Matteo Magnani
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Matteo Magnani has a total of 108 co-authors. Here are the top co-authors by number of shared publications:
- Danilo Montesi
Department of Computer Science and Engineering, University of Bologna, Mura Anteo Zamboni 7, 40126 Bologna, Italy
31 shared publications - Luca Rossi
IT University of Copenhagen, Copenhagen, Denmark
21 shared publications Mark E. Dickison
10 shared publications- Davide Vega
Uppsala University, Uppsala, Sweden
8 shared publications Luca Rossi
University of Copenhagen
6 shared publicationsAntonio Messina
University of Bologna
3 shared publications
Latest Publications
- Open Access
- 17 September 2026
- Published by Springer Nature in Data Mining and Knowledge Discovery
- Vol. 40 (6) , 92
MOUFLON: multi-group modularity-based fairness-aware community detection
In this paper, we propose MOUFLON, a fairness-aware, modularity-based community detection method that allows adjusting the importance of partition quality over fairness outcomes. MOUFLON uses a novel proportional balance fairness metric, providing consistent and comparable fairness scores across multi-group and imbalanced network settings. We evaluate our method under both synthetic and real-world network datasets, focusing primarily on performance and the trade-off between modularity and fairness in the resulting communities, along with the impact of network characteristics such as size, density, and group distribution. As structural biases can lead to strong alignment between demographic groups and network structure, we also examine scenarios with highly clustered homogeneous groups, to understand how such structures influence fairness outcomes. Finally, we benchmark our method against existing fairness-aware community detection approaches, and apply MOUFLON to register-derived population networks from two Swedish municipalities, demonstrating that proportionally balanced communities can be recovered at a small cost in modularity in real-world settings. Our findings showcase the effects of incorporating fairness constraints into modularity-based community detection, and highlight key considerations for designing and benchmarking fairness-aware social network analysis methods.
- 17 September 2026
- Published by Springer Nature in Data Mining and Knowledge Discovery
- Vol. 40 (6) , 92
Open AccessMOUFLON: multi-group modularity-based fairness-aware community detection
In this paper, we propose MOUFLON, a fairness-aware, modularity-based community detection method that allows adjusting the importance of partition quality over fairness outcomes. MOUFLON uses a novel proportional balance fairness metric, providing consistent and comparable fairness scores across multi-group and imbalanced network settings. We evaluate our method under both synthetic and real-world network datasets, focusing primarily on performance and the trade-off between modularity and fairness in the resulting communities, along with the impact of network characteristics such as size, density, and group distribution. As structural biases can lead to strong alignment between demographic groups and network structure, we also examine scenarios with highly clustered homogeneous groups, to understand how such structures influence fairness outcomes. Finally, we benchmark our method against existing fairness-aware community detection approaches, and apply MOUFLON to register-derived population networks from two Swedish municipalities, demonstrating that proportionally balanced communities can be recovered at a small cost in modularity in real-world settings. Our findings showcase the effects of incorporating fairness constraints into modularity-based community detection, and highlight key considerations for designing and benchmarking fairness-aware social network analysis methods.
