Knowledge Base Completion: Baseline strikes back (Again)

TL;DR

This paper demonstrates that training Knowledge Base Completion models with all available negative samples significantly improves performance, challenging prior assumptions about negative sampling efficiency and unifying various multiplicative methods.

Contribution

It introduces COMPLEX-V2, a training approach using all negative samples, and shows how this enhances existing models and questions the value of recent multiplicative methods.

Findings

Using all negative samples yields near state-of-the-art results.

Multiplicative KBC methods perform similarly when trained with all negatives.

Reassessment of the value of recent KBC methods is necessary.

Abstract

Knowledge Base Completion (KBC) has been a very active area lately. Several recent KBCpapers propose architectural changes, new training methods, or even new formulations. KBC systems are usually evaluated on standard benchmark datasets: FB15k, FB15k-237, WN18, WN18RR, and Yago3-10. Most existing methods train with a small number of negative samples for each positive instance in these datasets to save computational costs. This paper discusses how recent developments allow us to use all available negative samples for training. We show that Complex, when trained using all available negative samples, gives near state-of-the-art performance on all the datasets. We call this approach COMPLEX-V2. We also highlight how various multiplicative KBC methods, recently proposed in the literature, benefit from this train-ing regime and become indistinguishable in terms of performance on most datasets. Our work calls for a reassessment of their individual value, in light of these findings.