Data and artifacts for "Viewpoint–Task Alignment: When and Why Pattern-Derived Features Improve Symbolic Folk-Music Classification" (under review, TISMIR)

Derived data artifacts accompanying a manuscript under review at Transactions of the International Society for Music Information Retrieval (TISMIR). Two archives are included: - patterns-artifacts-v1.zip: pattern-mining feature matrices, per-corpus label CSVs, mined pattern breakdowns, and permutation-importance tables for four symbolic folk-music corpora (SymbTr, The Session, the Essen Folksong Collection, and a reconstructed Dance-9 benchmark). The analysis reproduces every table and figure in the paper from these feature matrices and labels, with no dependency on the pattern-mining implementation. - paper-folk-patterns-data-results.zip: the source score data and the full corrected results tree (classification scores, statistics, feature statistics, and figures). Released under CC-BY 4.0. Author and repository details are withheld for double-anonymous peer review. ===== Changes in version 1.1. Corrected the Essen rows of the aggregated viewpoint-comparison tables and the four figures derived from them: both cells of every viewpoint contrast now use each corpus's fixed classifier, whereas three Essen rows previously compared a gradient-boosting global against a logistic-regression combined. The affected files are results/viewpoint_comparison/ablation_significance.csv, results/stats_corrected/ablation_corrected.csv, results/viewpoint_comparison/variance_comparison.csv and results/figures/fig1–fig4. Substantive conclusions are unchanged: no Essen viewpoint configuration is significant after correction. patterns-artifacts-v1.zip is unchanged from version 1.0.