Modeling interurban mentioning relationships in the U.S. Twitter network using geo-hashtags

Twitter users mention cities in the context of tourist attractions or events, such as protests or games, thus forming a network between cities from which they tweet and cities that they tweet about. This study tackles the challenge of explaining why users tweet about cities outside of their own by analyzing an underlying network of city mentions on Twitter. It applies graph theory as well as various measures of network connectivity such as indegree, hub score, and authority score to examine the prominence of individual cities in the Twitter landscape and the connection patterns between cities. Closely related to communication ties is the sentiment of tweets about other cities, which can be extracted from the text of tweets that contain geohashtags, i.e., hashtags with names of other cities. The effect of distance between cities on user sentiments towards cities will be explored. Furthermore, Quadratic Assignment Procedure (QAP) network regression will be used to build a general socio-demographic and geographic model that helps to identify which characteristics of city pairs, e.g. separation distance, or similarity in employment data or population, increase or decrease the likelihood of mentions between those cities. Findings show that distance and network size (compactness) are major determinants in communication ties between cities. City popularity, when measured by indegree, follows a power-law distribution, and is closely tied to population, GDP, or visitor numbers. Larger cities reveal a higher percentage of self-mentions than smaller cities, showing the high level of attention these metropolitan areas attract from Twitter users due to the many opportunities, events, and sights offered. Future research in the field of analysis of geotagged tweets can further extend the network regression model with new covariates.