Sentiment AI for Maintenance Prioritization
The Sentiment AI (SAI) project explored how Natural Language Processing can help the Military Sealift Command (MSC) improve the way ship maintenance work orders are prioritized. Today, MSC relies on manual, subjective interpretation of maintenance descriptions, an approach that can lead to inconsistencies, delays, and reduced fleet readiness.
Old Dominion University evaluated whether sentiment analysis could automatically assess the urgency of work orders by analyzing the language used in maintenance logs. The study found that while SAI shows promise, several real-world challenges must be addressed before deployment. MSC’s data contained thousands of misspellings, inconsistent terminology, repeated failure descriptions, and conflicting criticality assignments. These issues limited the accuracy of existing NLP models, which are not designed to interpret maritime technical jargon.
The project concludes that successful adoption of SAI will require improved data quality, standardized maintenance language, and the development of a domain-specific maritime lexicon tailored for machine learning. With these foundations in place, sentiment-based automation could help MSC move toward faster, more objective, and more data-driven maintenance prioritization.