Knowledge graph-based recommendation method for cold chain logistics

The cold chain logistics context-aware recommendation containing time, location, environment, activity, user device status and other information has good recommendation accuracy. However, traditional cold chain logistics context-aware recommendation still has the drawbacks of lack of semantic features and insufficient interpretability. This study proposes a knowledge graph-based recommendation method for cold chain logistics (KGRCCL). KGRCCL includes four modules: data mining and dynamic fusion module of multi-source heterogeneous cold chain logistics, construction module of cold chain knowledge graph, context-aware recommendation module of cold chain and cold chain recommendation module fused with cold chain knowledge graph. Using one real-life dataset of cold chain info and six other recommendation methods, we demonstrate that KGRCCL can reduce the mean absolute error (MAE) and root mean square error (RMSE) of the dataset by about 12%–25%. It is anticipated that the outcome from this research would be useful in the recommendations of cold chain logistics distribution in the future.