Predicting the extension of biomedical ontologies

Area of Science:

• Biomedical Informatics
• Computational Biology
• Ontology Engineering

Background:

• Biomedical ontologies require continuous development due to evolving life science knowledge.
• Extending ontologies is labor-intensive, necessitating automated support for expert researchers.

Purpose of the Study:

• To develop and evaluate a strategy for automating change detection in biomedical ontology extension.
• To predict areas of an ontology likely to require extension in future versions using supervised learning.

Main Methods:

• Applied supervised learning to features of previous ontology versions to predict future extension areas.
• Utilized the Gene Ontology as a test case for the proposed prediction strategy.
• Compared the strategy's performance against existing state-of-the-art change capturing methods.

Main Results:

• Achieved an average f-measure of 0.79 for predicting extensions in a subset of biological process terms.
• Demonstrated superior performance compared to current state-of-the-art change capturing techniques.
• Identified key challenges and outlined a general framework for ontology evolution prediction.

Conclusions:

• The proposed supervised learning strategy effectively predicts areas for ontology extension.
• This approach can enhance the efficiency of manual or semi-automated biomedical ontology extension.
• The framework is applicable to any versioned biomedical ontology, optimizing expert focus.