Mapping the Intellectual Structure of Generative AI Research in Retail and E-commerce: A Bibliometric Review

Generative artificial intelligence (GenAI) has rapidly transitioned from a frontier laboratory technology into a mainstream commercial layer of retail and e-commerce, reshaping personalization, recommendation, content generation, and conversational commerce. Despite an explosion of empirical and conceptual work since late 2022, the intellectual architecture of this body of knowledge remains fragmented. This study aims to map the intellectual structure, thematic clusters, and emerging research frontiers of GenAI scholarship in retail and e-commerce, with particular attention to agentic AI and retrieval-augmented generation (RAG) as emerging concepts and to country-level impact differentials. A PRISMA-guided bibliometric review was conducted on 445 Scopus-indexed articles published between 2023 and April 2026 (5,460 cumulative citations, 75 contributing countries). Seven complementary VOSviewer analyses were performed: keyword co-occurrence with curated thesaurus harmonization (26 mappings), co-citation of sources, co-citation of first authors, bibliographic coupling of documents, bibliographic coupling of journals, co-authorship of authors using Scopus identifiers, and co-authorship of countries. Full counting and association-strength normalization were applied throughout. The corpus exhibits exponential growth (12→90→218 articles in 2023–2025) and is structured around five thematic clusters: GenAI core and customer outcomes, AI chatbots and adoption, NLP/ML methods, e-commerce and customer experience, and large language models with recommender systems. Two parallel intellectual traditions anchor the field: a technical-AI tradition (Vaswani, Devlin, Brown, Radford, Touvron) and an adoption-behavioral tradition (Davis, Venkatesh, Ajzen, Hair, Dwivedi, Mogaji). Emerging concepts of agentic AI, retrieval-augmented generation, and conversational AI (average publication years 2025.3–2025.7) constitute the research frontier. Country impact is highly heterogeneous: China leads in volume (112 papers, 7.48 citations per paper) but the United Kingdom, India, the United Arab Emirates, Romania, Taiwan, and France produce small-but-impactful portfolios with citation per paper ratios ranging from 29 to 89. This is the first comprehensive bibliometric mapping of GenAI scholarship at the retail/e-commerce intersection, the first to document the dual technical-behavioral intellectual structure of the field, and the first to position agentic AI and RAG as emerging research frontiers using empirical co-occurrence overlay temporal evidence.