Related Experiment Videos
Towards a medical question-answering system: a feasibility study.
Pierre Jacquemart1, Pierre Zweigenbaum
1STIM/DSI, Assistance Publique-Hôpitaux de Paris, France. pja@biomath.jussieu.fr
Studies in Health Technology and Informatics
|December 11, 2003
Summary
This study explores French healthcare question-answering (QA) systems. While a semantic model covers 90% of oral surgery questions, limited French online resources may restrict answer availability.
Area of Science:
- Natural Language Processing (NLP)
- Information Retrieval (IR)
- Knowledge Engineering
- Question-Answering (QA) Systems
Background:
- Question-answering systems represent the next evolution of search engines, integrating Information Retrieval (IR) and Natural Language Processing (NLP).
- Evaluating the feasibility of QA systems for the French healthcare domain is crucial for advancing medical information access.
Purpose of the Study:
- To assess the viability of developing a French-language QA system for the healthcare sector.
- To examine keyword selection for IR and question compatibility with existing QA prototypes.
Main Methods:
- Collected a corpus of student questions in oral surgery.
- Performed manual web searches to develop automated principles for IR query construction.
- Designed a semantic model using UMLS Semantic Network relations.
Main Results:
- Identified automatable principles for building effective IR queries from natural language questions.
- Developed a semantic model consistent with a QA prototype, covering 90% of the studied questions.
- Noted that domain specialization and limited French online resources may impact answer retrieval.
Conclusions:
- A QA system for French healthcare is feasible, with a robust semantic model for question understanding.
- Challenges remain due to the specialized nature of medical queries and the scarcity of French online health information, potentially limiting answer quantity.