Integrating Social Network Analysis and Topic Modeling in Social Media Research: A Systematic Literature Review of Open-Access Scopus Studies (2016-2026)

Abstract Background. Social media research increasingly needs to capture both what people say and how they are connected. Social network analysis (SNA) and topic modeling are the main computational answers, yet how they are combined has not been mapped systematically. Objective. To synthesize how SNA and topic modeling are integrated in social media studies published in 2016–2026 and to derive a research agenda. Methods. We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guideline. Scopus was searched for open-access, English-language research articles combining network, topic-modeling and social media terms. Of 560 records, 232 were screened and 194 included. Studies were coded for platform, domain, topic model, network type, integration pattern and validation by rule-based extraction from titles, abstracts and keywords, audited on 29 articles. Results. Output rose from one article in 2016 to 36 in 2025. X (formerly Twitter) supplied 52% of the data; health (34%) and politics (25%) dominated. LDA (latent Dirichlet allocation) appeared in 32% of articles, BERTopic in 9% and STM (structural topic model) in 6%; half named no algorithm in the abstract. Actor networks (59%) outnumbered semantic (27%) and hybrid (11%) networks. Parallel use of the two methods prevailed (71%); joint models were 9% and topic-built networks 5%. Evaluation was rarely visible in abstracts (2%). Conclusions. Integration is widespread but shallow. We propose a four-route taxonomy (parallel, network→topic, topic→network, joint model), a five-stage research model and an eight-point agenda centred on joint and temporal models, validation standards, under-studied platforms and regions, and transparent reporting.