Literature dataset for "Large Language Models for defense and security: A systematic review" (Maaiveld et al., 2026)

This dataset contains the raw literature-search data underlying the systematic review "Large Language Models for defense and security: A systematic review" (Maaiveld et al., to appear). It covers the identification stage of the PRISMA process. Records were retrieved on 9 March 2026 using Publish or Perish (v8.19) across four databases: Scopus, Google Scholar, Crossref, and Semantic Scholar. Each (query, source) pair was exported as both a search report (.rtf) and a BibTeX file (.bibtex), organized into one folder per database. The eleven query/source pairs correspond to the record counts reported in the paper's methodology section. The dataset also includes: duplicate_entries.bibtex: the 309 records identified and removed as duplicates during cross-database deduplication. overlap-analysis.log: the within- and between-database overlap analysis. The retrieval yielded 1,986 records, reduced to 1,677 unique records after deduplication. Screening against the inclusion and exclusion criteria was performed manually by the review team; no automated filtering was applied at this stage. Full details of the query construction, retrieval constraints, and selection process are given in the paper.