Defect factors and their relevant rules can be analyzed in depth by processing defect records which are often expressed in the form of text data. However

considering that defect text consists of both structured and unstructured data

it is necessary to excavate structured information from unstructured data. In this paper

a text mining method based on semantic framework technology is introduced to transform unstructured defect description into structured information such as components and defect attributes. Then

a deep analyzing model of a power equipment defect is established

which provides a scheme of defect mining based on historical defect texts. Case studies prove that the proposed deep analysis method has a guiding significance for equipment upgrading

selection and maintenance.