Where to search top-K biomedical ontologies?

Area of Science:

• Biomedical Informatics
• Ontology Engineering
• Information Retrieval

Background:

• Searching for precise biomedical terms in ontologies is challenging due to concept overlap.
• Existing search engines often return extensive, unfocused results, reducing user confidence.
• A systematic evaluation of ontology search engines is needed to understand their performance.

Purpose of the Study:

• To systematically evaluate and compare information retrieval (IR) ranking algorithms for biomedical ontologies.
• To assess the strengths and weaknesses of different ontology search engines.
• To provide recommendations for selecting appropriate search mechanisms.

Main Methods:

• Implemented and compared seven IR ranking algorithms against four ontology search engines.
• Conducted free-text queries and expert-based evaluation of search results.
• Developed a probabilistic ground truth for broader query evaluation.

Main Results:

• Identified key search factors and requirements for biomedical ontologies.
• Evaluated the performance of ranking algorithms and search engines against expert-based ground truth.
• Established a benchmark for evaluating ontology search tools.

Conclusions:

• The study provides recommendations to aid experts and engineers in selecting the best retrieval mechanisms.
• This evaluation enhances the reliability of current search techniques for biomedical data.
• Researchers can utilize these findings to improve their daily work with ontologies.