Leveraging Wikipedia knowledge to classify multilingual biomedical documents

Area of Science:

• Biomedical Informatics
• Natural Language Processing
• Knowledge Representation

Background:

• Classifying biomedical documents across multiple languages is challenging.
• Existing methods often rely on machine translation or complex linguistic tools.
• Leveraging structured knowledge bases like Wikipedia can offer an alternative approach.

Purpose of the Study:

• To develop and evaluate a classifier for multilingual biomedical documents using Wikipedia knowledge.
• To introduce and assess the effectiveness of a cross-language concept matching technique.
• To compare the proposed method against machine translation and MetaMap-based classifiers.

Main Methods:

• Document representation using concept weights derived from Wikipedia.
• Cross-language concept matching via Wikipedia interlanguage links.
• Experimental validation on two custom multilingual biomedical corpora (ML-UVigoMED and EFSG-UVigoMED).

Main Results:

• The proposed Wikipedia-based classifier significantly outperforms machine translation and MetaMap-based approaches.
• The cross-language concept matching technique effectively enables multilingual classification.
• Superior performance was observed across benchmark experiments.

Conclusions:

• Wikipedia knowledge is highly advantageous for multilingual biomedical document classification.
• The proposed method offers a robust and effective solution for cross-lingual information retrieval in the biomedical domain.
• This approach provides a scalable and efficient alternative to traditional multilingual classification techniques.