RECAP:面向个性化对话回复生成的检索增强型上下文感知前缀编码器
RECAP: Retrieval-Enhanced Context-Aware Prefix Encoder for Personalized Dialogue Response Generation
AI总结:
针对聊天机器人一致人设构建难题,提出名为RECAP的检索增强个性化对话回复生成方法,通过分层Transformer检索器和上下文感知前缀编码器实现信息融合,在英文Reddit对话数据集上表现优于现有最优基线。
摘要:
赋予聊天机器人一致的人设是打造有吸引力对话的关键,但这仍是一个尚未解决的难题。本研究提出了一种用于个性化回复生成的新型检索增强方法。具体而言,我们设计了一个在对话领域数据上训练的分层Transformer检索器以执行个性化检索,还设计了一个上下文感知前缀编码器,可将检索到的信息更有效地融合到解码器中。在真实世界数据集上开展的大量实验证明,我们的模型在生成更流畅、更具个性化的回复方面具备有效性。我们通过一套人工指标和自动指标对模型性能进行了定量评估,发现在英文Reddit对话任务上,该模型的表现优于现有最优基线模型。
英文摘要:
Endowing chatbots with a consistent persona is essential to an engaging conversation, yet it remains an unresolved challenge. In this work, we propose a new retrieval-enhanced approach for personalized response generation. Specifically, we design a hierarchical transformer retriever trained on dialogue domain data to perform personalized retrieval and a context-aware prefix encoder that fuses the retrieved information to the decoder more effectively. Extensive experiments on a real-world dataset demonstrate the effectiveness of our model at generating more fluent and personalized responses. We quantitatively evaluate our model's performance under a suite of human and automatic metrics and find it to be superior compared to state-of-the-art baselines on English Reddit conversations.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。