Comprehensive framework for implementing blockchain-enabled federated learning and full homomorphic encryption for chatbot security system

Comprehensive framework for implementing blockchain-enabled federated learning and full homomorphic encryption for chatbot security system

연구 분야: Cryptography

학회: Cluster Computing

Chatbot is an artificial intelligence application that can provide a conversational environment between humans and machines. Most organizations and industries are willing to lay out their services through chatbots because they can provide 24/7 customer support. Meanwhile, it raises security and privacy challenges like access control, data leakage during transmission, SQL injection attacks, and language model attacks, which make the users concerned about their data, performance, and accuracy. Therefore, this research paper proposes a comprehensive framework integrating blockchain, federated learning, and a fully homomorphic encryption algorithm with face recognition to solve the above-mentioned chatbot’s challenges. The experimental result shows that a distributed system improves chatbot accuracy (90%) and that more transactions in less time with more clients do not affect the performance. In contrast, more iterations and clients will decrease the accuracy, performance, and transactions in a centralized system. In addition, fully homomorphic encryption improves and speeds up the data encryption process. It encrypted more data (1792 MB) in a small amount of 1240 times per second, and conversations and transactions can be transferred via a secure network to ensure the confidentiality, integrity, and authenticity of users’ data. The implementation of such a comprehensive framework in real-life situations can improve chatbot security when it actively works as a customer agent in an organization.