Expressing Biomedical Ontologies in Natural Language for Expert Evaluation

Area of Science:

• Biomedical Informatics
• Natural Language Processing
• Ontology Engineering

Background:

• Biomedical ontologies are crucial for data standardization and knowledge representation.
• Translating complex ontological axioms into human-readable natural language is challenging.
• Automated tools can facilitate the interpretation of biomedical knowledge.

Purpose of the Study:

• To evaluate the efficacy of the custom Hootation software in generating natural language phrases from biomedical ontologies.
• To assess the clarity and accuracy of the generated natural language.
• To determine the fidelity of the natural language to the source axioms and represented domain knowledge.

Main Methods:

• Utilized Hootation software to process axioms from three biomedical ontologies.
• Employed multiple domain experts for evaluation.
• Applied three discrete rating scales to assess clarity, axiom fidelity, and knowledge representation fidelity.

Main Results:

• Hootation successfully produced relatively clear natural language equivalents for selected OWL axioms.
• The clarity of the generated phrases was contingent upon the accuracy and quality of the axiom representation within the ontology.
• Evaluations indicated moderate success in maintaining fidelity to the source axioms and represented knowledge.

Conclusions:

• Hootation demonstrates potential as a tool for generating natural language from biomedical ontologies.
• The quality of ontological axiom representation is critical for achieving clear and accurate natural language output.
• Further refinement of Hootation may enhance its utility in biomedical knowledge dissemination and interpretation.