How Pre-trained Language Models Capture Factual Knowledge? A Causal-Inspired Analysis
Shaobo Li, Xiaoguang Li, Lifeng Shang, Zhenhua Dong, Chengjie Sun, Bingquan Liu, Zhenzhou Ji, Xin Jiang
Abstract
Recently, there has been a trend to investigate the factual knowledge captured by Pre-trained Language Models (PLMs). Many works show the PLMs’ ability to fill in the missing factual words in cloze-style prompts such as ”Dante was born in [MASK].” However, it is still a mystery how PLMs generate the results correctly: relying on effective clues or shortcut patterns? We try to answer this question by a causal-inspired analysis that quantitatively measures and evaluates the word-level patterns that PLMs depend on to generate the missing words. We check the words that have three typical associations with the missing words: knowledge-dependent, positionally close, and highly co-occurred. Our analysis shows: (1) PLMs generate the missing factual words more by the positionally close and highly co-occurred words than the knowledge-dependent words; (2) the dependence on the knowledge-dependent words is more effective than the positionally close and highly co-occurred words. Accordingly, we conclude that the PLMs capture the factual knowledge ineffectively because of depending on the inadequate associations.
BibTeX
@inproceedings{li-etal-2022-pre,
title = "How Pre-trained Language Models Capture Factual Knowledge? A Causal-Inspired Analysis",
author = "Li, Shaobo and
Li, Xiaoguang and
Shang, Lifeng and
Dong, Zhenhua and
Sun, Chengjie and
Liu, Bingquan and
Ji, Zhenzhou and
Jiang, Xin and
Liu, Qun",
editor = "Muresan, Smaranda and
Nakov, Preslav and
Villavicencio, Aline",
booktitle = "Findings of the Association for Computational Linguistics: ACL 2022",
month = may,
year = "2022",
address = "Dublin, Ireland",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2022.findings-acl.136/",
doi = "10.18653/v1/2022.findings-acl.136",
pages = "1720--1732"
}