Exploring Influence Maximization: State-of-the-Art Methods, Taxonomies, and Trends

Influence maximization (IM) is a key problem in social network analysis, with numerous applications in areas such as viral marketing, advertising, public health, and more. Current surveys in IM primarily focus on traditional algorithms while overlooking leading advanced IM algorithms such as signed IM, fairness-enabled IM, and diversity-based IM. This survey paper seeks to address the existing gap by providing a thorough review of both classical and advanced IM algorithms. It discusses classical techniques like simulation-based, path-based, and so on, and mainly focuses on recent advanced IM algorithms. In this detailed discussion of advanced algorithms, the work analyzes various aspects, including complexity, network types, properties (monotonicity, sub-modularity), diffusion models, and comparative algorithms, along with their advantages and disadvantages. This work includes an analysis of classical and advanced algorithms, as well as experimental evaluations of 20 state-of-the-art algorithms across 10 datasets. The chosen state-of-the-art algorithms utilize various approaches like simulation-based, evolutionary-based, fairness-based, and path-based methods. This work also addresses the challenges associated with IM and highlights emerging trends and future directions. Overall, this survey serves as a comprehensive guide to the latest developments in IM while proposing new avenues for future research in this field.