Conversational AI with Rasa Core & NLU
summary
Live-coding builds a working ML chatbot with Rasa NLU & Rasa Core, using spaCy medium vectors for intent features, CRF for entities and an LSTM policy for dialogue; the demo trains on about six intents and basic stories, runs a model persisted after trained 400 epochs, adds a fallback at ~0.6 confidence, shows interactive correction that exports new stories, and wires custom actions that set slots and call APIs so the bot actually executes external calls and adapts from real conversations.
AI-written summary · about
chapters (35)
- 0:00Introduction to conversational AI
- 2:54Why choose Rasa
- 7:11Speaker background and audience engagement
- 8:37Chatbot roadmap with Rasa
- 9:21Setup and environment
- 10:02Natural language understanding fundamentals
- 11:28Intent classification
- 14:19Entity extraction
- 17:12Simple conversational example
- 18:20Coding with Jupyter
- 22:50Training data for NLU
- 26:13Rasa NLU pipeline
- 28:50Training NLU models
- 30:38Adding entity recognition
- 34:42Testing NLU model
- 37:34Handling out-of-domain
- 40:00Model performance evaluation
- 45:06Language models and vectors
- 46:34Dialogue handling intro
- 49:56ML for dialogue management
- 50:57State machines vs RL
- 54:11Rasa Core’s ML approach
- 58:46Training data for dialogue
- 1:02:22Training dialogue stories
- 1:08:09Interacting with chatbot
- 1:10:38Fallback policies
- 1:16:48Custom actions in Rasa
- 1:22:44Dialogue model evaluation
- 1:24:59Training from real convos
- 1:26:31Interactive learning demo
- 1:30:33Current Rasa research
- 1:33:24Combining machine learning with rule-based systems
- 1:35:16Continuous improvement
- 1:36:43Key resources
- 1:37:24Final thoughts
related talks
30mConversational AI: Building clever chatbots
01h 27mDeprecating the state machine: building conversational AI with the Rasa stack
31mRasa: open source conversational AI to build next generation chatbots
01h 31mGoing beyond “Sorry, I didn’t get that” building AI assistants that scale...
01h 29mGoing Beyond “Sorry, I Didn’t Get That”: Building AI Assistants that Scale...
34mWriting a Chatbot in PHP
33mBuilding the next-generation Conversational AI with Python and Deep Learning
38m3 things about Chatbot (you don't have to learn it the hard way)