Relating Complexity and Error Rates of Ontology Concepts. More Complex NCIt Concepts Have More Errors

Area of Science:

• Biomedical Informatics
• Knowledge Representation
• Ontology Engineering

Background:

• Ontologies are crucial knowledge structures supporting health-information systems.
• Assessing the quality of ontological concepts is vital for reliable data.
• Concept complexity is a potential factor influencing ontology quality.

Purpose of the Study:

• To evaluate the relationship between concept complexity and error rates in ontologies.
• To determine if a measure of lateral complexity can predict concept quality.
• To inform quality assurance strategies for health-related ontologies.

Main Methods:

• A lateral complexity measure (number of role types) was used to categorize concepts.
• Concepts from the National Cancer Institute thesaurus (NCIt) Biological Process hierarchy were analyzed.
• A two-phase quality assurance (QA) process identified and verified errors in concept modeling.

Main Results:

• Errors such as missing, incorrect, or misassigned roles were identified.
• A statistically significant difference in error rates was observed between complex and simple concepts.
• The study confirmed the hypothesis that increased concept complexity correlates with higher error rates.

Conclusions:

• Quality assurance (QA) is indispensable for maintaining ontologies.
• More complex concepts, defined by role types, tend to have higher error rates.
• These findings provide a basis for targeted QA efforts in ontology development, particularly for the NCIt.