Data and code for: A PRISMA-Guided Evidence Map of Business Intelligence, Artificial Intelligence, Natural-Language-to-SQL and Ubiquitous Computing

Data and analysis code supporting a PRISMA 2020 systematic mapping study of four research domains: business intelligence (including decision support), artificial intelligence, natural-language-to-SQL, and ubiquitous computing (including context-awareness). Contents. The search and screening records for both stages, the 5,826-record deduplicated pool and its decisions, the analysed 2,857-record corpus with its cluster codings under three schemes, the gap matrices, the truncation-conditional quasi-independence model fit, the pair-association outputs, the full-text reading results, the retraction and ISBN exclusions, and the two supplementary tables with their column guide. The code directory holds the harness and the analysis scripts that produce every count, marginal, matrix, test and probability reported in the study. Identifiers, not text. Third-party abstracts and full texts are not redistributed. Abstract-level records carry the DOI, the screening decision and a SHA-256 of the text that was screened, so a reader holding the paper can confirm that the same text was used, without this deposit reproducing publisher prose. The full-text reading records likewise carry verdicts and hashes rather than quoted passages. Reproduction. Every figure and table in the article is generated from these files by scripts in code/. MANIFEST.md lists each file with its size and SHA-256. Version 1.1. Adds the hashed abstract twins for the EC2 recovery set, the manual and PDF-supplied abstracts, the web-extracted abstracts and the retrieval queue, which version 1.0 omitted. No record or decision changed; the addition only extends what a reader can verify. Version 1.2. Corrects the record title to the manuscript's own title. The data and the file set are unchanged from 1.1.