BERTopic Modeling Analysis of Privacy Protection Trends in New Trade Rules for the Mobility Industry

The rapid transformation of the mobility industry has led to an unprecedented volume of personal data, posing new privacy challenges within evolving trade rules. This study applies BERTopic modeling, a BERT-based approach, to systematically analyze 187 screened academic papers (2020-2025) on privacy regulations in the mobility sector. The final corpus was assembled through database-driven filtering (initial n = 321, deduplication, relevance screening). Unlike traditional topic models (e.g., LDA), BERTopic's contextual embeddings facilitate superior semantic understanding, which is crucial for nuanced interdisciplinary analysis. Key hyperparameters include UMAP (nneighbors = 15), HDBSCAN (mincluster = 10), and c-TF-IDF weighting. Through BERTopic analysis, five distinct research themes emerged, health security, user acceptance, location protection, smart city innovation, and autonomous vehicle AI integration, demonstrating its superior performance over LDA in topic coherence (+26.6%), diversity (+8.9%), and keyword accuracy (+19.9%). A comparative policy analysis of GDPR, CCPA, and Korea's PIPA reveals significant regulatory gaps. The findings provide critical insights for policymakers, highlighting the need for harmonized mobility-specific privacy frameworks and offering technical recommendations such as differential privacy for UAM and federated learning for traffic prediction. This study demonstrates that context-aware topic modeling reveals intricate and evolving trends in mobility privacy, addressing the limitations of prior research that lacked the granularity to capture such complexities.