Managing and Assessing the Risk Management of Supply Chain Using the A-BiGRU-CNN Approach

Companies can now expand their competitive advantages, manufacturing flexibility, and cost savings by utilizing global supply networks that span the globe. Inadequate awareness of unknown geographical variances and changes often increases risks in supply chain operations and can even interrupt them totally, despite these benefits. This study proposes a two-stage, text-mining-based approach for managing risks in the world's supply chains. First, textual methods including phrase frequency, correlation, and bi-gram analysis were used to sift through the existing literature on the topic of global supply chain hazards. These studies shed insight on the significance of lexical content in the studied literature, and the topic model's ability to group similar topics together benefited in the identification of supply chain risks. Its results served as input into the construction of a risk classification (hierarchy) that takes into account seven different classes of risks and the variables contributing to them in the global supply chain. In the second stage, they selected online news items based on the specific type of risk, using these risk criteria, and then performed sentiment analysis to determine the pattern of risk variance. When applied to regional and global supply chains, the results of the risk hierarchy and sentiment analysis might be useful for risk assessment and supplier selection. The suggested method includes preparing data, extracting features, and training a model. Preprocessing makes use of data reduction, data cleansing, data integration, and data transformation. PCA and LDA are used for dimensionality reduction. A-BiGRU-LSTM is used for model training. The suggested method is compared to two established approaches, CNN and BiGRU, and fares better than both.