Enhancing Personalized Dialogue Generation with Contrastive Latent Variables: Combining Sparse and Dense Persona
This work addresses the challenge of noisy and uninformative persona resources in dialogue systems, offering an incremental improvement for AI assistants and chatbots.
The paper tackled the problem of modeling persona for personalized dialogue generation by combining sparse attributes, dense texts, and dialogue history to create a richer persona representation, resulting in a model that demonstrated superiority in personalization on Chinese and English datasets.
The personalized dialogue explores the consistent relationship between dialogue generation and personality. Existing personalized dialogue agents model persona profiles from three resources: sparse or dense persona descriptions and dialogue histories. However, sparse structured persona attributes are explicit but uninformative, dense persona texts contain rich persona descriptions with much noise, and dialogue history query is both noisy and uninformative for persona modeling. In this work, we combine the advantages of the three resources to obtain a richer and more accurate persona. We design a Contrastive Latent Variable-based model (CLV) that clusters the dense persona descriptions into sparse categories, which are combined with the history query to generate personalized responses. Experimental results on Chinese and English datasets demonstrate our model's superiority in personalization.