A hybrid approach for developing, extending, and implementing industrial maintenance knowledge graphs and semantic ontologies to support smart maintenance diagnostics
The unstructured historical data stored in Computerized Maintenance Management Systems (CMMS) is a mine of maintenance diagnostic information. This data is often underused due to its unstructured and informal nature. This thesis will propose a framework for transforming maintenance log data, which is often in the form of natural language text, into formal knowledge graphs. The proposed method generates a knowledge graph that encodes the semantic relationships between multiple maintenance entities based on the historical data that can be found in maintenance work orders. The knowledge graph is created semi-automatically through the hybrid application of text analytics techniques and human-assisted semantic tagging of maintenance work order text. The semantics of the knowledge graph proposed in this research will be provided jointly by a SKOS thesaurus and an OWL ontology. SKOS (Simple Knowledge Organization System) and OWL (Web Ontology Language) are both Semantic Web standards that will enhance the reusability and portability of the final knowledge graph. The knowledge graph created as an output of a java based tool will become an opensource shared industrial maintenance knowledge base that can be extended incrementally and be used for various decision support applications, including maintenance diagnostics and root-cause analysis. An online knowledge graph platform will be used to conduct querying and inferencing over the graph to support smart maintenance diagnosis.
Knowledge graph, Thesaurus, Natural Language Processing, Ontology
Tahsin, R. (2022). <i>A hybrid approach for developing, extending, and implementing industrial maintenance knowledge graphs and semantic ontologies to support smart maintenance diagnostics</i> (Unpublished thesis). Texas State University, San Marcos, Texas.