CAVG Balancing Privacy and Utility: A Novel PMKMA–AWOA Method for Social Network Anonymisation

ABSTRACT The acute issue of privacy protection in social networks is addressed in this study. While large volumes of personal data can improve services, they also run the risk of disclosing private information. This paper presents a novel hybrid strategy aimed at balanced clustering and improved anonymisation in social architectures: the partitioned mean k ‐means algorithm (PMKMA) in conjunction with the adaptive whale optimisation algorithm (AWOA). To provide an initial solution that complies with predetermined anonymisation constraints and goals, PMKMA first divides users into C clusters, each of which has a minimum of K users. AWOA is used to further optimise cluster configurations after clustering, guaranteeing better anonymisation for the network structure and data. The suggested PMKMA–AWOA approach is assessed using metrics, including CPU runtime, cost function, objective function, distortion rate, balancing error, and clustering error in comparison to other methods, including K ‐means clustering with coati optimisation algorithm (KMCOA), K ‐member k ‐means clustering and coati optimisation algorithm (2KMCOA), and K ‐member clustering with honeybee optimisation algorithm (K‐HOA). This strategy seeks to improve privacy protection in social networks while preserving data accuracy and usability by skilfully striking a balance between data utility and confidentiality.