Understandig Territory from an Online Perspective: Twitter and New Ways to do Research on Urban Phenomena

Abstract

In the era of the data revolution (Kitchin, 2014), new kinds of data and data sources allow researchers to share innovative ways of studying society and its dynamics to comprehend the peculiarities related to the territory. In fact, the analysis of space-time data extracted from social media can support the comprehension of complex social phenomena such as, for example, mobility (Noulas, Scellato, Lathia, & Mascolo, 2012; Yuan et al., 2017; Zhang et al., 2016) or where and when certain social activities are carried out, such e.g., the participation in big events, tourist experiences or the monitoring and management of emergencies and security.

Following the actual dissemination of use of social media in the cities, the analysis of the spatial dimension is considered particularly promising in the field of urban analysis (Singleton et al., 2018) and in the emerging line of digital studies.  Nowadays, the intentional action of many people is increasingly contributing to the production and sharing of spatial data with a relevant impact on the construction of Territorial Information Systems (Zupi, 2017). This new category of spatial data is commonly known "Volunteered Geographic Information" (VGI) (Goodchild, 2007). The analysis of this kind of data leads to an innovative scenario for the collection and diffusion of geographic information from millions of users all over the world. It provides valuable insight into their perceptions and needs, opinions on places, events, and daily routes, which allow researchers to better follow traces about the knowledge of the local identities of a given place (Campaign, 2014), information flows, and social networks within society (Stefanidis et al., 2013).

However, a clear awareness of the limits of these tools of investigation is needed. For instance, problems may arise regarding representativeness of population, or even technical issues, such as for scarping systems which are not able to offer all the effective amount of data available to the researcher.  Despite this, these kinds of studies are taking hold in the social sciences which are paying particular attention to both data and source contextualisation.  As a result, they are interesting and valid tools to study issues that classic instruments are not able to perform.

Among the different kinds of digital platforms, Twitter allows the combination of textual and geo-referenced data, making this platform an interesting choice, for instance, to detect or monitor different kinds of events (Korkmaz et al.,2015), or it could be used to study the dissemination of influence (Generous et al., 2014).  Algorithms are, in this way, extraordinary tools for organising and spreading information. According to these claims, our work aims to understand, both theoretically and methodologically, how the research about territory, as a field of study, has been adapted to prolifically use data coming from algorithms of geo-location or geoparsing (Middleton et al, 2018).Specifically, this work aims to critically discuss the elements and the steps that characterize this emerging approach in order to highlight how the use of geo-localized data can be advantageous for territorial analysis and for which kinds of phenomena..Thus,  after showing the most commonly used ways of collecting information about territory using algorithms, we will discuss strategies and techniques of analysis used to combine online (social media) and offline (territorial) sources and try to identify the possible applications for each type of technique. Moreover, a third step allows us to identify the different meanings and operational processes made regarding the concept of territory. That will allow us to recognize the different conceptions of sociological space emerging from the analysis and suggest how to avoid the use of geo-localized data with a simple data-driven perspective (Kitchin, 2014) that does not consider the sociological peculiarities of the space. We will try to answer these questions by examining papers that used Twitter as the source of data. In particular, the analysis of articles will be led thanks to a third kind of content analysis following Rositi’s categorisation (Rositi, 1988). Also, the extraction of the articles will be made using PRISMA methodology (Moher et. Al, 2018) for systematic review.