Building your first chatbot in Python - Rachael Tatman | PyData Jeddah

Build your first chatbot in Python using PyData Jeddah, covering task-oriented dialogue systems and more.

Key takeaways

Task Oriented Dialogue Systems: Focus on building agents that can have conversations with humans; use stories to train the agents.
Rasa: Framework for building conversational agents with natural language understanding (NLU) capabilities.
Intent Classification: System identifies the intent behind the user’s request (e.g., greet, purchase, etc.).
Dialogue Policy: Determines the next action to take based on the conversation history and user intent.
Transformer Models: Neural models that can handle long-range relationships and are suitable for sequence-to-sequence modeling.
Hybrid Approach: Use a combination of rule-based and machine learning based approaches to build a conversational agent.
Training Data: Provide extensive training data to improve the accuracy and robustness of the conversational agent.
Evaluation: Use testing data and evaluation metrics to assess the performance of the conversational agent.
Customization: Design a custom domain file for intent classification and use slots to carry state information across the conversation.
Action file: Define actions that the conversational agent can take in response to user input.

More talks