This study investigates how the relationships between cities are represented in local news by analyzing co-mentions of cities in 31,004 news articles from Alabama. Using a large language model, we extract geographic references and construct co-mention networks that reflect both spatial proximity and symbolic connections. To interpret these links, we develop a classification framework of relationships between cities, including common impacts and sequential dynamics. Our preliminary analysis reveals that different news categories produce distinct patterns of spatial association. This approach offers a novel methodology and a new lens for understanding media-driven spatial imaginaries, contributing to research on information geographies and shedding light on the relational dimensions of local information.