Related Experiment Videos
SAPHIRE International: a tool for cross-language information retrieval
1Division of Medical Informatics and Outcomes Research, School of Medicine, Oregon Health Sciences University, Portland, USA.
Proceedings. AMIA Symposium
|February 3, 1999
Summary
The SAPHIRE system now supports cross-language information retrieval in six languages, enhancing access to medical literature. Further development is needed for German language variants and synonym expansion in the UMLS Metathesaurus.
Area of Science:
- Medical Informatics
- Natural Language Processing
- Information Retrieval
Background:
- Medical literature is predominantly in English, creating barriers for non-native speakers.
- Cross-language information retrieval (CLIR) is crucial for global access to medical knowledge.
- Existing systems require enhancement to support multilingual querying and retrieval.
Purpose of the Study:
- To enhance the SAPHIRE concept-retrieval system for multilingual capabilities.
- To enable SAPHIRE to process text input and output concepts in multiple languages.
- To evaluate the system's performance in a real-world application like CliniWeb.
Main Methods:
- Enhanced the SAPHIRE system to map free-text documents and queries to concepts in the UMLS Metathesaurus.
- Implemented support for six languages: English, German, French, Russian, Spanish, and Portuguese.
- Utilized the CliniWeb catalogue as a test case for SAPHIRE International.
Main Results:
- SAPHIRE International successfully accepts text input and provides Metathesaurus concept output in six languages.
- Demonstrated the system's application in the CliniWeb catalogue for accessing clinical information.
- Formative evaluation of German terms identified areas for improvement, including handling pluralization and synonym expansion.
Conclusions:
- The enhanced SAPHIRE system significantly improves cross-language information retrieval for medical literature.
- Further linguistic processing, particularly for German, is necessary to optimize performance.
- Expanding the UMLS Metathesaurus with more synonyms will enhance retrieval accuracy.