Machine unlearning techniques aim to mitigate unintended memorization in large language models (LLMs). However, existing approaches predominantly focus on the explicit removal of isolated facts, often overlooking latent inferential dependencies and the non-deterministic nature of knowledge within LLMs. Consequently, facts presumed forgotten may persist implicitly through correlated information. To address these challenges, we propose a knowledge unlearning evaluation framework that more accurately captures the implicit structure of real-world knowledge by representing relevant factual contexts as knowledge graphs with associated confidence scores. We further develop an inference-based evaluation protocol leveraging powerful LLMs as judges; these judges reason over the extracted knowledge subgraph to determine unlearning success. Our LLM judges utilize carefully designed prompts and are calibrated against human evaluations to ensure their trustworthiness and stability. Extensive experiments on our newly constructed benchmark demonstrate that our framework provides a more realistic and rigorous assessment of unlearning performance. Moreover, our findings reveal that current evaluation strategies tend to overestimate unlearning effectiveness. Our code is publicly available at https://github.com/Graph-COM/Knowledge_Unlearning.git.

Adversarial attacks on knowledge graph embeddings (KGE) aim to disrupt the model's ability of link prediction by removing or inserting triples. A recent black-box method has attempted to incorporate textual and structural information to enhance attack performance. However, it is unable to generate human-readable explanations, and exhibits poor generalizability. In the past few years, large language models (LLMs) have demonstrated powerful capabilities in text comprehension, generation, and reasoning. In this paper, we propose LLMAtKGE, a novel LLM-based framework that selects attack targets and generates human-readable explanations. To provide the LLM with sufficient factual context under limited input constraints, we design a structured prompting scheme that explicitly formulates the attack as multiple-choice questions while incorporating KG factual evidence. To address the context-window limitation and hesitation issues, we introduce semantics-based and centrality-based filters, which compress the candidate set while preserving high recall of attack-relevant information. Furthermore, to efficiently integrate both semantic and structural information into the filter, we precompute high-order adjacency and fine-tune the LLM with a triple classification task to enhance filtering performance. Experiments on two widely used knowledge graph datasets demonstrate that our attack outperforms the strongest black-box baselines and provides explanations via reasoning, and showing competitive performance compared with white-box methods. Comprehensive ablation and case studies further validate its capability to generate explanations.

Despite being the go-to choice for link prediction on knowledge graphs, research on interpretability of knowledge graph embeddings (KGE) has been relatively unexplored. We present KGEx, a novel post-hoc method that explains individual link predictions by drawing inspiration from surrogate models research. Given a target triple to predict, KGEx trains surrogate KGE models that we use to identify important training triples. To gauge the impact of a training triple, we sample random portions of the target triple neighborhood and we train multiple surrogate KGE models on each of them. To ensure faithfulness, each surrogate is trained by distilling knowledge from the original KGE model. We then assess how well surrogates predict the target triple being explained, the intuition being that those leading to faithful predictions have been trained on "impactful" neighborhood samples. Under this assumption, we then harvest triples that appear frequently across impactful neighborhoods. We conduct extensive experiments on two publicly available datasets, to demonstrate that KGEx is capable of providing explanations faithful to the black-box model.

In knowledge graph completion (KGC), predicting triples involving emerging entities and/or relations, which are unseen when the KG embeddings are learned, has become a critical challenge. Subgraph reasoning with message passing is a promising and popular solution. Some recent methods have achieved good performance, but they (i) usually can only predict triples involving unseen entities alone, failing to address more realistic fully inductive situations with both unseen entities and unseen relations, and (ii) often conduct message passing over the entities with the relation patterns not fully utilized. In this study, we propose a new method named RMPI which uses a novel Relational Message Passing network for fully Inductive KGC. It passes messages directly between relations to make full use of the relation patterns for subgraph reasoning with new techniques on graph transformation, graph pruning, relation-aware neighborhood attention, addressing empty subgraphs, etc., and can utilize the relation semantics defined in the ontological schema of KG. Extensive evaluation on multiple benchmarks has shown the effectiveness of techniques involved in RMPI and its better performance compared with the existing methods that support fully inductive KGC. RMPI is also comparable to the state-of-the-art partially inductive KGC methods with very promising results achieved. Our codes and data are available at https://github.com/zjukg/RMPI.

We study the problem of explaining link predictions in the Knowledge Graph Embedding (KGE) models. We propose an example-based approach that exploits the latent space representation of nodes and edges in a knowledge graph to explain predictions. We evaluated the importance of identified triples by observing progressing degradation of model performance upon influential triples removal. Our experiments demonstrate that this approach to generate explanations outperforms baselines on KGE models for two publicly available datasets.

We study the problem of generating data poisoning attacks against Knowledge Graph Embedding (KGE) models for the task of link prediction in knowledge graphs. To poison KGE models, we propose to exploit their inductive abilities which are captured through the relationship patterns like symmetry, inversion and composition in the knowledge graph. Specifically, to degrade the model's prediction confidence on target facts, we propose to improve the model's prediction confidence on a set of decoy facts. Thus, we craft adversarial additions that can improve the model's prediction confidence on decoy facts through different inference patterns. Our experiments demonstrate that the proposed poisoning attacks outperform state-of-art baselines on four KGE models for two publicly available datasets. We also find that the symmetry pattern based attacks generalize across all model-dataset combinations which indicates the sensitivity of KGE models to this pattern.

Despite the widespread use of Knowledge Graph Embeddings (KGE), little is known about the security vulnerabilities that might disrupt their intended behaviour. We study data poisoning attacks against KGE models for link prediction. These attacks craft adversarial additions or deletions at training time to cause model failure at test time. To select adversarial deletions, we propose to use the model-agnostic instance attribution methods from Interpretable Machine Learning, which identify the training instances that are most influential to a neural model's predictions on test instances. We use these influential triples as adversarial deletions. We further propose a heuristic method to replace one of the two entities in each influential triple to generate adversarial additions. Our experiments show that the proposed strategies outperform the state-of-art data poisoning attacks on KGE models and improve the MRR degradation due to the attacks by up to 62% over the baselines.

Representing entities and relations in an embedding space is a well-studied approach for machine learning on relational data. Existing approaches, however, primarily focus on improving accuracy and overlook other aspects such as robustness and interpretability. In this paper, we propose adversarial modifications for link prediction models: identifying the fact to add into or remove from the knowledge graph that changes the prediction for a target fact after the model is retrained. Using these single modifications of the graph, we identify the most influential fact for a predicted link and evaluate the sensitivity of the model to the addition of fake facts. We introduce an efficient approach to estimate the effect of such modifications by approximating the change in the embeddings when the knowledge graph changes. To avoid the combinatorial search over all possible facts, we train a network to decode embeddings to their corresponding graph components, allowing the use of gradient-based optimization to identify the adversarial modification. We use these techniques to evaluate the robustness of link prediction models (by measuring sensitivity to additional facts), study interpretability through the facts most responsible for predictions (by identifying the most influential neighbors), and detect incorrect facts in the knowledge base.