Realistic Re-evaluation of Knowledge Graph Completion Methods: An Experimental Study
This work highlights a critical flaw in evaluating embedding models for knowledge graph completion, calling for re-investigation of approaches due to excessive data leakage and unrealistic scenarios.
The study systematically re-evaluates knowledge graph completion methods by removing unrealistic triples from benchmark datasets, revealing that embedding models are much less accurate than previously reported, with poor accuracy rendering link prediction ineffective.
In the active research area of employing embedding models for knowledge graph completion, particularly for the task of link prediction, most prior studies used two benchmark datasets FB15k and WN18 in evaluating such models. Most triples in these and other datasets in such studies belong to reverse and duplicate relations which exhibit high data redundancy due to semantic duplication, correlation or data incompleteness. This is a case of excessive data leakage---a model is trained using features that otherwise would not be available when the model needs to be applied for real prediction. There are also Cartesian product relations for which every triple formed by the Cartesian product of applicable subjects and objects is a true fact. Link prediction on the aforementioned relations is easy and can be achieved with even better accuracy using straightforward rules instead of sophisticated embedding models. A more fundamental defect of these models is that the link prediction scenario, given such data, is non-existent in the real-world. This paper is the first systematic study with the main objective of assessing the true effectiveness of embedding models when the unrealistic triples are removed. Our experiment results show these models are much less accurate than what we used to perceive. Their poor accuracy renders link prediction a task without truly effective automated solution. Hence, we call for re-investigation of possible effective approaches.