Text mining and topic modeling of the global laying hen welfare literature: An LDA-based synthesis

The global laying hen welfare research landscape has expanded rapidly in recent decades, yet large-scale, data-driven quantitative synthesis of this field remains limited to descriptive bibliometric analyses. This study aimed to systematically map the global spatiotemporal patterns, core research topics, and critical knowledge gaps in laying hen welfare research using standardized text mining and topic modeling approaches. A systematic search was conducted in the Scopus database for peer-reviewed literature on laying hen welfare published from 1990 to 2024. A final corpus of 1,102 eligible abstracts was curated following the PRISMA 2020 standard, with core analyses implemented in R software, including text mining, Latent Dirichlet Allocation (LDA) topic modeling, and geospatial-temporal trend analysis. Descriptive analyses revealed a consistent and substantial growth in publication output, with 78% of all documents published between 2013 and 2024. Geographically, Europe (45.0%), North America (22.0%), and Asia (20.0%) dominated global research output, while Africa accounted for only 1.1% of publications; BRIC (Brazil, Russia, India, China) nations contributed 54% of global egg production but just 23% of scholarly output. LDA modeling identified 5 distinct, semantically coherent research topics, with housing systems, stocking density, and feather pecking confirmed as the most intensively investigated welfare factors. The five topics were selected based on the C_V coherence peak at k = 5 (Section 2.4). This study provides the first comprehensive text mining-based map of global laying hen welfare research, revealing a stark mismatch between research investment and egg production scale. The findings offer an evidence-based foundation for global research prioritization, international collaboration, and evidence-based policy development for laying hen welfare.