Collaborative filtering based talent development algorithm in sustainable modern logistics management project

Abstract

ABSTRACT Logistics is crucial to the global economy, but traditional logistics management models have issues such as low informationization, single structures, low efficiency, and insufficient recommendation accuracy. This study aims to improve personalized recommendation in logistics management using the collaborative filtering algorithm and comparing it with Logistic Regression (LR) and Factorization Machine (FM) algorithms. The modified cosine similarity calculation method had the lowest absolute error at 0.91. After improving the algorithm, the PR value increased to 0.9952, showing a better balance between accuracy and recall. Combining the improved collaborative filtering recommendation algorithm with the actual logistics management model resulted in higher accuracy in personalized recommendations to users. The accuracy also increased with the number of recommended companies. Overall, the improved algorithm showed promising effectiveness and feasibility in the context of logistics management.