Retrieval protocol and full-text extraction design from four AI-assisted instruments chosen to fail differently, supporting an integrative review of the remediation of hazardous tailings

Abstract

This deposit contains the inputs for an integrative review on the remediation of hazardous tailings, metal-, radionuclide-, and reagent-bearing residues from mineral processing that remain sources of contamination long after closure. The review asks whether remediation evidence moves from laboratory containers to whole facilities and receiving landscapes and whether the endpoints measured at a small scale are those that closure and discharge criteria read. The procedure builds on an earlier one (Zenodo 10.5281/zenodo.22230688) in which retrieval instruments chosen to fail differently are run against a predefined review structure. Four AI-assisted instruments were used. Field-tagged searching in the Web of Science Core Collection defined a corpus of 1,958 records, which was crossed with query blocks for each indicator axis. CiteSpace was used to map the co-citation structure. Consensus answered 39 claim-level questions. Elicit extracted 21 fields, each with its supporting quote, from the full texts of 52 papers, 35 targeted at rare facility- and landscape-scale cases and 17 drawn at random from abstract strata. Each quote was then automatically searched in its PDF, and codes requiring judgement were checked by a second language model, with no manual checking of cells. The deposit contains the control file of 22 September 2026, the query architecture with dated counts, the exported record sets, the claim-level questions and their exports, the coupling of instruments to review sections, and the full-text selection and prompts. These inputs can be reused for syntheses on tailings, mine waste, and contaminated land, or reprocessed with new questions; the processed data accompany the review.