Tracing trends and topics in two decades of social commerce research: a bibliometric and topic modeling analysis

Abstract

As a new genre of electronic commerce, social commerce has rapidly evolved in both academia and practice, necessitating continuous knowledge updates to comprehensively recognize its research trends and topics. While existing reviews have identified research themes, gaps remain to better utilize unstructured data such as abstracts, reduce the labor intensity of manual reviews, and enhance semantic understanding through advanced techniques. To address these gaps, this study combines updated bibliometric analysis with BERTopic, a BERT-based topic modeling technique with superior contextual understanding, to identify research topics. Following SPAR-4-SLR protocol, 1,279 articles from 2005 to March 2025 were collected from Web of Science. Bibliometric analysis first examines publication trends and influential entities via Biblioshiny and visualizes knowledge networks via VOSviewer, followed by BERTopic topic modeling of abstracts that reveals 14 topics. The findings highlight the sustained importance of deepening core topics, research opportunities from business-oriented topics, and the necessity of integrating research perspectives. Based on these topics, an integrative conceptual framework is proposed, organizing topics into four antecedent dimensions (technology, social, commerce, people) and one outcome dimension (behavior and attitude). This study contributes by demonstrating methodological novelty in applying advanced BERTopic technique to extract latent topics with superior semantic precision, providing researchers with comprehensive insights into major trends, the research framework and future research directions, and offering actionable implications for platforms, enterprises, and governments.