Language Models as Knowledge Embeddings

25 Jun 2022  ·  Xintao Wang, Qianyu He, Jiaqing Liang, Yanghua Xiao ·

Knowledge embeddings (KE) represent a knowledge graph (KG) by embedding entities and relations into continuous vector spaces. Existing methods are mainly structure-based or description-based. Structure-based methods learn representations that preserve the inherent structure of KGs. They cannot well represent abundant long-tail entities in real-world KGs with limited structural information. Description-based methods leverage textual information and language models. Prior approaches in this direction barely outperform structure-based ones, and suffer from problems like expensive negative sampling and restrictive description demand. In this paper, we propose LMKE, which adopts Language Models to derive Knowledge Embeddings, aiming at both enriching representations of long-tail entities and solving problems of prior description-based methods. We formulate description-based KE learning with a contrastive learning framework to improve efficiency in training and evaluation. Experimental results show that LMKE achieves state-of-the-art performance on KE benchmarks of link prediction and triple classification, especially for long-tail entities.

PDF Abstract

Results from the Paper

Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Link Prediction FB15k-237 C-LMKE(BERT-tiny) MRR 0.410 # 1
Hits@10 0.571 # 2
Hits@3 0.445 # 1
Hits@1 0.319 # 1
MR 132 # 6
Link Prediction WN18RR C-LMKE(bert-base) MRR 0.598 # 4
Hits@10 0.806 # 3
Hits@3 0.675 # 3
Hits@1 0.480 # 4
MR 72 # 5