- 8 August 2026
- Published by Springer Nature
- p. 1-17
Path-Based and Whole-Network Measures
- 16 June 2026
- Published by Edward Elgar Publishing
- p. 191-214
Network science for social communication studies
- 16 June 2026
- Published by Edward Elgar Publishing
- p. 191-214
Network science for social communication studies
- 18 April 2026
- Published by Springer Nature
- p. 340-353
Fair Multilayer Community Detection: A Research Agenda
- 18 April 2026
- Published by Springer Nature
- p. 340-353
Fair Multilayer Community Detection: A Research Agenda
- Open Access
- 17 January 2026
- Published by Elsevier in Information Sciences
- Vol. 737
Multimodal coordinated online behavior: Trade-offs and strategies
- 17 January 2026
- Published by Elsevier in Information Sciences
- Vol. 737
Open AccessMultimodal coordinated online behavior: Trade-offs and strategies
- Open Access
- 19 September 2025
- Published by SAGE Publications in Social Science Computer Review
- Vol. 44 (3) , 572-591
Leveraging VLLMs for Visual Clustering: Image-to-Text Mapping Shows Increased Semantic Capabilities and Interpretability
As visual content becomes increasingly prominent on social media, automated image categorization is vital for computational social science efforts to identify emerging visual themes and narratives in online debates. However, the methods based on convolutional neural networks (CNNs) currently used in the field are unable to fully capture the connotative meaning of images, and struggle to produce easily interpretable clusters. In response to these challenges, we test an approach that leverages the ability of Vision-and-Large-Language-Models (VLLMs) to generate image descriptions that incorporate connotative interpretations of the input images. In particular, we use a VLLM to generate connotative textual descriptions of a set of images related to climate debate, and cluster the images based on these textual descriptions. In parallel, we cluster the same images using a more traditional approach based on CNNs. In doing so, we compare the connotative semantic validity of clusters generated using VLLMs with those produced using CNNs, and assess their interpretability. The results show that the approach based on VLLMs greatly improves the quality score for connotative clustering. Moreover, VLLM-based approaches, leveraging textual information as a step towards clustering, offer a high level of interpretability of the results.
- 19 September 2025
- Published by SAGE Publications in Social Science Computer Review
- Vol. 44 (3) , 572-591
Open AccessLeveraging VLLMs for Visual Clustering: Image-to-Text Mapping Shows Increased Semantic Capabilities and Interpretability
As visual content becomes increasingly prominent on social media, automated image categorization is vital for computational social science efforts to identify emerging visual themes and narratives in online debates. However, the methods based on convolutional neural networks (CNNs) currently used in the field are unable to fully capture the connotative meaning of images, and struggle to produce easily interpretable clusters. In response to these challenges, we test an approach that leverages the ability of Vision-and-Large-Language-Models (VLLMs) to generate image descriptions that incorporate connotative interpretations of the input images. In particular, we use a VLLM to generate connotative textual descriptions of a set of images related to climate debate, and cluster the images based on these textual descriptions. In parallel, we cluster the same images using a more traditional approach based on CNNs. In doing so, we compare the connotative semantic validity of clusters generated using VLLMs with those produced using CNNs, and assess their interpretability. The results show that the approach based on VLLMs greatly improves the quality score for connotative clustering. Moreover, VLLM-based approaches, leveraging textual information as a step towards clustering, offer a high level of interpretability of the results.
- Open Access
- 23 April 2025
- Published by Center for Open Science in SocArXiv
Anatomy of a Swedish population-scale network
With the increasing interest in large-scale social network analysis, recent research has expanded into nation-wide networks generated from administrative data. We construct a multilayer population-scale social network for Sweden using public register data from 2000 to 2017, covering approximately 8.3 million individuals aged 15 and older. We analyze the structure and connectivity patterns in the network across six layers: close family, extended family, household, school, neighbors, and work. We compare our findings to a similar study of the Netherlands, revealing similar degree distributions and small-world characteristics, but also discrepancies in the embeddedness and closure distributions, mainly driven by differences in size and population density.
- 23 April 2025
- Published by Center for Open Science in SocArXiv
Open AccessAnatomy of a Swedish population-scale network
With the increasing interest in large-scale social network analysis, recent research has expanded into nation-wide networks generated from administrative data. We construct a multilayer population-scale social network for Sweden using public register data from 2000 to 2017, covering approximately 8.3 million individuals aged 15 and older. We analyze the structure and connectivity patterns in the network across six layers: close family, extended family, household, school, neighbors, and work. We compare our findings to a similar study of the Netherlands, revealing similar degree distributions and small-world characteristics, but also discrepancies in the embeddedness and closure distributions, mainly driven by differences in size and population density.
- Open Access
- 8 April 2025
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 20 (4) , e0318338
Visual digital intermediaries and global climate communication: Is climate change still a distant problem on YouTube?
