Contrasting lexical similarity and formal definitions in SNOMED CT: Consistency and implications

- SNOMED CT is a large and complex terminology and imperfections are inevitable.
- We propose an algorithmic quality assurance method to find inconsistent concepts.
- We formulate similarity sets: groups of concepts with similar lexical structure.
- Five different set types are formed and a sample of each is evaluated.
- Evaluation of the samples revealed 38-70% of the sets having inconsistencies.