Connected Cities: Deciphering City Relationships Using Collocation Analysis

Cities form a variety of relationships that facilitate cooperative development. While it is well-recognized that intercity relationships play critical roles in boosting city performance and shaping regional policy, conducting large-scale empirical analyses of these intercity relationships presents substantial challenges. A primary obstacle is the scarcity of city relational data. Collocation analysis, also known as toponym co-occurrence in geography, presents a novel approach to this challenge. While preliminary investigations have shown collocation patterns are effective in capturing city relationships, there remain several unanswered questions. This thesis explored the potential of collocation analysis to extract and interpret city relations from a large textual database, guided by the central question: To what extent and how can collocation analysis be used to extract, categorize, analyze, and evaluate relations between cities? This central question is further explored through four sub-questions, each addressed in a dedicated chapter. Chapter 1 is the introduction, presenting the motivation, theoretical background, research questions, and outline of the thesis. Chapter 2 introduced an efficient and straightforward method to extract collocation patterns from a large text database, and illustrated through a case study on extracting collocation patterns of 293 Chinese cities from Common Crawl web archive. This proposed method is adaptable to other Common Crawl archives, and the generated dataset offers a rich resource for subsequent research in city network analysis. Chapter 3 delves into the multiplexity of city relationships, acknowledging that the general collocation patterns identified earlier provide a broad overview. To address this, Chapter 3 introduced a lexicon-based method to refine the collocation-based city relationships captured in Chapter 2, classifying them into six categories: industry, information technology (IT), finance, research, culture, and government. Chapter 4 highlighted the value of collocation-based relationships in addressing empirical challenges at the city scale. It highlighted that due to the lack of city relational data, there were few empirical studies on the comparative advantages of a city’s network position versus its agglomeration benefits. Using collocation patterns derived from Chapter 2, an intercity network was constructed and analyzed, revealing that the retrieved collocation patterns reflected reality to some extent, especially mirroring the backbone structure of national networks identified through other intercity relationship data. The analysis showed that city network externalities were more important in explaining urban performance than agglomeration externalities, with smaller cities benefiting more from a strong network position. Chapter 5 highlighted the potential advantages of using collocation patterns in confronting planning on regional and national scales, with a focus on the development of large-scale regional plans, or “megaregions.” It pointed out that there was an empirical gap in creating effective policies for megaregion planning, attributed to a lack of understanding of the functional coherence within these vast planned areas. To address this gap, Chapter 5 developed a three-dimensional evaluation framework to assess the functional coherence of planned megaregions: inclusion, integration, and consistency. Chapter 6 summarizes the main conclusions and discusses policy implications. It also discusses the limitations of this dissertation and suggests directions for future research.