Analysis and auditing of operational data using blockchain and NLP technology

With the increasing requirements for production safety, the authenticity and tamper-resistance of operation data in maintenance have become key challenges. This paper proposes a maintenance operation data processing and auditing system that combines blockchain [1] and natural language processing (NLP) technologies. By using the GPT2 [2] model and the spaCy library [3], it extracts key operation information operation quadruples <time, subject, object, action>from unstructured maintenance data. It combines hash algorithms and blockchain technology to achieve tamper-proof data storage and auditing verification. Experiments show that this study improves the accuracy of information extraction by 41.2% compared to traditional regular matching methods, 11.5% compared to the BERT model, and 10.2% compared to the Roberta model, while saving more than 60% of training time. It also supports fine-grained traceability of unstructured logs. This paper provides an innovative solution for data transparency and security in multi-link maintenance environments.