Summary of Infusing Clinical Knowledge Into Tokenisers For Language Models, by Abul Hasan et al.
Infusing clinical knowledge into tokenisers for language models
by Abul Hasan, Jinge Wu, Quang Ngoc Nguyen, Salomé Andres, Imane Guellil, Huayu Zhang, Arlene Casey, Beatrice Alex, Bruce Guthrie, Honghan Wu
First submitted to arxiv on: 20 Jun 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The novel knowledge enhanced tokenisation mechanism, K-Tokeniser, is introduced for clinical text processing. This technique populates global representations of tokens based on semantic types from domain concepts like Unified Medical Language System or task-related corpus training data. The optimal global token representation is chosen using sentence-level localised context at training or inference stage. To avoid pretraining, an embedding initialisation approach generates new token representations. Three transformer-based language models are used to evaluate K-Tokeniser on four real-world datasets for clinical text analytics tasks like concept and relation extraction, automated coding, phenotype identification, and research article classification. The models demonstrate consistent improvements over counterparts in all tasks, with a 13% increase in Micro F1 score in the automated clinical coding task. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This study creates a new way to process medical texts using knowledge enhanced tokenisation. It helps language models work better by giving them more information about what words mean. The researchers tested this method on four different datasets and it worked well for tasks like finding specific concepts, classifying articles, and automatically coding patient records. |
Keywords
» Artificial intelligence » Classification » Embedding » F1 score » Inference » Pretraining » Token » Transformer
