Deep Bidirectional Language-Knowledge Graph Pretraining
Michihiro Yasunaga, Antoine Bosselut, Hongyu Ren, Xikun Zhang, Christopher D. Manning, Percy Liang, Jure Leskovec
Abstract
Pretraining a language model (LM) on text has been shown to help various downstream NLP tasks. Recent works show that a knowledge graph (KG) can complement text data, offering structured background knowledge that provides a useful scaffold for reasoning. However, these works are not pretrained to learn a deep fusion of the two modalities at scale, limiting the potential to acquire fully joint representations of text and KG. Here we propose DRAGON (Deep Bidirectional Language-Knowledge Graph Pretraining), a self-supervised method to pretrain a deeply joint language-knowledge foundation model from text and KG at scale. Specifically, our model takes pairs of text segments and relevant KG subgraphs as input and bidirectionally fuses information from both modalities. We pretrain this model by unifying two self-supervised reasoning tasks, masked language modeling and KG link prediction. DRAGON outperforms existing LM and LM+KG models on diverse downstream tasks including question answering across general and biomedical domains, with +5% absolute gain on average. In particular, DRAGON achieves strong performance on complex reasoning about language and knowledge (+10% on questions involving long contexts or multi-step reasoning) and low-resource QA (+8% on OBQA and RiddleSense), and new state-of-the-art results on various BioNLP tasks. Our code and trained models are available at https://github.com/michiyasunaga/dragon . Glamping GNN 1st Layer Glamping GNN Final Layer camp trip beach school beach RoBERTa: A. camp (✗) GreaseLM: C. camp (✗) Glamping: B. school (✓) school trip Glamping GNN 1st Layer Glamping GNN Final Layer garage trip beach school beach garage movie Glamping GNN 1st Layer Glamping GNN Final Layer entertain ment station movie Glamping GNN 1st Layer Glamping GNN Final Layer live movie movie entertain ment station live Model Prediction RoBERTa: B. school (✗) GreaseLM: B. school (✗) Glamping: C. beach (✓) Model Prediction B. movie theater (✓) GreaseLM: B. movie theater (✓) Glamping: B. movie theater (✓) Model Prediction B. movie theater (✗) GreaseLM: B. movie theater (✗)
Ask about this paper
Your agent reads all of it.
Lune indexed this paper to the last equation, along with the top-tier papers that cite it. Ask a question and the answer quotes them.
Your agent calls
Luneget_paper_fulltext
Free to start. No credit card required.
Terminal
Install the CLIlune papers fulltext 04062e7c-d186-4d97-85d7-ade74f228a9aCited by top-tier papers44
- HippoRAG: Neurobiologically Inspired Long-Term Memory for Large Language ModelsBernal Jimenez Gutierrez, Yiheng Shu, Yu Gu, Michihiro Yasunaga et al.NeurIPS 2024 · 395 citations
- Knowledge Graph Prompting for Multi-Document Question AnsweringYu Wang, Nedim Lipka, Ryan A. Rossi, Alexa F. Siu et al.AAAI 2024 · 290 citations
- UnifiedSKG: Unifying and Multi-Tasking Structured Knowledge Grounding with Text-to-Text Language ModelsTianbao Xie, Chen Henry Wu, Peng Shi, Ruiqi Zhong et al.EMNLP 2022 · 222 citations
- MediQ: Question-Asking LLMs and a Benchmark for Reliable Interactive Clinical ReasoningShuyue Stella Li, Vidhisha Balachandran, Shangbin Feng, Jonathan Ilgen et al.NeurIPS 2024 · 215 citations
- Talk like a Graph: Encoding Graphs for Large Language ModelsBahare Fatemi, Jonathan Halcrow, Bryan PerozziICLR 2024 · 194 citations
Builds on16
- Language Models are Few-Shot LearnersTom B. Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah et al.NeurIPS 2020 · 64,255 citations
- Chain-of-Thought Prompting Elicits Reasoning in Large Language ModelsJason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma et al.NeurIPS 2022 · 22,562 citations
- Retrieval-Augmented Generation for Knowledge-Intensive NLP TasksPatrick Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni et al.NeurIPS 2020 · 19,162 citations
- PIQA: Reasoning about Physical Commonsense in Natural LanguageYonatan Bisk, Rowan Zellers, Ronan Le Bras, Jianfeng Gao et al.AAAI 2020 · 2,916 citations
- Improving Language Models by Retrieving from Trillions of TokensSebastian Borgeaud, Arthur Mensch, Jordan Hoffmann, Trevor Cai et al.ICML 2022 · 1,629 citations
Related papers
- LinkBERT: Pretraining Language Models with Document LinksMichihiro Yasunaga, Jure Leskovec, Percy LiangACL 2022 · 463 citations
- Exploiting Structured Knowledge in Text via Graph-Guided Representation LearningTao Shen, Yi Mao, Pengcheng He, Guodong Long et al.EMNLP 2020 · 60 citations
- ReasoningLM: Enabling Structural Subgraph Reasoning in Pre-trained Language Models for Question Answering over Knowledge GraphJinhao Jiang, Kun Zhou, Wayne Xin Zhao, Yaliang Li et al.EMNLP 2023 · 26 citations
- GreaseLM: Graph REASoning Enhanced Language ModelsXikun Zhang, Antoine Bosselut, Michihiro Yasunaga, Hongyu Ren et al.ICLR 2022 · 285 citations
- JAKET: Joint Pre-training of Knowledge Graph and Language UnderstandingDonghan Yu, Chenguang Zhu, Yiming Yang, Michael ZengAAAI 2022 · 171 citations
