An NLP-Driven Analysis of Technical Contributions and Governance Influence in 5G

Abstract

This thesis addresses a critical gap in understanding how companies and countries influence 5G technical standards within the Third Generation Partnership Project (3GPP). Given the immense volume and complexity of 3GPP documentation, traditional analysis methods are impractical. To overcome this, a novel Natural Language Processing (NLP) pipeline was developed, featuring a four-tier impact scoring system for Change Requests (CRs) and a semi-supervised annotation approach using GPT-4o, followed by fine-tuning a RoBERTa-base model for large-scale classification. This methodology quantifies technical contributions, assesses influence through governance roles (like chair positions), and identifies strategic priorities across technology areas in 5G Release 18. Key findings reveal that technical influence is highly concentrated, with Huawei, Nokia, and Ericsson leading among companies, and China dominating at the country level. National strategies distinctly vary: China and the United States present broad industrial contributions, whereas Finland, Sweden, and South Korea exhibit a concentrated contribution model, where one or two dominant firms drive the majority of technical input. The United States demonstrates significant governance influence, often disproportionate to its technical contributions, highlighting a strategic focus on procedural control. Analysis of technology areas shows China's broad dominance, the U.S.'s focus on foundational radio technologies, and South Korea's emphasis on advanced network architecture. This study provides the first quantified, large-scale analysis of 5G standardization influence, validating geopolitical narratives and offering empirical insights for policymakers, while also demonstrating a robust NLP methodology for analyzing complex technical datasets.