This article addresses the role of digital intermediaries in visual climate communication, and specifically their contribution to the persistence of a ‘green ghetto’ of traditional communicators and repertoires online. We argue for a comparative sensibility: global platforms convey global issues to global audiences, yet the same platform may distribute conditions of visibility for compelling communication unevenly around the world. The study analyses how a major global visual platform, YouTube (Search), articulates climate change in 232 countries in their official languages. It combines API research, channel coding and computational image analysis to assess the processing and presentation of top-ranked results with respect to their diversity and proximity to local context. The findings show that YouTube Search establishes visibility winners who typically sustain the classic visual repertoire of climate change as a distant problem, and that Global North sources dominate irrespective of region. However, there are notable exceptions to these patterns.
- 8 April 2025
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 20 (4) , e0318338
Open AccessVisual digital intermediaries and global climate communication: Is climate change still a distant problem on YouTube?
This article addresses the role of digital intermediaries in visual climate communication, and specifically their contribution to the persistence of a ‘green ghetto’ of traditional communicators and repertoires online. We argue for a comparative sensibility: global platforms convey global issues to global audiences, yet the same platform may distribute conditions of visibility for compelling communication unevenly around the world. The study analyses how a major global visual platform, YouTube (Search), articulates climate change in 232 countries in their official languages. It combines API research, channel coding and computational image analysis to assess the processing and presentation of top-ranked results with respect to their diversity and proximity to local context. The findings show that YouTube Search establishes visibility winners who typically sustain the classic visual repertoire of climate change as a distant problem, and that Global North sources dominate irrespective of region. However, there are notable exceptions to these patterns.
- Open Access
- 11 September 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Visualization and Computer Graphics
- Vol. 31 (1) , 1028-1038
Improved Visual Saliency of Graph Clusters with Orderable Node-Link Layouts
Graphs are often used to model relationships between entities. The identification and visualization of clusters in graphs enable insight discovery in many application areas, such as life sciences and social sciences. Force-directed graph layouts promote the visual saliency of clusters, as they bring adjacent nodes closer together, and push non-adjacent nodes apart. At the same time, matrices can effectively show clusters when a suitable row/column ordering is applied, but are less appealing to untrained users not providing an intuitive node-link metaphor. It is thus worth exploring layouts combining the strengths of the node-link metaphor and node ordering. In this work, we study the impact of node ordering on the visual saliency of clusters in orderable node-link diagrams, namely radial diagrams, arc diagrams and symmetric arc diagrams. Through a crowdsourced controlled experiment, we show that users can count clusters consistently more accurately, and to a large extent faster, with orderable node-link diagrams than with three state-of-the art force-directed layout algorithms, i.e., ‘Linlog’, ‘Backbone’ and ‘sfdp’. The measured advantage is greater in case of low cluster separability and/or low compactness. A free copy of this paper and all supplemental materials are available at https://osf.io/kc3dg/.
- 11 September 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Visualization and Computer Graphics
- Vol. 31 (1) , 1028-1038
Open AccessImproved Visual Saliency of Graph Clusters with Orderable Node-Link Layouts
Graphs are often used to model relationships between entities. The identification and visualization of clusters in graphs enable insight discovery in many application areas, such as life sciences and social sciences. Force-directed graph layouts promote the visual saliency of clusters, as they bring adjacent nodes closer together, and push non-adjacent nodes apart. At the same time, matrices can effectively show clusters when a suitable row/column ordering is applied, but are less appealing to untrained users not providing an intuitive node-link metaphor. It is thus worth exploring layouts combining the strengths of the node-link metaphor and node ordering. In this work, we study the impact of node ordering on the visual saliency of clusters in orderable node-link diagrams, namely radial diagrams, arc diagrams and symmetric arc diagrams. Through a crowdsourced controlled experiment, we show that users can count clusters consistently more accurately, and to a large extent faster, with orderable node-link diagrams than with three state-of-the art force-directed layout algorithms, i.e., ‘Linlog’, ‘Backbone’ and ‘sfdp’. The measured advantage is greater in case of low cluster separability and/or low compactness. A free copy of this paper and all supplemental materials are available at https://osf.io/kc3dg/.