Development of a Combined System Based on Data Mining and Semantic Web for the Diagnosis of Autism

Introduction: Autism is a nervous system disorder, and since there is no direct diagnosis for it, data mining can help diagnose the disease. Ontology as a backbone of the semantic web, a knowledge database with shareability and reusability, can be a confirmation of the correctness of disease diagnosis systems. This study aimed to provide a system for diagnosing autistic children with a combination of semantic web and data mining. Method: Data is taken from the UCI database. There were a total of 292 data records available of which 80% (234 records) were used for modeling through the decision tree. Knowledge about patients and autism disease was presented via ontology using the Protégé 5 software. The ontology has four classes and 12 properties to communicate between the individuals in the classes. The rules extracted from the decision tree were transformed into a comprehensible form (SWRL) for interpretation in the ontology using a converter. Results: Whether the child is healthy or not can be determined by the rules obtained in the decision tree. In addition, the output of the ontology using the interpretation of 25 rules confirmed the diagnosis of an Autistic child using the decision tree. The evaluation of the ontology also confirmed its correctness. Conclusion: According to the similarity between the result of the ontology and the decision tree regarding the diagnosis of the disease, the accuracy of the proposed method can be emphasized.