Supply Chain Optimization Based on Large Language Models (LLMs): Global Supply Chain Risk Assessment

Abstract

In recent years, the rapid development of Large language models (LLMs) provides a new idea for supply chain risk assessment. In this paper, a global supply chain risk assessment method based on LLMs is proposed. By using named entity identification, relationship extraction, text classification and sentiment analysis, a knowledge map of supply chain risk is constructed, and a risk assessment model is constructed by combining machine learning or deep learning algorithm to achieve accurate assessment and early warning of global supply chain risks. Taking the "Ever Given" Suez Canal stranded in 2021 as an example, LLMs is used to extract features and assess risks. The results show that the model is significantly superior to the traditional methods in terms of comprehensiveness, timeliness and accuracy of risk assessment, and it has won the emergency response time for enterprises. However, LLMs still faces many challenges in data quality and diversity, model interpretability and robustness, and practical application. Therefore, this paper puts forward some countermeasures, such as strengthening data quality management, improving the interpretability and robustness of the model, and optimizing the practical application effect, in order to promote the wide application of supply chain risk assessment methods based on LLMs.