AI news ·
Revolutionizing Navy Supply Chains: How AI-Powered Sentiment Analysis Enhances Contractor Performance Evaluations
The STARS project uses AI to enhance Navy supply chain by automating contractor evaluations with sentiment analysis. It improves accuracy and flags inconsistencies in feedback.
STARS Project Enhances Navy Supply Chain with AI
The Naval Undersea Warfare Center Division, Keyport, in collaboration with Virginia Tech and the Naval Sea Logistics Center, is advancing Navy supply chain efficiency through AI. Their initiative, the Sentiment and Topic Analysis for Reliable Supply (STARS) project, focuses on using large language models (LLMs) to improve the consistency and accuracy of contractor performance evaluations.
Currently, contractor assessments combine numerical ratings with narrative comments that often don’t match, complicating decision-making for future contracts. STARS addresses this by automating the analysis of both scores and text to deliver more objective and reliable evaluations.
How AI Brings Clarity to Contractor Evaluations
At the core of STARS is sentiment analysis powered by LLMs trained to detect positive, negative, or neutral sentiment in text. These models identify discrepancies where written feedback contradicts numerical scores, helping to flag inconsistencies for review.
Brett Davis from the Naval Sea Logistics Center highlights that improving narrative quality enhances supply chain risk management. This approach reduces the impact of different writing styles, experience levels, and the inherent difficulty in interpreting tone through text.
However, challenges remain. For example, LLMs struggle with sarcasm or implied tone, as John Greener, NSLC’s CTO, explains. Written phrases like “This individual was great” can be interpreted differently depending on context, which AI cannot always discern.
Training AI Models with Proxy Data
Due to the sensitive nature of actual contractor data, Virginia Tech researchers use publicly available datasets, such as movie reviews, to train and refine the AI models. These proxies simulate sentiment analysis tasks while the team seeks approval for access to Navy-specific data.
Broader Implications and Future Applications
The STARS project has potential beyond contractor assessments. It could improve how acquisition-related documents are drafted and help predict contractor performance more accurately. This initiative may benefit not only the Navy but also the broader Department of Defense and federal agencies.
John Greener notes that two years of collaboration with Virginia Tech have revealed additional use cases for this technology, indicating promising opportunities for adoption across government supply chains.
About NUWC Division, Keyport
Located near Seattle, Washington, NUWC Division, Keyport supports Fleet operations across major Navy Pacific homeports. It operates detachments in San Diego and Honolulu, with additional sites in Guam, Japan, Nevada, and Virginia. The center’s team of engineers, scientists, and technicians focus on maintaining undersea warfare superiority for the United States.
For IT professionals interested in AI applications similar to those in the STARS project, exploring courses on sentiment analysis and large language models can be a valuable step. Resources like Complete AI Training's latest AI courses offer practical learning paths.