Knot. Interfacce per l'esplorazione di reti sociali per la ricerca storica

My research aims at better understanding the role information visualization and interaction design in the scholarly activity of humanities, proposing a case study specifically focused on social network visualization within the project Mapping the Republic of Letters. In recent years humanists have become increasingly involved in what is often referred as the “visual-computational turn” in scholarship. The adoption of digital tools and visualization became fundamental in modern humanities studies but, while natural and social sciences have a long tradition with these technologies, these disciplines are still facing a context where well-established research methods and practices are not well integrated. The main reason behind this situation is due mainly to the lack of digital tools designed specifically for humanities studies. This induce scholars to use existing tools build for other disciplines like Gephi, Tableau etc. designed to produce graphs, charts, diagrams, and other visual artefacts often associated with data visualization, the expression of quantifiable or quantitative information in graphic form. While these visualizations and interactive processes tend to work with data sets such as social networks, digitized corpora, and demographic data, the same tools can’t completely succeed in the analysis and observation of incomplete and fragmented data as the one of the project Mapping the Republic of Letters. The visual outputs of these tools illustrate data through process of aggregation and distillation and the interaction process underlying the production of these images follow rules than can be synthesized most of the times in the famous Ben Shneiderman formula “Overview first, zoom and filter, then details on demand”. In this way the user plays the role of consumer and not co-author, because the design of the software sets pre existing conditions and the control of the tool is just an illusion. This logic is evident in tools like Gephi, where the interface relegate the user in a position where he can simply choose between a pre set series of features that analyse and modify the visualization of the network, starting from the a priori assumption that the data are complete and the image created is certain. What is dangerous here is the appropriation of a methodology designed for some specific purpose, that most of times lose a series of philosophical and theoretical caveats during the translation into another disciplines. We should always keep in mind that the purpose of historians is very different from that of the computer scientist who designed the algorithm and interactions behind Gephi and similar softwares. However I’m not saying that the design of the new tool, Knot, should completely abandon all the logics behind social network analysis and visualization, but consider all the limits and think about a new approach that combine the quali-quantitative side with a more humanist approach. The project aims at building a research environment where the user plays an active role in the process of discovering the multidimensional and heterogeneous data, helping him discovering, and create, where necessary, explicit and not explicit relationships between the intellectuals and the people present in the data of MRofL. Instead of a tool to create social network visualizations and perform network analysis, we want to create an environment where the user can work with the network to create new epistemological processes. To do so I divide the design of the tool in 5 macro areas of interest that I consider important, comparing the limits of the existing tools. The entry point in the existing tools (Gephi) is the entire network made out of all the nodes and edges present in the dataset. In Knot the construction of the network is radically different. The user can decide where to start building the network. It can start from a single person, a group of people related to each other by some attributes (e.g.: born in the same city) or a group of people arbitrarily chosen by the user. From these nodes the exploration and construction of the network begins. Each node can be selected and the user can expand the nodes explicitly linked to them (explicit relations in MRofL are based on correspondence, family and meetings). The user can decide how many nodes to expand, which kind of relationship and the degree (only first degree, second degree, etc.) and also find the shortest path in our database between two given nodes. The visualization in Knot is conceived as a tool to help the scholars raise new research questions and to explore explicit and not explicit networks of the MRofL database. The final output is not necessarily an image addressed to the public, but a visual artefact of the research process of the scholar. The user has a huge freedom on all the visual properties of the graph network and decides how and when to use colours, edges and different spatial layouts. In this way the scholar has the possibility to exploit and work on the narrative aspects provided by the data, using a visual language that is not completely pre designed, avoiding in this way bias derived from the uncertain nature of the data. The heterogeneous and multidimensional data of MRofL, presents many different possibilities for exploring not explicit relationships. We can describe these relationships as a broad spectrum of possible connectedness that can be very vague (born in the same city) or more evident (went to the same school in the same year). The interaction should allow the user to quickly explore the nature of the networks through non hierarchical queries and set different geo-spatial contexts for the research. The exploration is supported by different types of selection (multiple selection, common nodes, nodes having a certain attribute, etc) transforming the interaction and manipulation of the network much easier than in the existing tools. The interaction and exploration of the network allow the user to not only uncover and visualize the possible relationships but also to facilitate a process of categorization based on user-defined and research-based attributes. This process is made possible through the direct manipulation of the network and and of the visualization. The user can add and edit nodes and edges when he/she discovers errors or gaps in the data of the database. Once the user is sure about the data editing and enrichment, he can also decide to share and publish his work for other scholars making the collaborative work easier. These kind of processes in the existing tools is quite elaborated to do and force the user to transform and re-upload the dataset, losing the important visual aid of the network graph. Another important aspect of Knot is the direct link with Athanasius that allows the user revealing and understanding the nature of the data and the sources. Every moment during the work the scholar can decide which sources to work with and trace the origin of the information used to create the visualization.

Negli ultimi anni i processi di ricerca nelle discipline umanistiche sono state coinvolte in quella che viene definita come “visual-computational turn”. L’adozione di strumenti digitali e della visualizzazione nei processi epistemologici è diventato un fatto fondamentale nei moderni progetti di ricerca umanistica ma, al contrario di altre discipline che hanno una lunga tradizione con il mezzo digitale, le discipline umanistiche sono ancora in una fase di definizione di modelli e pratiche da adottare. Il motivo principale di questa situazione è dovuta principalmente alla mancanza di strumenti digitali progettati specificatamente per gli studi umanistici che ha costretto i ricercatori ad utilizzare strumenti progettati per altre tipologie di ricerca, orientate alla visualizzazione di grafici, diagrammi e grafi con un approccio tipico dell’information visualization. Poichè i dati provenienti dalle ricerche umanistiche tradizionali spesso provengono da risorse lontane nel tempo o da un’operazione di estrapolazione e interpretazione di un fenomeno, questi strumenti non sempre danno risposte convincenti ai ricercatori. Gli artefatti visivi prodotti da questi strumenti sono il risultato di processi di aggregazione e riduzione, in cui l’utente ha un controllo limitato. La tesi ha l’obiettivo di meglio comprendere il ruolo che il design ha nella definizione dei processi epistemologici nelle digital humanities, proponendo un caso specifico orientato alla progettazione di un’interfaccia per l’analisi delle reti sociali storiche all’interno del progetto “Mapping the Republic of Letters”. Il progetto ha lo scopo di fornire un approccio innovativo alla visualizzazione di reti sociali, attraverso uno strumento che coniuga l’accesso ai dati con un processo più adatto ai contesti umanistici. La visualizzazione non è concepita come il prodotto finale della ricerca ma come uno strumento per permettere un analisi interpretative delle informazioni.