Leveraging Social Media Big Data and Large Language Models to Quantify the Temporal Dynamics of Cross-Border Spatial Identity: Evidence from Shenzhen-Hong Kong Region

Abstract

Cross-border spatial identity refers to the multifaceted, individually constructed perceptions that residents form through cross-border interactions and daily behaviors. It has emerged as a critical metric for evaluating the integration of cross-border regions, playing a pivotal role in fostering regional community cohesion and advancing integration initiatives. In the context of the Guangdong-Hong Kong-Macau Greater Bay Area, the Shenzhen-Hong Kong corridor exemplifies as a significant yet complex cross-border zone, where divergent political systems, administrative practices, and cultural norms impede seamless integration despite a shared national affiliation. Consequently, examining residents’ subjective identity perceptions within cross-border spaces, as reflected in their daily activities, becomes imperative. However, the social, cultural, and linguistic heterogeneity in cross-border spaces complicates the clear definition of measurement dimensions, hampers the collection of long-term, high-quality data, and limits the ability to discern complex, multi-dimensional emotions. As a result, existing research has not adequately captured the temporal dynamics and addressed the multi-dimensional heterogeneity of cross-border spatial identity. To address these challenges, this study introduces an innovative measurement framework that integrates extensive long-term social media data with the semantic analysis capabilities of large language models. Focusing on the Shenzhen-Hong Kong cross-border space as a case study, we analyzed a comprehensive dataset of social media text from platforms such as Weibo and Xiaohongshu, covering the period from 2018 to 2023. By employing the chain of thinking prompt engineering of large language models (LLMs), we quantitatively assess the identity perceptions of Shenzhen and Hong Kong residents across three dimensions: spatial imagery, spatial satisfaction, and spatial sense of belonging. Our findings reveal that the spatial identity of Hong Kong residents experienced marked temporal fluctuations during critical events and policy shifts, such as the COVID-19 pandemic and the implementation of the Greater Bay Area development plan. Moreover, distinct variations in identity perception were observed across different types of cross-border spaces, with shopping and consumption areas displaying a notably positive spatial identity. This research provides a robust, efficient and scalable framework for the long-term tracking of cross-border spatial identity, offering valuable theoretical insights and quantitative evidence to better understand the temporal and spatial evolution of identity perceptions in cross-border populations. It offers important theoretical support and quantitative evidence for the evaluation of the cross-border region integration process.