Relation Prediction as an Auxiliary Training Objective for Improving Multi-Relational Graph Representations
This work addresses knowledge base completion for AI systems by improving entity ranking, though it is incremental as it builds on existing objectives.
The paper tackles the problem of learning representations on multi-relational graphs for knowledge base completion by proposing a self-supervised training objective that incorporates relation prediction, resulting in significant improvements such as a 6.1% increase in MRR and 9.9% increase in Hits@1 on FB15k-237.
Learning good representations on multi-relational graphs is essential to knowledge base completion (KBC). In this paper, we propose a new self-supervised training objective for multi-relational graph representation learning, via simply incorporating relation prediction into the commonly used 1vsAll objective. The new training objective contains not only terms for predicting the subject and object of a given triple, but also a term for predicting the relation type. We analyse how this new objective impacts multi-relational learning in KBC: experiments on a variety of datasets and models show that relation prediction can significantly improve entity ranking, the most widely used evaluation task for KBC, yielding a 6.1% increase in MRR and 9.9% increase in Hits@1 on FB15k-237 as well as a 3.1% increase in MRR and 3.4% in Hits@1 on Aristo-v4. Moreover, we observe that the proposed objective is especially effective on highly multi-relational datasets, i.e. datasets with a large number of predicates, and generates better representations when larger embedding sizes are used.