Enhancing Dialogue Generation via Dynamic Graph Knowledge Aggregation

TL;DR

This paper introduces a dynamic graph knowledge aggregation framework that integrates external graph knowledge into dialogue generation models, effectively bridging the semantic gap between text and graph representations to produce more informative responses.

Contribution

It proposes a novel hierarchical graph attention method with pseudo nodes for better feature aggregation, improving dialogue generation performance over existing models.

Findings

Outperforms state-of-the-art baselines on dialogue generation tasks.

Fills the semantic gap between text and graph knowledge representations.

Enables the language model to select more relevant knowledge triples for responses.

Abstract

Incorporating external graph knowledge into neural chatbot models has been proven effective for enhancing dialogue generation. However, in conventional graph neural networks (GNNs), message passing on a graph is independent from text, resulting in the graph representation hidden space differing from that of the text. This training regime of existing models therefore leads to a semantic gap between graph knowledge and text. In this study, we propose a novel framework for knowledge graph enhanced dialogue generation. We dynamically construct a multi-hop knowledge graph with pseudo nodes to involve the language model in feature aggregation within the graph at all steps. To avoid the semantic biases caused by learning on vanilla subgraphs, the proposed framework applies hierarchical graph attention to aggregate graph features on pseudo nodes and then attains a global feature. Therefore, the framework can better utilise the heterogeneous features from both the post and external graph knowledge. Extensive experiments demonstrate that our framework outperforms state-of-the-art (SOTA) baselines on dialogue generation. Further analysis also shows that our representation learning framework can fill the semantic gap by coagulating representations of both text and graph knowledge. Moreover, the language model also learns how to better select knowledge triples for a more informative response via exploiting subgraph patterns within our feature aggregation process. Our code and resources are available at https://github.com/tangg555/SaBART.