Social Network Optimization Algorithm via Centrality and Influence-aware Learning
SOCIAL is a structure-aware optimization project developed within the Applied Data Science and Artificial Intelligence environment at SRH University Heidelberg. It reframes optimization as networked intelligence: candidate solutions are modeled as agents in a small-world social graph, where centrality, influence diffusion, elite memory, and adaptive mutation guide the search. The approach is designed for complex engineering and scientific problems where gradients may be unavailable, search spaces are multimodal or noisy, and constraints matter.
Why SOCIAL?
- Transforms scattered search into structured, network-based learning.
- Uses centrality-weighted knowledge diffusion instead of only global-best attraction.
- Combines exploration and exploitation through elite memory, synchronization, and adaptive mutation.
- Published in Applied Soft Computing as a new social-network-inspired metaheuristic.
Where can it be used?
- Engineering design optimization: gear train, pressure vessel, welded beam, speed reducer.
- Materials science and cheminformatics: MOF discovery, graph-based materials exploration, design-space search.
- Drug discovery and molecular design, including docking and QSAR-oriented optimization.
- Neural architecture search, logistics, smart grids, and other networked decision environments.
Research context at SRH
At SRH University Heidelberg, our work in Applied Data Science and Artificial Intelligence includes optimization algorithms, graph-based intelligence, scientific AI, and data-driven solutions for engineering and materials science.