charlesXu86/Chatbot_CN

Combines deep learning, reinforcement learning, and dialogue management within a modular architecture that supports multi-turn task-oriented conversations across financial and legal domains. The system decomposes into 13+ specialized submodules (NLU, dialogue policy, retrieval-based QA, knowledge graphs, entity recognition) with RASA as the core dialogue engine, supplemented by utility components for query correction and coreference resolution. Deployable across multiple channels (web, DingTalk, WeChat) via Django integration, with optional voice interaction and call center capabilities.

1,293 stars. No commits in the last 6 months.