This presentation gives overview and key lessons from building up a customer service chat automation platform with built-in natural language understanding system from ground zero. Hendrik outlines the challenges involved in interpreting language, which is just a surface representation of the vastly complex conceptual model every human carries around in their brain, and addresses the real world limitations imposed by the chat interface, scarcity of training data and dynamic content of the customer service.
• Human-level language understanding is much more complicated than matching token patterns
• Training data is scarce
• Customisation is the key to good user experience
• Automation pays off when volumes are large