We investigate the suitability of case-based reasoning supporting experts during technical diagnosis. Diagnosis is not considered as a classification task, but as a process to be guided by computer assisted experience. This corresponds to a flexible "case completion" approach. Flexibility is also needed for the expert view with predominant interest in the unexpected, unpredictable cases. Integration of further knowledge sources (domain knowledge, common knowledge) is investigated in the project. Two different approaches based on semantic nets are investigated: Fallexperte-D makes use of Case Retrieval Nets (CRN) that are based on spreading activation techniques. Falldaten-DL employs description logics enhanced by concrete domains.
展开▼