Toward greener logistics: uncovering key enablers of the physical internet using AI-powered theme analysis

Abstract

Transitioning to a zero-carbon, sustainable logistics ecosystem requires a fundamental shift in how physical, digital, and organizational systems interact. The Physical Internet (PI) presents a transformative vision for logistics and supply chain management by providing a blueprint for decarbonized, circular, and resilient operations. However, the complex, interdisciplinary nature of its knowledge base presents challenges for coordinated global implementation. This study introduces a dual-model natural language processing (NLP) approach combining transformer-based topic modeling (BERTopic) with maximal marginal relevance (MMR) and generative pretrained transformer (GPT) techniques. This hybrid approach enables the extraction and synthesis of key research themes from over 2600 scientific publications on PI. Thematic analysis revealed eight critical domains, ranging from smart infrastructure and energy systems to cybersecurity and governance that are foundational to PI’s sustainable development and adoption. Furthermore, we evaluated the alignment of these themes with the PI roadmaps and the UN sustainable development goals (SDGs), especially SDG 9 (Industry, Innovation and Infrastructure), SDG 11 (Sustainable Cities and Communities), and SDG 13 (Climate Action). Results highlight the importance of interoperability, digital twin technologies, renewable energy integration, and secure data exchange for achieving greener and more adaptive logistics networks. This work provides a scalable, data-driven methodology for strategic decision-making and knowledge synthesis, thereby supporting the sustainable transformation of logistics and supply chains.