Neural Generation Meets Real People: Building a Social, Informative Open-Domain Dialogue Agent
This work addresses the problem of building engaging and socially fluent dialogue agents for users in open-domain settings, though it appears incremental as it builds upon existing chatbot frameworks.
The authors tackled the challenge of creating an open-domain social chatbot that is both informative and conversational, resulting in Chirpy Cardinal, which placed second out of nine bots in the Alexa Prize Socialbot Grand Challenge with an average user rating of 3.58/5.
We present Chirpy Cardinal, an open-domain social chatbot. Aiming to be both informative and conversational, our bot chats with users in an authentic, emotionally intelligent way. By integrating controlled neural generation with scaffolded, hand-written dialogue, we let both the user and bot take turns driving the conversation, producing an engaging and socially fluent experience. Deployed in the fourth iteration of the Alexa Prize Socialbot Grand Challenge, Chirpy Cardinal handled thousands of conversations per day, placing second out of nine bots with an average user rating of 3.58/5.