The Systematic Review of Machine Learning Models in Social Media Marketing

Abstract

The paper analyzes the degree of using machine learning (ML) models in marketing. It identifies a research focus in the social media marketing (SMM) sector. The study systematically investigates the literature published between 2020 and 2025 through the Web of Science (WOS) platform. Within this platform, articles that address the concept of marketing in tandem with ML are identified. In this way, 1763 articles are identified, but only 37 explicitly address SMM. The analysis of the 37 articles highlighted 6 main research directions and identified the models Logistic Regression, Random Forest, XGBoost, K-Nearest Neighbors, and Deep Learning as the primary models addressed within the main topic. The study addressed perspectives on researchers' interest in disseminating the results of treating SMM concerning ML. Thus, between the years 2020 and 2025, a few studies were observed, which 2020 targeted six articles and remained at low levels until 2025. Additionally, in the five years of research, only 29 articles were of the original contributions type, and the fields that focused on the collaboration between SMM and ML were predominantly those from business and computer science. Regarding the geographical distribution, the USA was the country with the most research in this field, and in terms of publications, Springer Nature had 9 articles disseminating such research. The co-occurrence analysis of key concepts identified three clusters. Of the three clusters, the central cluster highlighted the correlation between AI technologies and SMM technologies. The small number of articles identified in the specialized literature suggests the need to intensify research in the field of SMM-ML.