System and methods for semiautomatic generation and tuning of natural language interaction applications

This invention describes a system that helps build and fine-tune applications that understand natural language, like chatbots. It works by taking example conversations, automatically organizing them, and then generating the underlying rules for the application to respond. The system then tests these rules and uses real user interactions to continuously improve the application. The claims specify that this system relies on preconfigured, hierarchical models for understanding language and domains.