Application of Large Language Models for Container Throughput Forecasting: Incorporating Contextual Information in Port Logistics

2026-02-24

AI Summary

This study addresses the challenge of effectively integrating multi-source contextual information in port container throughput forecasting. It introduces, for the first time, a large language model (LLM) to this domain and proposes an innovative prompting mechanism that incorporates multi-level contextual features of port operations. Through tailored prompt engineering, the approach guides the LLM to synthesize heterogeneous, multidimensional data sources relevant to container traffic prediction. Experimental results demonstrate that the proposed method significantly outperforms existing benchmark models, achieving higher forecasting accuracy and underscoring the effectiveness and potential of generative AI in complex logistics scenarios.

Technology Category

• Planning, Routing, and Scheduling: Planning with Language Models
• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Generation

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

Recent advancements in generative artificial intelligence (AI) have demonstrated its substantial potential in various fields. However, its application in port logistics remains underexplored. Ports are complex operational environments where diverse types of contextual information coexist, making them a promising domain for the implementation of generative AI and highlighting the urgency of related research. In this study, we applied a large language model (LLM)-a leading generative AI technique-to forecast container throughput, which is a critical challenge in port logistics. To this end, we adopted a state-of-the-art LLM approach and proposed a novel prompt structure designed to incorporate the contextual characteristics of port operations. Extensive experiments confirm the superiority of our method, showing that the proposed approach outperforms competitive benchmark models. Furthermore, additional experiments revealed that LLMs can effectively learn and utilize multiple layers of contextual information for inference in port logistics. Based on these findings, we explore the key constraints affecting LLM adoption in this domain and outline future research directions aimed at addressing them. Accordingly, we offer both technical and practical insights to support the effective deployment of generative AI in port logistics.

Problem

• container throughput forecasting
• port logistics
• large language models
• contextual information
• generative AI

Innovation

• Large Language Models
• Container Throughput Forecasting
• Prompt Engineering
• Contextual Information
• Port Logistics