Terminological resources for text mining over biomedical scientific literature

Area of Science:

• Biomedical Informatics
• Computational Biology
• Natural Language Processing

Background:

• Biomedical literature contains complex terminology with significant ambiguity.
• Accurate identification and disambiguation of biological entities are critical for reliable text mining.
• Existing terminological resources may lack comprehensiveness or suffer from ambiguity issues.

Purpose of the Study:

• To create a unified terminological resource for biomedical text mining.
• To enable the detection and grounding of specific biological entities to standard identifiers.
• To serve as a foundational component for advanced text mining applications.

Main Methods:

• Aggregating multiple manually curated terminological sources.
• Analyzing lexical properties, focusing on term ambiguity, particularly for protein names.
• Developing and applying term normalization and disambiguation algorithms.

Main Results:

• A comprehensive terminological knowledge base was compiled.
• An annotation application demonstrated the resource's utility on a protein interaction corpus.
• The annotation process included term normalization and disambiguation steps.

Conclusions:

• The developed resource effectively addresses the challenge of term ambiguity in biomedical text.
• Understanding lexical properties, especially ambiguity, is vital for efficient term grounding.
• This resource facilitates more accurate and reliable biomedical text mining.