CEGRL-TKGR: A Causal Enhanced Graph Representation Learning Framework for Temporal Knowledge Graph Reasoning

TL;DR

This paper introduces CEGRL-TKGR, a novel framework that incorporates causal structures into temporal knowledge graph reasoning to improve prediction accuracy by mitigating biases and spurious correlations.

Contribution

It proposes a causal-enhanced graph learning framework that disentangles causal and confounding factors, utilizing causal intervention theory for more robust temporal knowledge graph reasoning.

Findings

Outperforms existing models on six benchmark datasets.

Effectively reduces bias and spurious correlations in predictions.

Achieves superior link prediction accuracy.

Abstract

Temporal knowledge graph reasoning (TKGR) is increasingly gaining attention for its ability to extrapolate new events from historical data, thereby enriching the inherently incomplete temporal knowledge graphs. Existing graph-based representation learning frameworks have made significant strides in developing evolving representations for both entities and relational embeddings. Despite these achievements, there's a notable tendency in these models to inadvertently learn biased data representations and mine spurious correlations, consequently failing to discern the causal relationships between events. This often leads to incorrect predictions based on these false correlations. To address this, we propose an innovative Causal Enhanced Graph Representation Learning framework for TKGR (named CEGRL-TKGR). This framework introduces causal structures in graph-based representation learning to unveil the essential causal relationships between events, ultimately enhancing the performance of the TKGR task. Specifically, we first disentangle the evolutionary representations of entities and relations in a temporal knowledge graph sequence into two distinct components, namely causal representations and confounding representations. Then, drawing on causal intervention theory, we advocate the utilization of causal representations for predictions, aiming to mitigate the effects of erroneous correlations caused by confounding features, thus achieving more robust and accurate predictions. Finally, extensive experimental results on six benchmark datasets demonstrate the superior performance of our model in the link prediction task.