Related Experiment Videos
Customization in a unified framework for summarizing medical literature
N Elhadad1, M-Y Kan, J L Klavans
1Department of Computer Science, Columbia University, New York, NY 10027, USA. noemie@cs.columbia.edu
Artificial Intelligence in Medicine
|April 7, 2005
Summary
This study demonstrates that customized medical document summaries are achievable within a digital library. The system tailors information for physicians and laypersons, enhancing accessibility and relevance.
Area of Science:
- Medical Informatics
- Natural Language Processing
- Information Retrieval
Background:
- The PErsonalized Retrieval and Summarization of Images, Video and Language (PERSIVAL) medical digital library aims to improve information access.
- Existing systems often provide generic summaries, failing to meet diverse user needs.
Purpose of the Study:
- To present a summarization system for the PERSIVAL medical digital library.
- To focus on strategies for defining and generating customized medical document summaries.
Main Methods:
- A unified user model is employed to create tailored summaries for physicians or laypersons.
- The system leverages regularities in medical literature structure and content.
- Summaries are generated using a combination of machine-generated and extracted text from multiple documents.
Main Results:
- Customized summaries were successfully generated for distinct user groups (physicians, laypersons).
- Both group-based and individually-driven (patient record-based) models were implemented.
- The system effectively combined machine-generated and extracted text for comprehensive summaries.
Conclusions:
- Customization of information retrieval and summarization is feasible within a medical digital library.
- This approach enhances the utility of medical digital libraries by providing personalized content.