Building Knowledge-Grounded Dialogue Systems with Graph-Based Semantic Modeling

TL;DR

This work tackles knowledge-grounded dialogue by introducing Grounded Graph ($G^2$), a graph-structured representation of both dialogue and knowledge, to enable robust knowledge selection and integration. It couples $G^2$ with a Grounded Graph Aware Transformer ($G^2AT$) that fuses sequential and graphical knowledge via dual encoders and a graph-sequence fusion decoder. Empirical results on Wizard of Wikipedia and CMU_DoG show >$10\%$ improvements in response generation and ~ $20\%$ gains in factual consistency over state-of-the-art baselines, with strong generalization and robustness. The proposed approach demonstrates that incorporating explicit semantic structures as priors in neural models can significantly enhance knowledge-grounded language generation and reliability.

Abstract

The knowledge-grounded dialogue task aims to generate responses that convey information from given knowledge documents. However, it is a challenge for the current sequence-based model to acquire knowledge from complex documents and integrate it to perform correct responses without the aid of an explicit semantic structure. To address these issues, we propose a novel graph structure, Grounded Graph ($G^2$), that models the semantic structure of both dialogue and knowledge to facilitate knowledge selection and integration for knowledge-grounded dialogue generation. We also propose a Grounded Graph Aware Transformer ($G^2AT$) model that fuses multi-forms knowledge (both sequential and graphic) to enhance knowledge-grounded response generation. Our experiments results show that our proposed model outperforms the previous state-of-the-art methods with more than 10\% gains in response generation and nearly 20\% improvement in factual consistency. Further, our model reveals good generalization ability and robustness. By incorporating semantic structures as prior knowledge in deep neural networks, our model provides an effective way to aid language generation.

Figures (7)

• Figure 1: An example of knowledge-grounded dialogue with responses from models. Text in orange denotes the the information from knowledge and node in purple and green denotes the node from dialogue and knowledge respectively.
• Figure 2: An example illustrates the distance of tokens in sequence and graph, which will affect the modeling of long-distance relationships.
• Figure 3: An example of Grounded Graph construction procedure. To simplify the graph, we only choose the last utterance and the grounded knowledge sentence to construct the graph. In the actual processing, we consider all knowledge (about 60 sentences) and long-distance context (about three utterances) and filters out sub-graphs that do not contain any nodes from the dialogue context by graph augmentation.
• Figure 4: Illustration of our $G^2AT$ architecture.
• Figure 5: Comparison of different graph structures.