Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

KEPLER: A Unified Model for Knowledge Embedding and Pre-trained Language Representation

About

Pre-trained language representation models (PLMs) cannot well capture factual knowledge from text. In contrast, knowledge embedding (KE) methods can effectively represent the relational facts in knowledge graphs (KGs) with informative entity embeddings, but conventional KE models cannot take full advantage of the abundant textual information. In this paper, we propose a unified model for Knowledge Embedding and Pre-trained LanguagE Representation (KEPLER), which can not only better integrate factual knowledge into PLMs but also produce effective text-enhanced KE with the strong PLMs. In KEPLER, we encode textual entity descriptions with a PLM as their embeddings, and then jointly optimize the KE and language modeling objectives. Experimental results show that KEPLER achieves state-of-the-art performances on various NLP tasks, and also works remarkably well as an inductive KE model on KG link prediction. Furthermore, for pre-training and evaluating KEPLER, we construct Wikidata5M, a large-scale KG dataset with aligned entity descriptions, and benchmark state-of-the-art KE methods on it. It shall serve as a new KE benchmark and facilitate the research on large KG, inductive KE, and KG with text. The source code can be obtained from https://github.com/THU-KEG/KEPLER.

Xiaozhi Wang, Tianyu Gao, Zhaocheng Zhu, Zhengyan Zhang, Zhiyuan Liu, Juanzi Li, Jian Tang• 2019

Related benchmarks

TaskDatasetResultRank
Natural Language UnderstandingGLUE (dev)
SST-2 (Acc)94.7
504
Relation ExtractionTACRED (test)
F1 Score72.5
194
Relation ExtractionTACRED
Micro F171.7
97
Link PredictionWikidata5M (test)
MRR0.402
58
Few-shot Relation ClassificationFewRel 1.0 (test)--
36
Knowledge ProbingLAMA
Google-RE P@110
20
Few-shot Relation ClassificationFewRel 2.0
Accuracy (5-way 1-shot)0.6723
18
Relation ClassificationWiki80 (test)
Accuracy93.4
15
Link PredictionWikidata5M (transductive)
MR1.45e+4
14
Knowledge ProbingLAMA-UHN
P@1 (Google-RE)5.9
14
Showing 10 of 20 rows

Other info

Code

Follow for update