Dynamic Twitter friend grouping based on similarity, interaction, and trust to account for ever-evolving relationships

Abstract

Online social networks have become ubiquitous, allowing users to share opinions on various topics. However, oversharing can compromise privacy, leading to potential blackmail or fraud. Current platforms lack friend categorization based on trust levels. This study proposes simulating real-world friendships by grouping users into three categories: acquaintances, friends, and close friends, based on trust and engagement. It also introduces a dynamic method to adjust relationship status over time, considering users' past and present offenses against peers. The proposed system automatically updates friend lists, eliminating manual grouping. It calculates relationship strength by considering all components of online social networks and trust variations caused by user attacks. This method can be integrated with clustering algorithms on popular platforms like Facebook, Twitter, and Instagram to enable constrained sharing. By implementing this system, users can better control their information sharing based on trust levels, reducing privacy risks. The dynamic nature of the relationship status adjustment ensures that the system remains relevant as user interactions evolve over time. This approach offers a more nuanced and secure social networking experience, reflecting real-world relationship dynamics in the digital sphere.