Automated Interactive Domain-Specific Conversational Agents that Understand Human Dialogs

Explore the AutoConcierge system, an innovative approach to creating domain-specific conversational agents that truly understand human dialogs, presented in this 51-minute conference talk. Dive into the mechanics of how AutoConcierge leverages GPT-3 to convert human utterances into predicates, employs s(CASP) answer set programming for commonsense reasoning, and generates natural language responses. Learn how this framework combines large language models with answer set programming to create realistic, task-specific chatbots capable of engaging in human-like conversations, specifically demonstrated in the domain of restaurant recommendations. Gain insights into the potential applications of this technology for developing other domain-specific conversational agents.

[PADL'24] Automated Interactive Domain-Specific Conversational Agents that Understand Huma...