Clustering Social Media Data for Marketing Strategies

Abstract

With the rise of social media platforms for marketing purposes, the central dilemma for researchers and policymakers lies in choosing effective data analysis tools to improve marketing decisions. In the academic literature, numerous articles have discussed clustering techniques for analysing social media data, from a perspective of data mining or social media marketing. However, few studies have attempted to synthesise results obtained from both perspectives. This research aims to (1) offer a structured overview of existing literature on clustering methods for marketing strategies and (2) compare three topic modelling techniques applied to extract the main topics evoked in the corpus of papers. Indeed, topic modelling emerges as a valuable tool for extracting relevant information from big data in general and more specifically from extensive scientific papers. Based on a thematic analysis, the extracted topics were classified according to the following categories: fields, marketing strategies and technologies. Results prove that latent Dirichlet allocation (LDA) is the most effective technique in this context. Furthermore, this study provides an overview of clustering techniques and technologies used for marketing strategies in studied fields. These findings help researchers and practitioners to select the best techniques and technologies for extracting marketing knowledge from big data.