Sketch-Fill-A-R: A Persona-Grounded Chit-Chat Generation Framework
This addresses the challenge of creating engaging and consistent conversational agents for applications like chatbots, though it is incremental as it builds on existing persona-based methods.
The paper tackled the problem of generating human-like chit-chat responses by proposing Sketch-Fill-A-R, a persona-grounded framework that outperformed a state-of-the-art baseline with a 10-point lower perplexity and was preferred by 55% in single-turn and 20% higher in consistency in multi-turn user studies.
Human-like chit-chat conversation requires agents to generate responses that are fluent, engaging and consistent. We propose Sketch-Fill-A-R, a framework that uses a persona-memory to generate chit-chat responses in three phases. First, it generates dynamic sketch responses with open slots. Second, it generates candidate responses by filling slots with parts of its stored persona traits. Lastly, it ranks and selects the final response via a language model score. Sketch-Fill-A-R outperforms a state-of-the-art baseline both quantitatively (10-point lower perplexity) and qualitatively (preferred by 55% heads-up in single-turn and 20% higher in consistency in multi-turn user studies) on the Persona-Chat dataset. Finally, we extensively analyze Sketch-Fill-A-R's responses and human feedback, and show it is more consistent and engaging by using more relevant responses and questions.