Unlearning of Knowledge Graph Embedding via Preference Optimization

TL;DR

This paper introduces GraphDPO, a novel approximate unlearning framework for knowledge graph embeddings that effectively removes outdated or erroneous information while preserving remaining knowledge, outperforming existing methods.

Contribution

The paper proposes GraphDPO, a preference optimization-based unlearning method that improves removal of targeted triples and preserves boundary knowledge in KGs.

Findings

GraphDPO outperforms baselines by up to 10.1% in MRR_Avg.

GraphDPO achieves up to 14.0% improvement in MRR_F1.

The method effectively removes targeted knowledge while maintaining overall KG integrity.

Abstract

Existing knowledge graphs (KGs) inevitably contain outdated or erroneous knowledge that needs to be removed from knowledge graph embedding (KGE) models. To address this challenge, knowledge unlearning can be applied to eliminate specific information while preserving the integrity of the remaining knowledge in KGs. Existing unlearning methods can generally be categorized into exact unlearning and approximate unlearning. However, exact unlearning requires high training costs while approximate unlearning faces two issues when applied to KGs due to the inherent connectivity of triples: (1) It fails to fully remove targeted information, as forgetting triples can still be inferred from remaining ones. (2) It focuses on local data for specific removal, which weakens the remaining knowledge in the forgetting boundary. To address these issues, we propose GraphDPO, a novel approximate unlearning framework based on direct preference optimization (DPO). Firstly, to effectively remove forgetting triples, we reframe unlearning as a preference optimization problem, where the model is trained by DPO to prefer reconstructed alternatives over the original forgetting triples. This formulation penalizes reliance on forgettable knowledge, mitigating incomplete forgetting caused by KG connectivity. Moreover, we introduce an out-boundary sampling strategy to construct preference pairs with minimal semantic overlap, weakening the connection between forgetting and retained knowledge. Secondly, to preserve boundary knowledge, we introduce a boundary recall mechanism that replays and distills relevant information both within and across time steps. We construct eight unlearning datasets across four popular KGs with varying unlearning rates. Experiments show that GraphDPO outperforms state-of-the-art baselines by up to 10.1% in MRR_Avg and 14.0% in MRR_F1.