This academic article introduces PK-NCLI, a novel method aimed at improving the quality of responses generated by conversational agents leveraging AI and deep learning. The method achieves this by accurately and efficiently identifying relevant auxiliary information like persona and background knowledge. The article highlights the importance of respecting external knowledge and personalizing to user preferences in conversational AI. The experiment results indicate that PK-NCLI outperforms the state-of-the-art method, PK-FoCus, in terms of perplexity, knowledge grounding, and training efficiency.
Publication date: 4 Dec 2023
Project Page: Not provided
Paper: https://arxiv.org/pdf/2312.00774
