Link prediction in supply chain enhanced projection networks based on time-aware similarity metrics

Link prediction is a challenging and costly issue in supply chain networks. However, most existing models rely on simple static networks, which are insufficient for complex, dynamic networks. This paper proposes a link prediction model for enhanced supply chain projection networks using time-aware similarity metrics. By analyzing a dataset of domestic new energy passenger vehicle supply chains, we construct a supplier-manufacturer network and create multiple time-sliced snapshots. Projection analysis on these snapshots reveals potential link sets, and a time-aware similarity measure based on temporal events is used to determine prediction scores. Empirical analysis shows that the proposed model outperforms conventional static similarity metrics in accuracy and efficiency.