International Journal of Social Network Mining (3 papers in press)

Regular Issues

• Community detection using imperialist competitive algorithm  
  by Zahra Mahmoodabadi, Abdorreza Savadi 
  Abstract: The clustering of complex networks has become one of the most fascinating fields in recent years. The clustering of complex networks, including social networks, provides useful information. This information leads to optimal structures in complex networks and is very useful for solving real-world problems. Social network graphs have an important feature that random graphs do not have: community structure. The problem of community detection in social network graphs corresponds to graph partitioning and is an NP-hard problem. Therefore, the application of evolutionary algorithms can be an appropriate solution to this problem. This paper uses the imperialist competitive algorithm (ICA) to identify communities within a social network graph. In addition, a parallel version of the ICA was applied to the problem. The results show that the parallel version has better convergence rather than the serial ICA. To compare the performance of ICA with other evolutionary algorithms, we use the PSO algorithm for the community detection problem. The results confirm that parallel ICA (PICA) is superior in both cost value and convergence rate.
  Keywords: parallel ICA; PICA; community detection; social networks; evolutionary algorithms.
  DOI: 10.1504/IJSNM.2023.10059293
   
  
• Comparative analysis of SIR, SEIR, SIRS, and SEIRS models for information diffusion in social networks using Instagram data  
  by Manoj Kumar Srivastav, Somsubhra Gupta, Subhranil Som 
  Abstract: Social networks enable users to share information and interact rapidly. Therefore, understanding information diffusion patterns is essential for analysing engagement behaviour. In this study, four epidemiological models R, SEIR, SIRS, and SEIRS are compared for information diffusion on Instagram. For this purpose, a publicly available dataset comprising 200 Instagram influencers is employed. In the proposed framework, user engagement is represented through epidemiological compartments, and model parameters are estimated from normalised Instagram features. A 30-day discrete-time simulation is conducted to compare the four models. Furthermore, the models are evaluated using peak engagement, peak day, average engagement, parameter sensitivity analysis, mean absolute error (MAE), root mean square error (RMSE), and the Friedman statistical test. The results indicate that the four models produce distinct diffusion patterns, with the susceptible-infected-recovered (SIR) and susceptible-infected-recovered susceptible (SIRS) models achieving higher engagement levels than the susceptible-exposed-infected-recovered (SEIR) and susceptible-exposed infected-recovered-susceptible (SEIRS) models. Overall, this study compares epidemiological models for information diffusion using Instagram engagement data.
  Keywords: social networks; susceptible-infected-recovered; SIR; SEIR; SIRS; SEIRS; engagement.
  DOI: 10.1504/IJSNM.2026.10081132
   
  
• Transformations in social media content production lines following the adoption of AI-generated text and images: workflow reconfiguration, role repositioning, and media differentiation  
  by Chia-Sung Yen 
  Abstract: This study situates itself in platform-based social media contexts and examines how the adoption of AI-generated texts and images reshapes content production workflows and professional roles, as well as how these changes relate to content credibility. Using a qualitative research design, in-depth interviews were conducted with six media and social media content practitioners, and their experiences and judgements regarding copywriting, design, multimedia processing, and publication gatekeeping were synthesised. The findings indicate that AI tools can enhance production efficiency and support rapid iteration, shifting the workflow from conventional production to a generate-select-revise mode. At the same time, AI-generated content amplifies the importance of verification and quality control, making editing and review more critical gatekeeping checkpoints. At the role level, design and copywriting work tends to move away from repetitive output toward more strategic judgement and the maintenance of professional standards. Divergent orientations were also observed across media backgrounds: platform-side practitioners placed greater emphasis on creativity and formal innovation, whereas interviewees with traditional media backgrounds emphasised credibility and professionalism. Overall, the diffusion of AI-generated content may drive increasing differentiation in content positioning between social media and traditional media, while raising new challenges for future content governance.
  Keywords: social media; generative artificial intelligence; content production workflow; professional roles; content credibility.
  DOI: 10.1504/IJSNM.2026.10081258