On Inductive Abilities of Latent Factor Models for Relational Learning

Trouillon, Théo; Gaussier, Éric; Dance, Christopher R.; Bouchard, Guillaume

doi:10.1613/jair.1.11305

“…In the literature besides adversarial attacks, Lawrence et al (2020), Nandwani et al (2020 and Zhang et al (2019b) generate post-hoc explanations to understand KGE model predictions. Trouillon et al (2019) study the inductive abilities of KGE models as binary relation properties for controlled inference tasks with synthetic datasets. Allen et al (2021) interpret the structure of knowl-edge graph embeddings by comparison with word embeddings.…”

Section: Related Workmentioning

confidence: 99%

Poisoning Knowledge Graph Embeddings via Relation Inference Patterns

Bhardwaj¹,

Kelleher²,

Costabello³

et al. 2021

Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Confer

View full text Add to dashboard Cite

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.

show abstract

“…To give an idea of the types of reasoning necessary for models to perform well on CODEX, we analyze the presence of learnable binary relation patterns within CODEX. The three main types of such patterns in knowledge graphs are symmetry, inversion, and compositionality (Trouillon et al, 2019;. We address symmetry and compositionality here, and omit inversion because we specifically removed inverse relations to avoid train/test leakage ( § 3.2).…”

Section: Analysis Of Relation Patternsmentioning

confidence: 99%

CoDEx: A Comprehensive Knowledge Graph Completion Benchmark

Safavi¹,

Koutra²

2020

Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP)

View full text Add to dashboard Cite

We present CODEX, a set of knowledge graph COmpletion Datasets EXtracted from Wikidata and Wikipedia that improve upon existing knowledge graph completion benchmarks in scope and level of difficulty. In terms of scope, CODEX comprises three knowledge graphs varying in size and structure, multilingual descriptions of entities and relations, and tens of thousands of hard negative triples that are plausible but verified to be false. To characterize CODEX, we contribute thorough empirical analyses and benchmarking experiments. First, we analyze each CODEX dataset in terms of logical relation patterns. Next, we report baseline link prediction and triple classification results on CODEX for five extensively tuned embedding models. Finally, we differentiate CODEX from the popular FB15K-237 knowledge graph completion dataset by showing that CODEX covers more diverse and interpretable content, and is a more difficult link prediction benchmark. Data, code, and pretrained models are available at https://bit.ly/2EPbrJs.

show abstract

“…To give an idea of the types of reasoning necessary for models to perform well on CODEX, we analyze the presence of learnable binary relation patterns within CODEX. The three main types of such patterns in knowledge graphs are symmetry, inversion, and compositionality (Trouillon et al, 2019;Sun et al, 2019). We address symmetry and compositionality here, and omit inversion because we specifically removed inverse relations to avoid train/test leakage ( § 3.2).…”

Section: Analysis Of Relation Patternsmentioning

confidence: 99%

CoDEx: A Comprehensive Knowledge Graph Completion Benchmark

Safavi¹,

Koutra²

2020

Preprint

0

View full text Add to dashboard Cite

We present CODEX, a set of knowledge graph COmpletion Datasets EXtracted from Wikidata and Wikipedia that improve upon existing knowledge graph completion benchmarks in scope and level of difficulty. In terms of scope, CODEX comprises three knowledge graphs varying in size and structure, multilingual descriptions of entities and relations, and tens of thousands of hard negative triples that are plausible but verified to be false. To characterize CODEX, we contribute thorough empirical analyses and benchmarking experiments. First, we analyze each CODEX dataset in terms of logical relation patterns. Next, we report baseline link prediction and triple classification results on CODEX for five extensively tuned embedding models. Finally, we differentiate CODEX from the popular FB15K-237 knowledge graph completion dataset by showing that CODEX covers more diverse and interpretable content, and is a more difficult link prediction benchmark. Data, code, and pretrained models are available at https://bit.ly/2EPbrJs.

show abstract

On Inductive Abilities of Latent Factor Models for Relational Learning

Cited by 24 publications

References 44 publications

Poisoning Knowledge Graph Embeddings via Relation Inference Patterns

Poisoning Knowledge Graph Embeddings via Relation Inference Patterns

CoDEx: A Comprehensive Knowledge Graph Completion Benchmark

CoDEx: A Comprehensive Knowledge Graph Completion Benchmark

Contact Info

Product

Resources

About