# Context Variance Evaluation of Pretrained Language Models for Prompt-based Biomedical Knowledge Probing

arXiv: https://arxiv.org/abs/2211.10265 · HTML: https://mlsanity.com/paper/2211.10265/

## Key facts

- Cited by 15 papers in the map (data through 14 Oct 2026).
- Builds on 22 of its 31 references that are in the map.
- First posted on arXiv 18 Nov 2022.

## Builds on (most cited first, 22 of 22)

- [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://mlsanity.com/paper/1810.04805/) (2018)
- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [RoBERTa: A Robustly Optimized BERT Pretraining Approach](https://mlsanity.com/paper/1907.11692/) (2019)
- [BioBERT: a pre-trained biomedical language representation model for biomedical text mining](https://mlsanity.com/paper/1901.08746/) (2019)
- [Language Models as Knowledge Bases?](https://mlsanity.com/paper/1909.01066/) (2019)
- [Publicly Available Clinical BERT Embeddings](https://mlsanity.com/paper/1904.03323/) (2019)
- [How Can We Know What Language Models Know?](https://mlsanity.com/paper/1911.12543/) (2019)
- [Transfer Learning in Biomedical Natural Language Processing: An Evaluation of BERT and ELMo on Ten Benchmarking Datasets](https://mlsanity.com/paper/1906.05474/) (2019)
- [Factual Probing Is [MASK]: Learning vs. Learning to Recall](https://mlsanity.com/paper/2104.05240/) (2021)
- [AutoPrompt: Eliciting Knowledge from Language Models with Automatically Generated Prompts](https://mlsanity.com/paper/2010.15980/) (2020)
- [How Context Affects Language Models' Factual Predictions](https://mlsanity.com/paper/2005.04611/) (2020)
- [Breaking the Softmax Bottleneck: A High-Rank RNN Language Model](https://mlsanity.com/paper/1711.03953/) (2017)
- [Don't Say That! Making Inconsistent Dialogue Unlikely with Unlikelihood Training](https://mlsanity.com/paper/1911.03860/) (2019)
- [Knowledgeable or Educated Guess? Revisiting Language Models as Knowledge Bases](https://mlsanity.com/paper/2106.09231/) (2021)
- [Can Language Models be Biomedical Knowledge Bases?](https://mlsanity.com/paper/2109.07154/) (2021)
- [Can Prompt Probe Pretrained Language Models? Understanding the Invisible Risks from a Causal View](https://mlsanity.com/paper/2203.12258/) (2022)
- [Training Dynamics for Text Summarization Models](https://mlsanity.com/paper/2110.08370/) (2021)
- [Rewire-then-Probe: A Contrastive Recipe for Probing Biomedical Knowledge of Pre-trained Language Models](https://mlsanity.com/paper/2110.08173/) (2021)
- [MedJEx: A Medical Jargon Extraction Model with Wiki's Hyperlink Span and Contextualized Masked Language Model Score](https://mlsanity.com/paper/2210.05875/) (2022)
- [Extracting Biomedical Factual Knowledge Using Pretrained Language Model and Electronic Health Record Context](https://mlsanity.com/paper/2209.07859/) (2022)
- [Zero-shot Entity Linking with Efficient Long Range Sequence Modeling](https://mlsanity.com/paper/2010.06065/) (2020)
- [Improving Formality Style Transfer with Context-Aware Rule Injection](https://mlsanity.com/paper/2106.00210/) (2021)

## Built on by (most cited first, 15 of 15)

- [Prompt Engineering for Healthcare: Methodologies and Applications](https://mlsanity.com/paper/2304.14670/) (2023)
- [JMLR: Joint Medical LLM and Retrieval Training for Enhancing Reasoning and Professional Question Answering Capability](https://mlsanity.com/paper/2402.17887/) (2024)
- [PaniniQA: Enhancing Patient Education Through Interactive Question Answering](https://mlsanity.com/paper/2308.03253/) (2023)
- [Multi-label Few-shot ICD Coding as Autoregressive Generation with Prompt](https://mlsanity.com/paper/2211.13813/) (2022)
- [README: Bridging Medical Jargon and Lay Understanding for Patient Education through Data-Centric NLP](https://mlsanity.com/paper/2312.15561/) (2023)
- [UMASS_BioNLP at MEDIQA-Chat 2023: Can LLMs generate high-quality synthetic note-oriented doctor-patient conversations?](https://mlsanity.com/paper/2306.16931/) (2023)
- [SYNFAC-EDIT: Synthetic Imitation Edit Feedback for Factual Alignment in Clinical Summarization](https://mlsanity.com/paper/2402.13919/) (2024)
- [Improving Summarization with Human Edits](https://mlsanity.com/paper/2310.05857/) (2023)
- [Automated Identification of Eviction Status from Electronic Health Record Notes](https://mlsanity.com/paper/2212.02762/) (2022)
- [Synthetic Imitation Edit Feedback for Factual Alignment in Clinical Summarization](https://mlsanity.com/paper/2310.20033/) (2023)
- [Comparing Template-based and Template-free Language Model Probing](https://mlsanity.com/paper/2402.00123/) (2024)
- [EHRTutor: Enhancing Patient Understanding of Discharge Instructions](https://mlsanity.com/paper/2310.19212/) (2023)
- [Do Physicians Know How to Prompt? The Need for Automatic Prompt Optimization Help in Clinical Note Generation](https://mlsanity.com/paper/2311.09684/) (2023)
- [Enhancing LLMs for Identifying and Prioritizing Important Medical Jargons from Electronic Health Record Notes Utilizing Data Augmentation: A Comparative Study](https://mlsanity.com/paper/2502.16022/) (2025)
- [Eliciting associations between clinical variables from LLMs via comparison questions across populations](https://mlsanity.com/paper/2605.06335/) (2026)

## Abstract (arXiv)

Pretrained language models (PLMs) have motivated research on what kinds of knowledge these models learn. Fill-in-the-blanks problem (e.g., cloze tests) is a natural approach for gauging such knowledge. BioLAMA generates prompts for biomedical factual knowledge triples and uses the Top-k accuracy metric to evaluate different PLMs' knowledge. However, existing research has shown that such prompt-based knowledge probing methods can only probe a lower bound of knowledge. Many factors like prompt-based probing biases make the LAMA benchmark unreliable and unstable. This problem is more prominent in BioLAMA. The severe long-tailed distribution in vocabulary and large-N-M relation make the performance gap between LAMA and BioLAMA remain notable. To address these, we introduce context variance into the prompt generation and propose a new rank-change-based evaluation metric. Different from the previous known-unknown evaluation criteria, we propose the concept of "Misunderstand" in LAMA for the first time. Through experiments on 12 PLMs, our context variance prompts and Understand-Confuse-Misunderstand (UCM) metric makes BioLAMA more friendly to large-N-M relations and rare relations. We also conducted a set of control experiments to disentangle "understand" from just "read and copy".