CHAT-EDA: An Event-Driven Architecture proposal for intelligent chatbots

Abstract

This paper presents CHAT-EDA, an Event-Driven Architecture (EDA) for developing intelligent chatbots in a distributed systems fashion. This proposal addresses challenges related to module coupling while ensuring scalability, flexibility, data consistency, and persistence, along with maintainability. The architecture comprises three main modules: frontend, backend, and an artificial intelligence (AI) module, interconnected by an event bus. Each module operates independently and asynchronously, promoting modularity, decoupling, and data integrity. The model was implemented in Techtalk, a chatbot project developed to optimize the registration and analysis of technical service calls in a network of technical support centers. The system was evaluated through load and resilience tests, validating quality attributes such as performance, consistency, and observability. The results demonstrated the architecture’s capability to handle high volumes of simultaneous messages while maintaining data integrity and efficiency in asynchronous processing. The CHAT-EDA evaluation results indicate the suitability to meet demands of complex corporate systems, highlighting its robustness for future expansions. Furthermore, the architectural decisions reinforces the acthitecture feasibility when applied to large-scale applications, ensuring efficient integration with AI technologies and meeting business requirements.