Building a Sustainable Knowledge Management System from Dark Data in Industrial Maintenance

As digitalization exponentially generates vast data volumes, concerns about its environmental impact have surged. Dark data, characterized by being stored but underutilized, not only presents challenges to organizations but also aligns with the imperative of digital decarbonization. Notably, the manufacturing domain lacks prior exploration of addressing dark data issues. In response, this study introduces a knowledge management perspective to tackle the challenge through the construction of a Knowledge Management System (KMS). The proposed KMS is built upon the foundation of a Knowledge Graph (KG). Data ingestion analysis reveals that the identified data sources are sparse and incomplete, prompting the application of data scraping and enrichment techniques. Data from various sources is integrated, and knowledge is extracted from the enriched datasets. A fault KG containing the physical level and the failure level is manually constructed to ensure enhanced accuracy and reliability. The proposed KMS framework, grounded in the power of KGs, serves as a comprehensive solution to the challenges posed by dark data in the manufacturing sector, and provides industrial maintenance applications such as fault analysis and decision-making guidance, to improve knowledge reuse and promote digital decarbonization.