Hausam Consulting

Previous work has suggested the potential advantage of including advanced description logic reasoning capabilities extending beyond the OWL 2 EL profile for SNOMED CT classification. Rector and Brandt argued that many of the difficulties identified in SNOMED could be more easily dealt with using a more expressive description logic language beyond the EL profile which is used for SNOMED CT. A more expressive language could bring significant benefits, including a uniform structure for context and negation, particularly for representation of part-whole relations and determining semantic equivalence between findings and observables. To achieve this, the logic formalism would need to include negation and disjunction, which the EL profile does not provide. Prior work looked at modular reasoning techniques, distributing the reasoning workload between an expressive DL reasoner for the advanced reasoning tasks coupled with an efficient reasoner for the majority of the reasoning tasks, in theory enhancing the reasoning performance overall. Unfortunately, in practice the performance using modular reasoning has not been shown to live up to the promise. Other expressive and capable DL reasoners may be considered to achieve the goal of high performance classification of SNOMED CT while allowing the desired additional expressive reasoning capabilities. Konclude is one such reasoner showing potential for providing these capabilities for SNOMED CT classification. Key Konclude features include a highly optimized implementation of tableau calculus enhanced with preprocessing and tableau saturation, and additionally the use of multiple cores within a shared memory environment for increased reasoning performance.