Negated and Misprimed Probes for Pretrained Language Models: Birds Can Talk, But Cannot Fly
Nora Kassner, Hinrich Schütze Center for Information and Language Processing (CIS) LMU Munich, Germany
Abstract
Building on Petroni et al. (2019), we propose two new probing tasks analyzing factual knowledge stored in Pretrained Language Models (PLMs). (1) Negation. We find that PLMs do not distinguish between negated (“Birds cannot [MASK]”) and non-negated (“Birds can [MASK]”) cloze questions. (2) Mispriming. Inspired by priming methods in human psychology, we add “misprimes” to cloze questions (“Talk? Birds can [MASK]”). We find that PLMs are easily distracted by misprimes. These results suggest that PLMs still have a long way to go to adequately learn human-like factual knowledge.
中文速览
预训练语言模型（Pretrained Language Models，PLMs）虽然在问答任务上表现抢眼，但它们对事实知识的掌握其实相当脆弱。研究者设计了两类探测实验：一是把陈述性问题加上否定词（如"鸟不能[MASK]"），二是在问题前插入干扰词（misprime，如"Talk？鸟能[MASK]"），发现BERT等主流模型对否定句和原句的预测结果几乎一样，完全无视语义上的否定，同时仅凭一个不相关的干扰词就足以让模型给出错误答案。进一步用合成语料实验表明，在无监督预训练阶段模型根本学不会区分"真"与"假"，但加入有监督微调后就能正确处理否定，说明问题出在预训练机制本身对否定语义的忽视。这项研究揭示了现有语言模型的知识存储方式更像表面的共现匹配而非真正的语义理解，为评估和改进模型的事实推理能力提供了新的基准视角。
原文 arXiv:1911.03343;中英对照 + 大白话阅读 https://aha.fim.ai/paper/1911.03343v3