Commonsense-augmented Memory Construction and Management in Long-term Conversations via Context-aware Persona Refinement
This work addresses a specific issue in conversational AI for multi-session settings, offering an incremental improvement over prior methods that focused on avoiding contradictions.
The paper tackles the problem of uninformative persona sentences in long-term conversations by introducing a framework that uses commonsense-based persona expansion to refine contradictory personas into rich speaker information, resulting in improved response generation.
Memorizing and utilizing speakers' personas is a common practice for response generation in long-term conversations. Yet, human-authored datasets often provide uninformative persona sentences that hinder response quality. This paper presents a novel framework that leverages commonsense-based persona expansion to address such issues in long-term conversation. While prior work focuses on not producing personas that contradict others, we focus on transforming contradictory personas into sentences that contain rich speaker information, by refining them based on their contextual backgrounds with designed strategies. As the pioneer of persona expansion in multi-session settings, our framework facilitates better response generation via human-like persona refinement. The supplementary video of our work is available at https://caffeine-15bbf.web.app/.