Positive-unlabeled graph ranking for assembly-center-oriented resilience-green sustainability evaluation of China Railway Express node cities

China Railway Express (CR Express) has become an important rail-based logistics corridor for improving Eurasian connectivity, regional openness, and sustainable freight organization. However, evaluating its node cities remains challenging because assembly-center labels are incomplete, city performance is network-dependent, and logistics development is increasingly shaped by resilience and green-sustainability requirements. This study proposes a positive-unlabeled graph ranking framework for CR Express node-city evaluation. The framework integrates statistical indicators, logistics operation variables, railway logistics relation structures, ecological governance indicators, and policy-text semantic representations. Existing assembly centers are treated as observed policy-designation positive anchors, while other node cities are modeled as unlabeled samples rather than negative cases. Accordingly, the model output is interpreted as relative assembly-center-oriented planning suitability under the observed operational, network, resilience, green-sustainability, and policy characteristics rather than as a causal estimate or a probability of future official designation. To reduce post-treatment information leakage, both ex-ante validation using 2011–2019 data and current assessment using 2011–2023 data are conducted. Experiments on 33 CR Express node cities show that the proposed model achieves an NDCG@5 of 0.932 ± 0.025, with auxiliary AUC and macro-F1 values of 0.948 ± 0.015 and 0.875 ± 0.021, respectively. Changsha, Wuhan, Jinan, Shenyang, and Hefei are identified as unlabeled cities with relatively high model-based planning suitability. The results indicate that railway logistics node evaluation benefits from jointly modeling incomplete labels, composite logistics relations, operational resilience, policy semantics, and green-sustainability constraints.