همايش ملی پيشرفت های معماری سازمانی

صفحه اصلی / نهمین همایش پیشرفت‌های معماری سازمانی ایران

Agile and Fair Enterprise Architecture Analytics with Federated Graph Contrastive Learning and Nash Bargaining

نویسندگان :

Saghar Shafaati¹ Mona Rahmani² S. Hossein Erfani³

1- Department of Computer Engineering, South Tehran Branch, Islamic Azad University, Tehran, Iran 2- Department of Corporate Entrepreneurship, Faculty of Entrepreneurship, University of Tehran, Tehran, Iran 3- Department of Computer Engineering, South Tehran Branch, Islamic Azad University, Tehran, Iran

کلمات کلیدی :

Enterprise Architecture،Federated Learning،Graph Neural Networks،Contrastive Learning،Nash Bargaining،Digital Twin

چکیده :

Modern enterprises face challenges from decentralized information, dispersed decision-making, and rapidly evolving technologies, which hinder effective Enterprise Architecture (EA) governance. To address this, we propose a novel framework that integrates federated graph learning, contrastive representation learning, and game-theoretic optimization. EA is modeled as a decentralized graph of entities, where Graph Attention Networks (GAT) capture structural dependencies, contrastive learning yields robust cross-silo embeddings, and Nash Bargaining ensures fairness and stability in federated aggregation. Experiments on multi-firm financial data show consistent gains over baseline federated GNNs, with lower prediction error and stronger resilience to noisy, non-IID data. To complement the accuracy, the system enhances EA practice by making decentralization, equity, and flexibility work—isolated data converges to create shared intelligence, bargaining grants equal weight, and contrastive embeddings support agility. The system is then a digital twin of EA, enabling collaborative, privacy-driven analytics that improve governance, accountability, and strategic decision-making.