A cross-validated prevalence dataset of 87 cosmetic active ingredients from four public sources

<title>Abstract</title> <p>We present a prevalence dataset of 87 cosmetic active ingredients drawn from two ecological niches: an EU-leaning crowd-sourced corpus (Open Beauty Facts, n = 19,045)[1]; and North-American retail, spanning premium (Sephora, n = 1,472)[2] and multi-platform (e-commerce catalogue, n = 6,600 with INCI text)[3] channels; plus a US regulatory negative control (California Safe Cosmetics Program, n = 34,502)[4]. A reusable 87-concept INCI ontology was applied by a rule-based matcher validated on a 50-product gold standard (precision 1.00, recall 0.996). Pairwise Spearman correlations (common-filter ρ = 0.56–0.81; full-87 upper bound 0.93) indicate strong cross-channel agreement; removing the shared Sephora subset raises correlation (0.81→0.92 common filter), showing the signal is not an overlap artefact. The registry covers 3 of 87 concepts with no rank agreement (ρ = −0.03 to 0.13, non-significant). Formula skeletons (Eclat itemsets) are included and exploratory temporal analyses are provided at the end of this Descriptor; all rank statistics are tie-corrected. The dataset supports reproducible research on cosmetic ingredient co-occurrence and formulation structure.</p>