Scientific case

MathildeGuillemot_Carte 12_2013
© Mathilde Guillemot, Carte 12, 2013

Introduction

Maps are commonly defined as the visual representation of an area. They are codified depictions highlighting relationships between elements of a given space such as objects, regions and themes. The craft of map-making has a long and rich history (Macchi & Mullender, 1980). Initially used for navigational purposes, early maps sought to represent geographical space by positioning elements on a two-dimensional surface. As representational devices, maps quickly became integrated with science and the state (Turnbull, 2000). Indeed, maps have played a major role in the ways of representing, navigating, understanding, organizing and disputing the spaces we inhabit (Lacoste, 1976).

In recent years, computers and satellites have profoundly changed cartography. Geographical Information Systems now capture, store, manipulate, analyze and manage spatial data to inform research and decision-making. Satellite systems provide geo-spatial positioning to monitor movement anywhere in the world while millions access interactive map applications through their personal computers and smart phones.

Parallel to these developments, the visual representation of non-spatial datasets using digital mapping tools has fostered growing interest in the social sciences. These representations are often labeled as maps of unchartered territories where the emergence of complex social phenomena can be studied through heterogeneous data contained in documents, structured databases or the World Wide Web. The technological and epistemological shift this entails has been described by some scholars as a move towards the Digital Humanities. Understandably, these developments are particularly attractive to social scientists. Though statistical tools have played a major role in the study of social existence (Desrosières, 2008), the variety of tools available to social scientists has been rather limited compared to those available to colleagues in the natural sciences. Consequently, the advent of tools to collect relational data and analyze networks has fostered “great expectations”. For instance, some hope they will help bridge the methodological gap that separates the study of specific interactions from that of global structures (Venturini & Latour, 2010).

Considering the ever-expanding production and circulation of digital information it is imperative that scholars in the social sciences aquaint themselves with these new sources of data and the emerging methods that permit their analysis. An approach based on counting hits, nodes and links, though incomplete, may help clarify questions in various fields from sociology to science and technology studies and the political sciences about the increasing imbrication between web-based activity, algorithims and data in everyday life. For instance, we are already witnessing that, far from the utopic discourse that characterized early descriptions of the Web’s collective dynamics and decentralized architecture, the production and sharing of content on the web have not resulted in some neo-Habermasian ideal of the public-sphere. Instead, research in the political sciences has shown that the Web is a mosaic space (Rogers, 2013), or a balkanized landscape (Sunstein, 2008) that does not fit with the unified “small world” narrative usually associated with online research (Barbier & Cointet, 2012). This seminar, as described further below, will provide participants with a conception of the web as a space of social action and present participants an emerging set of empirical tools designed to apprehend the nature of this action.

Scientific case

Digital mapping tools (DMTs) of various types like Gephi, the CorTexT Manager, ScienceScape and Hyphe have been developed to help social scientists tracing and analyzing the content and circulation of textual data. Such tools are beginning to be applied in the study of various social phenomena (see for instance: Parasie & Cointet, 2012; Diminescu, 2012), mainly by scholars active in STS and Media Studies. Though prior methods may offer some assistance (i.e. statistical regression and correspondence analysis), social scientists trying to construct explanations using relational datasets and DMTs must still cross a “methodological no man’s land” (Gläser & Laudel, 2001). Indeed, basic questions related to the construction, interpretation and use of DMTs have received little attention in the social sciences. For instance, how to choose data, parameters and algorithms to investigate specific social phenomena is seldom addressed. Similarly, literature on how to explore and make sense of maps obtained through the configuration of these elements is generally lacking, and the articulation between quantitative and qualitative results also remains an understudied area. The politics of algorithms would also benefit from further inquiry (Gillespie, forthcoming), as would the responsibility of researchers towards the actors mapped using DMTs.

The development of DMTs has been driven by multidisciplinary work at the crossroads of computational linguistics, data mining, artificial intelligence, dynamical systems and network analysis. Discussions within these communities focus on internal issues such as the extraction, parsing, disambiguation, clustering, filtering and visualization of data. Though interesting, research pursued in computer science and complex systems analysis provides an insufficient methodological basis to support the application of DMTs in the social sciences. Publications in these fields mainly revolve around modeling, development and optimization activities. They offer little insights for social scientists trying to figure out how to mobilize DMTs to answer their own research questions.

Similarly, though the representation of data in the form of geographical maps, statistical tables and charts or diagrams has been addressed in the seminal works of Jacques Bertin and Edward Tufte (Bertin, 1999 ; Tufte, 1990 & Tufte, 2001), little has been written on the graphic design of relational data to study social phenomena (Healy and Moody, 2013). The growing popularity of information design (McCandless, 2009) has underlined the cognitive advantages of making information “meaningful, entertaining and beautiful”. Whether and how this could/should be achieved in the social sciences provides a rich and underexplored field of inquiry.

Our workshop aims to create a space of exchange where social scientists, information designers and researchers involved in the theoretical and technical development of DMTs can discuss issues related to the dynamic mapping of networks and other web-based data. We argue that creating a collaborative space to examine DMTs can help social scientists harness the potential of existing tools and understand how results can be woven into rich and robust narratives and research agendas (Venturini, 2012). Our collaborative investigation seeks to underline the strengths and weaknesses of DMTs as well as identify the specific needs of social scientists using these tools. This should be of particular interest for researchers involved in DMT development, providing detailed feedback on the practices, critiques and needs of a user community mobilizing their software. Feedback can inform the refinement of existing tools and models or the development of new functionalities. The seminar should also be relevant for information designers interested in data visualisation approaches developed to explore, demonstrate and communicate complex phenomena. Encouraging interactions between these different communities will have mutually beneficial effects.