds; kappa input file.04_derived_data Item-similarity matrix (1,023 cross-instrument pairs of 49 threat-register items, raw and agent-masked Jaccard); OpenAlex replication input and output; keyword thesaurus and stop-terms; full-text access status.05_scripts Eight R and Python scripts: corpus construction, full-text retrieval, cross-citation matrix with E-I index and year-stratified permutation nulls, historiograph, Salton-normalised coupling network, coupling chord diagram, author-level bridging, OpenAlex replication.06_measurement_audit Four-axis coding form (agent, referent, register, horizon) and the manuscript reference list in RIS format. NOT INCLUDED, AND WHYScopus abstracts and cited-reference strings are licensed content and are not redistributableunder the Elsevier licence; they have been removed from every file, including the screeningworkbooks. Author affiliations, correspondence addresses and author identifiers were removed aspersonal data. Full-text PDFs of the reviewed articles are third-party copyrighted works and arenot deposited. The DOI and EID of every record are retained, so any user with Scopus access canre-download the full records and reproduce the pipeline exactly. LICENSINGData files, derived matrices, coding forms and documentation: CC BY 4.0.Scripts in 05_scripts/: MIT Licence (see LICENSE.txt in the package).Bibliographic metadata originate from Scopus (Elsevier) and OpenAlex (CC0). SOFTWARE VERSIONSR 4.6, bibliometrix 5.0 / biblioshiny, Python 3.11 (pandas, networkx, matplotlib).