Self-Healing Test Automation Dataset: 11,100 Experiment Rows with DOM-Similarity ML Models

Version 2.0.0 of the self-healing test automation research dataset. This release adds a cross-Subject-Under-Test (SUT) mutation corpus derived from OWASP Juice Shop: 192 broken-locator events across 6 pages and 4 mutated versions, applying the same 7 refactoring classes (id_change, attribute_removed, placeholder_change, class_change, element_reorder, wrapper_added, name_change) used for the SauceDemo headline SUT in v1.0.0. The cross-SUT addition addresses the external-validity threat of single-application methodology. The two SUTs differ in framework family (Angular Material vs plain HTML), routing model (SPA hash-router vs multi-page), component library, and DOM depth (median 18 vs median 7), providing a structurally distinct second subject. The new Juice Shop content is packaged as zenodo_v2_juiceshop.zip, containing: - juiceshop_sut2/juiceshop_v1/ through v5/ : 30 HTML pages (6 pages x 5 versions) - juiceshop_sut2/mutation_report_juiceshop.csv : 192 mutation events - juiceshop_sut2/juiceshop_summary.json : preliminary HSR summary - juiceshop_sut2/juiceshop_hsr_results.csv : per-mode HSR with Wilson 95% CI - juiceshop_sut2/README_juiceshop_sut2.md : subfolder documentation - DATA_README.md : v2.0.0 release notes (replaces v1 README) - CITATION.cff : updated citation info for v2.0.0 All v1.0.0 data files (heal_decisions.csv, heal_events.csv, healing_ranker_gb.pkl, healing_ranker_xgb.pkl, model metrics JSON files, healing_training_data.csv, model_comparison.json, mutation_report.csv, and all 5 result CSVs covering heuristic, ML-GB, ML-XGB, hybrid-GB, hybrid-XGB ranking configurations - 11,100 experiment rows in total) are preserved unchanged. Companion paper: "Self-Healing Test Automation via Tree-Ensemble Locator Ranking: An Empirical Evaluation on 2,400 Mutation Events Across Seven Refactoring Classes" - under peer review at Wiley STVR, submitted 14 June 2026. Honest methodology caveat: end-to-end Juice Shop HSR figures require retraining the Gradient Boosting and XGBoost rankers against the Juice Shop feature distribution. This v2.0.0 release publishes the cross-SUT corpus and mutation pipeline; full retraining is marked as immediate follow-up work and will appear in a subsequent version. License: CC-BY-4.0 Source code: https://github.com/javvadivijayprasad/self_healing_stage_a