AI-driven Influencer and Market Analysis: A Social Network Approach to Measure E-Commerce Relationships

Authors:
Open Access
Publisher website
Google Scholar
Abstract
Social network analysis is a process of studying social structures and relationships using graph theory and data analysis techniques. It involves mapping and measuring connections and entities in a network. However, on online selling platforms, identifying influential entities such as individuals and high-value products remains a challenge due to the complexity of customer and seller interactions. This study aims to assess seller performance and product lifetime value using AI-driven network analysis involving a measure of centrality. AI-driven network analysis utilises artificial intelligence (AI) to identify influential individuals and predict emerging trends in consumer engagement. It uses weighted degree and betweenness centrality to assess their effectiveness in identifying influential entities, including sellers, products, or organisations in a commercial network. Weighted degree centrality measures the strength and frequency of direct connections, while betweenness centrality identifies entities that act as intermediaries across different network segments. The analysis reveals that weighted degree centrality, with a value of 3190 for annual seller performance, is more closely aligned with actual sales performance and stakeholder assessments, making it a more suitable metric for supporting business decisions in this context. The findings demonstrate that AI-driven analytics enable businesses to consistently identify high-performing sellers and products based on their structural positions within the network. It contributes to the development of more targeted marketing strategies, systematic recognition of top performers, and enhanced customer engagement through data-informed decision-making. Future research may explore the integration of dynamic network modelling with multi-layered e-commerce networks, thereby increasing the depth of analysis across various platforms and industries.

This publication has 20 references indexed in Scilit: