What is it about?
Natural language processing (NLP) of clinical reports is an important area of development for accelerating our understanding of disease. We present arguments for a cognitive-inspired framework for medical NLP which Includes the need for symbol grounding, semantic activation, semantic composition, and hierarchical predictive coding.
Featured Image
Photo by Accuray on Unsplash
Why is it important?
Currently, there are no accepted strategies for community development of medical NLP systems. Furthermore, it is important to appreciate the requirement differences of medical applications compared to the general world-wide web applications. We present a design that can incorporate various language understanding perspectives in order to fully address difficult language ambiguities and performance needs in medical applications.
Perspectives
The general computing community is currently driven by deep learning numerical methods. There has been a resistance to symbolic methods, especially those that require extensive domain knowledge that must be manually specified. However, there is a theoretical limit as to the extent to which large language models can serve as a foundation for medical applications. We believe that a larger cognitive framework is a better strategic direction and which large language models can be used as one of many knowledge sources.
Ricky Taira
University of California Los Angeles
Read the Original
This page is a summary of: Design considerations for a hierarchical semantic compositional framework for medical natural language understanding, PLoS ONE, March 2023, PLOS,
DOI: 10.1371/journal.pone.0282882.
You can read the full text:
Contributors
The following have contributed to this page
