Traditional Graph-Based vs. AI-Driven Approaches in Social Network Analysis: A Comparative Study

Abstract

Traditionally, graph-based data mining used social network analysis algorithms such as measurement and influence propagation. These models adopted an analytical and mathematical model that was well suited in analyzing static and semi static social networks (Tabassum et al., 2018). But with the adoption of neural networks and deep learning techniques (Qiu et al., 2018), newer approaches have been designed for analyzing complex and dynamic networks.

In this paper, a review of both the traditional as well as the artificial intelligence-based approaches is made with respect to their strengths, features, and complementary aspects. Various concepts such as community discovery, link prediction, influence evaluation, and anomaly detection have been discussed.

In brief, the use of graph-based data mining methods in social network analysis was based on conventional models that concentrated on measurement and influence detection. These approaches were designed using a mathematical and statistical approach that could be used to analyze both static and semi-static networks. However, with the current fast development in artificial intelligence and machine learning technology, new methods that are better suited for the analysis of complex data structures and dynamic networks have emerged. In this research, an analysis is conducted on traditional and artificial intelligence data mining methods and the advantages and dimensions of the two approaches will be highlighted.