At the Infolab we do research on all aspects of social data science: we define mathematical models to represent contemporary social data, so that it can be processed by computers, we develop computational analysis methods, we implement our models and methods in publicly available software, and we apply them to perform empirical social data analyses. These applications allow us to test our theoretical advances and provide requirements for new developments.

Research process at the Infolab

Research areas

Models and methods for feature-rich networks

A temporal text network
A communication network, with actors (A) exchanging messages (M), with timestamps (Vega and Magnani, 2020).

Networks have been a popular model to study social systems for longer than one century, and their applicability has increased in the last decades thanks to increased accessibility to social data, for example from online social networks. However, contemporary social data can very heterogeneous, consisting of different types of actors and social ties, multimedia content, and temporal information. In addition, relevant social connections are often not directly available in the data, which implies that we cannot always be certain about which connections exist. Therefore, at the Infolab we study extended network models (also known as feature-rich) that can be used to represent the complexity of contemporary social data.

One model we have extensively studied since 20111 are multilayer networks. In addition to original research articles, we have also produced material to organise some of the vast knowledge in this area: a book on multilayer social networks, software libraries2 for multilayer network analysis, and survey articles on preprocessing3, layer comparison4, community detection5, and diffusion processes6. We have also introduced multilayer network models including textual and temporal information, called temporal text networks, and studied measures and algorithms for probabilistic networks where edge existence is uncertain.

Empirical study of online information networks

A world map showing the relative frequency of YouTube videos including a given visual theme
A world map showing the relative frequency of top-relevant YouTube videos including a given visual theme in different geographical regions, identified using deep neural network classifiers (Magnani and Segerberg, 2021).

Our main application area is the study of online communication. Two current areas of applied research are online disinformation and visual political communication.

Most of the existing computational work on online disinformation focuses on a single platform. However, we can expect organised disinformation campaigns and other misinformation spreading processes to happen across platforms. The importance of the content that is spread and the existence of interconnected platforms make online disinformation a timely and appropriate application for our work on feature-rich networks. Some aspects of our interest are: (1) the detection of coordinated behaviors, that is, groups of users or groups of social media accounts collaborating to boost spreading, (2) the role of change of platform in the dynamics of spreading, e.g., to what extent moving from one platform to another (and possibly back) allows reaching new audiences, and (3) the study of how cultural and political conditions of the Nordic countries relate to the observed disinformation processes if compared with other more studied context, e.g., the US. This research is done as part of the Nordic Observatory for Social Media and Information Disorder, of which we are the Swedish academic node.

Visual content, such as images, videos, and memes, plays a particularly important role in online communication: it can be persuasive, emotive, and affective, and is widely shared in online environments. For example, visual content plays an important symbolic, emotional and (dis)connecting role in political movements, where it becomes the focus of social and algorithmic negotiation. Part of our research addresses the methodological and empirical challenges related to the study of how visual content propagates online and how it becomes a beacon of political aggregation and polarisation.

Sponsors

The Infolab is or has been funded by the following projects:

International research projects and funded collaborations

VISUAL PERSUASION IN A TRANSFORMING EUROPE: the affective and polarising power of visual content in online political discourse (PolarVis)
Funding from: Chanse programme (FORTE)
Period: 2022-2025
In short: PolarVis examines the role of digital visual content as a beacon of belonging and polarisation in contemporary political life.

NORdic observatory for digital media and information DISorder (NORDIS)
Funding from: EU CEF Telecom
Period: 2021-2023
In short: The Nordic Hub of the European Digital Media Observatory (EDMO), consisting of four academic nodes and four fact-checking organisations.

Online disinformation: an integrated view
Funding from: NOS-HS
Period: 2019-2020
In short: Resources to organise a series of Nordic workshop on online disinformation and establish an interdisciplinary network of researchers.

VIRT-EU: Values and Ethics in Innovation for Responsible Technology in Europe
Funding from: The European Union’s Horizon 2020 research and innovation programme
Period: 2017-2019
In short: A project where we developed a new version of our analysis software and applied it to the study of the online IoT community.

Initiation grant
Funding from: STINT
Period: 2017-2018
In short: Seed funding for the establishment of a collaboration with TokyoTech.

National research projects and funded collaborations

eSSENCE
Funding from: Swedish Government
Period: 2022-
In short: A Swedish strategic collaborative research programme in e-Science

Visual Politics: A Deep Learning Approach to the Spread and Stick of Political Ideas
Funding from: Swedish Research Council (VR)
Period: 2022-2025
In short: A project on methodological and empirical advances in the analysis of online visual political communication.

Funding from Uppsala University

AI4Research Sabbatical
Funding from: AI4Research
Period: 2024
In short: Internal sabbatical (50%) at the AI4Research center to write a textbook on social data mining.

Digital politics (Research Network)
Funding from: Centre for Integrated Research on Culture and Society (CIRCUS)
Period: 2022-2023
In short: Seed funding for the establishment of a cross-Faculty research network.

Self-referential perception at the macro level
Funding from: Centre for Integrated Research on Culture and Society (CIRCUS)
Period: 2021
In short: Seed funding to prepare larger project proposals involving the analysis of longitudinal population-level social networks from registry data.

AI and political communication
Funding from: AI4Research
Period: 2020-2021
In short: Internal sabbatical (50%) at the AI4Research center on using deep neural networks to study online visual political communication.

Travel grants

Research stay
Funding from: AccelNet-MultiNet
Period: 2025
In short: Research stay at Indiana University (recipient: Georgios Panayiotou).

Research stay
Funding from: Karl Staaff Foundation
Period: 2025
In short: Research stay at New York University (recipient: Matias Piqueras).

Research stay
Funding from: SoBigData
Period: 2024
In short: University of Pisa (recipient: Georgios Panayiotou).

Research visit
Funding from: Institut français de Suède
Period: 2023
In short: University of Bordeaux (recipients: Matteo Magnani, Georgios Panayiotou).

We also thank Uppsala University for providing funding to create the International Master’s programme in Data Science (started in the Fall 2020) and our interdisciplinary introductory course on computational social science (first taught in the Spring 2022), and the Division of Computing Science at the Department of Information Technology for continuous support.

  1. Matteo Magnani and Luca Rossi (2011). The ML-Model for Multi-Layer Social Networks. International conference on social network analysis and mining (ASONAM). IEEE. 

  2. Matteo Magnani, Luca Rossi and Davide Vega. Multiplex network analysis with R. Journal of Statistical Software. 

  3. Roberto Interdonato, Matteo Magnani, Diego Perna, Andrea Tagarelli and Davide Vega. Multilayer network simplification: approaches, models and methods. Computer Science Review, Elsevier. 

  4. Piotr Brodka, Anna Chmiel, Matteo Magnani and Giancarlo Ragozini (2018). Quantifying layer similarity in multiplex networks: a systematic study. Royal Society Open Science, 5(8). 

  5. Obaida Hanteer, Roberto Interdonato, Matteo Magnani, Luca Rossi and Andrea Tagarelli. Community Detection in Multiplex Networks, and Cécile Bothorel, Juan David Cruz, Matteo Magnani, and Barbora Micenkova (2015). Clustering Attributed Graphs: Models, Measures and Methods. Network Science 3 (3). Cambridge University Press: 408–44. 

  6. Mostafa Salehi, Rajesh Sharma, Moreno Marzolla, Matteo Magnani, Payam Siyari, and Danilo Montesi (2015). Spreading Processes in Multilayer Networks. IEEE Transactions on Network Science and Engineering 2 (2): 65–83.