Technical Report of Team GraphMIRAcles in the WikiKG90M-LSC Track of OGB-LSC @ KDD Cup 2021

TL;DR

This paper presents a comprehensive link prediction framework for large-scale knowledge graphs, combining a basic model, rule mining, and inference, achieving high accuracy on the WikiKG90M dataset.

Contribution

The paper introduces an integrated framework for large-scale knowledge graph link prediction, combining ComplEx-CMRC, rule mining, and inference, with ablation studies on knowledge distillation.

Findings

Achieved an MRR of 0.9707 on test data.

Model without knowledge distillation achieved 0.9533 MRR.

Demonstrated effectiveness of combined components in large-scale link prediction.

Abstract

Link prediction in large-scale knowledge graphs has gained increasing attention recently. The OGB-LSC team presented OGB Large-Scale Challenge (OGB-LSC), a collection of three real-world datasets for advancing the state-of-the-art in large-scale graph machine learning. In this paper, we introduce the solution of our team GraphMIRAcles in the WikiKG90M-LSC track of OGB-LSC @ KDD Cup 2021. In the WikiKG90M-LSC track, the goal is to automatically predict missing links in WikiKG90M, a large scale knowledge graph extracted from Wikidata. To address this challenge, we propose a framework that integrates three components -- a basic model ComplEx-CMRC, a rule miner AMIE 3, and an inference model to predict missing links. Experiments demonstrate that our solution achieves an MRR of 0.9707 on the test dataset. Moreover, as the knowledge distillation in the inference model uses test tail candidates -- which are unavailable in practice -- we conduct ablation studies on knowledge distillation. Experiments demonstrate that our model without knowledge distillation achieves an MRR of 0.9533 on the full validation dataset.