Data supplement: Search strings, screening log, and coding tables for a narrative review and bibliometric mapping of AI governance in agriculture (2023-2026)
Description
This dataset supports the article "Governing Algorithms in the Field: A Narrative Review and Bibliometric Mapping of Inclusive Data Governance, Algorithmic Accountability, and Explainability in Agricultural AI Policy" (Muhamad Ridwan, Universitas Hasanuddin, Indonesia). The study combines a bibliometric mapping of 2,620 Scopus-indexed articles and reviews (2023–2026) with a narrative synthesis of 40 articles on inclusive data governance, algorithmic accountability, and explainability in agricultural AI. Searches were run in Scopus on 21 October 2026. Contents: - 00_README.txt: overview of files, settings, and licence. - 01_Scopus_Search_Strings.txt: the three Scopus Advanced Search strings (String 1: 2,641 documents; String 2: 2,620 documents, bibliometric corpus; String 3: 667 articles, narrative synthesis). - 02_Screening_Log_667_records.csv: title and abstract screening of 667 records, with decision, exclusion reason, and relevance score. - 03_Fulltext_Assessment_48_articles.csv: full-text assessment of 48 articles, with eligibility decision (40 included) and the research questions addressed. - 04_Coding_Tables_RQ2-RQ4.xlsx: cross-study comparison tables used in the narrative synthesis. - 05_Selection_Criteria_Coding_Scheme.xlsx: selection-flow counts, inclusion and exclusion criteria, and the coding scheme. - Figure1_bibliometric_workflow.png and Figure2_selection_flow.png: the bibliometric workflow and the article selection flow. The full Scopus export is not redistributed because of the database licence. It can be regenerated with the search strings by users with Scopus access; results may differ slightly because Scopus is updated continuously. Bibliometric maps were produced with VOSviewer 1.6.21 and Tableau Desktop. Elicit was used to assist full-text data extraction; all records were screened by the author.
Files
Steps to reproduce
1. Run the three search strings in 01_Scopus_Search_Strings.txt in Scopus Advanced Search (requires Scopus access). String 2 gives the bibliometric corpus (2,620 documents); String 3 gives the narrative sample (667 articles). Results may differ slightly because Scopus is updated continuously; the original searches were run on 21 October 2026. 2. Export String 2 results as CSV (all fields, including author keywords, affiliations, and citation counts). 3. Keyword co-occurrence and country collaboration maps: open the CSV in VOSviewer 1.6.21; use author keywords with a minimum occurrence of 20 (7,331 keywords, 58 meet the threshold, 56 items in the final map, 4 clusters); use full counting for countries; other settings at VOSviewer defaults. 4. Annual output, citations, source ranking, and country distribution: build in Tableau Desktop from the same CSV. 5. Narrative sample: apply the criteria in 05_Selection_Criteria_Coding_Scheme.xlsx to the 667 articles of String 3 (decisions and reasons in 02_Screening_Log_667_records.csv), assess 48 articles at full text (03_Fulltext_Assessment_48_articles.csv), and code the 40 included articles with the scheme in 05_Selection_Criteria_Coding_Scheme.xlsx. Cross-study comparison tables are in 04_Coding_Tables_RQ2-RQ4.xlsx.
Institutions
- Hasanuddin UniversitySouth Sulawesi, Makassar