Natural language processing and network modeling for policy analysis: Explaining car sharing governance in Rome and Milan

Car sharing has emerged as a strategic tool for advancing sustainable urban mobility and reshaping business models within the transportation sector. This study compares the managerial and policy factors of car sharing in Rome and Milan by text mining 199 public documents and institutional reports published between 2009 and 2025. Employing natural language processing (NLP) methods, including TF-IDF keyword analysis, temporal evolution analysis, and co-occurrence network modeling, the research identifies the dominant discourses and strategic priorities surrounding car sharing in both cities. Findings indicate that Rome relies on a highly interconnected governance framework focused on foundational infrastructure, spatial planning, and action-oriented regulations. In contrast, Milan exhibits a modular and decentralized ecosystem approach, prioritizing service innovation, electrification, and multi-modal integration through adaptive governance. Temporal tracking shows that policy narratives evolve not only across broad developmental phases, but also through different patterns of policy duration, continuity, and intermittent salience over time. By contextualizing these policy narratives within actual market conditions, the study yields critical practical implications for each city. Ultimately, it provides a valuable historical roadmap for emerging smart cities to bypass preliminary experimental stages and directly implement mature sustainable transport strategies.