Influence maximization in social networks: Theories, methods and challenges

Abstract

Influence maximization (IM) is the process of choosing a set of seeds from a social network so that the most individuals will be influenced by them. Calculating the social effect of a given seed set and identifying the smallest seed set that maximizes the social influence present two theoretical challenges in the IM problem, so the majority of current research projects focus on finding solutions to these two problems. This paper presents a comparative study of advanced approaches to IM algorithms, with particular attention to the following key aspects: (1) To establish the foundation of the IM problem, three types of diffusion models are first reviewed and their features are compared. Based on the classical diffusion models, research works on context-aware diffusion models and deep learning-based simulation of diffusion processes are summarized; (2) a complete analysis and comparison of classic IM algorithms and context-aware IM algorithms are presented using the algorithmic framework of IM algorithms; and (3) key challenges and future research directions in this field are explored. For researchers who are new to the IM field, our study offers the most recent advancements in the sector, a deeper grasp of IM issues, and a solid foundation from which to pursue further work in the area.

Keywords

• Social networks
• Information diffusion
• Social influence
• Influence maximization