Integrating data mining with KJ method to classify bridge construction defects

► The authors analyzed common bridge construction defects and classifed them into appropriate groups. ► A constraint-based clustering method and affinity diagram (KJ method) are proposed. This method simultaneously treats mixed data types and incorporate user-specified constraints. ► The results revealed five major defect classifications: safety and environment, construction site defects, supervision/control process, construction quality documents, and others. ► Among the five groups, construction site defects such as backfill, structure ranks the highest, approximately 33%. The construction quality document ranks second at approximately 29%. Each unit-in-charge may refer to the results to prevent recurrences.