Stylized Knowledge-Grounded Dialogue Generation via Disentangled Template Rewriting

TL;DR

This paper introduces a novel disentangled template rewriting method for stylized knowledge-grounded dialogue generation, enabling style control without supervision while maintaining factual accuracy and coherence.

Contribution

It proposes a new end-to-end differentiable framework that combines style and content templates for stylized dialogue generation without requiring paired training data.

Findings

Significant improvement over previous stylized dialogue methods.

Achieves comparable performance to standard knowledge-grounded dialogue models.

Effective style control without supervision.

Abstract

Current Knowledge-Grounded Dialogue Generation (KDG) models specialize in producing rational and factual responses. However, to establish long-term relationships with users, the KDG model needs the capability to generate responses in a desired style or attribute. Thus, we study a new problem: Stylized Knowledge-Grounded Dialogue Generation (SKDG). It presents two challenges: (1) How to train a SKDG model where no <context, knowledge, stylized response> triples are available. (2) How to cohere with context and preserve the knowledge when generating a stylized response. In this paper, we propose a novel disentangled template rewriting (DTR) method which generates responses via combing disentangled style templates (from monolingual stylized corpus) and content templates (from KDG corpus). The entire framework is end-to-end differentiable and learned without supervision. Extensive experiments on two benchmarks indicate that DTR achieves a significant improvement on all evaluation metrics compared with previous state-of-the-art stylized dialogue generation methods. Besides, DTR achieves comparable performance with the state-of-the-art KDG methods in standard KDG evaluation setting.