Atalaya: A Calibrated Multi-Evidence Affinity Graph over an Open-Data Catalog, with an Empirical False-Discovery Gate

Open-data portals publish thousands of datasets as a flat searchable list, but the relational knowledge that makes the data useful, which datasets can be joined, which describe the same thing, and which actually correlate when aligned by a common key, stays implicit and is infeasible to recover by hand. Existing dataset-discovery tools each rank relatedness on a single signal (value containment for joinability, embedding cosine for semantic search, statistical association for correlation mining), and each signal is individually misleading: two datasets can be embedding-similar yet share no join key, joinable on year yet describe unrelated phenomena, or correlate spuriously through a shared driver. This preprint presents Atalaya, a system that harvests a national open-data catalog (1,017 datasets from Chile's Data Observatory; a 303-dataset, 24.8 GB government subset mirrored and profiled in full), mines five orthogonal kinds of cross-dataset relation into an explorable knowledge graph, and ranks pairwise relatedness with a novel calibrated multi-evidence affinity. The affinity fuses three orthogonal evidences into one interpretable [0,1] score after passing each through its empirical null distribution, so the score means 'stronger than chance' rather than 'large raw number', and it down-weights any evidence that contradicts the others. Two design choices make the graph trustworthy rather than merely plausible. First, the correlation edges pass an empirical false-discovery gate: on the same alignments with the keys shuffled, 343 candidate correlations yield 0 survivors under a permutation null with Benjamini-Hochberg control (empirical false-discovery rate 0.00), so the 24 correlations reported on real data are believable. Second, the semantic encoder is measured honestly against a classical lexical baseline, and the modest true margin is reported (+0.014 top-5 neighbour-theme coherence over TF-IDF) rather than oversold. Every edge carries its evidence, the whole affinity is pure numpy so the browser recomputes it live as an analyst reweights the evidences, and all figures are regenerated deterministically from the committed artifacts. The manuscript gives the method, the adversarial validation, and an explicit account of what the graph does and does not license. Code, committed artifacts, and a live instance are open (MIT): https://github.com/fsantibanezleal/CAOS_ATALAYA ; https://atalaya.fasl-work.com .