Predicting potential members of a logistics network platform using recommender systems

Abstract

Logistics network platforms help logistics service providers (LSP) to digitize their operations. Choosing the right LSP as potential platform members is a crucial task for the platform’s growth. In this work, a recommender system will be applied to a data set with the goal of predicting potential members of a logistics network platform, which represents a novel method in this field. In the scope of this work, data of European LSP was collected, analyzed, and provided to a neural networks-based model, namely a two-tower recommender system. The recommender system was generally able to deliver proper results, having a top K accuracy of around 60\% in training. The application of recommender systems in logistics remains a promising prospect, where network data for the relationship between networks and LSP can yield a robust model.