Monday, July 10th, 2023, Vienna
Seminarraum SR 2
Network
Structure
The Causes and Effects of Social Phenomena
Networks are complex systems, many of which are formed by mechanisms such as homophily, preferential attachment, reciprocity, and triadic closure. The interplay of these mechanisms often leads to the formation of structures that determine the relevance of nodes in the network. For instance, important nodes are often located in the core, while less important nodes lie at the periphery. Similarly, nodes may be grouped into communities where they share common interests or attributes with others.
The challenge is then to understand how these mechanisms, that shape the structure of networks, affect our daily lives, decisions, omissions, and vice versa. Social, computer, and network scientists have extensively studied how networks form to better understand human behavior, and several societal issues including diseases and their spread, supply chain disruptions, socio-economic segregation, polarization, displacements, and privacy threats in online platforms.
In this (second edition of the) satellite, we aim to open up a wide range of questions, discussions and possible new directions on explaining and solving societal issues using network research. Furthermore, these discussions will allow us to reflect on the challenges and opportunities to mitigate societal issues with (or without) algorithmic interventions.
Call for abstracts
Extended abstracts about published or unpublished research (1 page max for main content, and 1 page extra max for references, an image or a table).
Submissions must be formatted using to the official LATEX template or MS WORD template. The abstract file must be submitted in PDF format and should be no larger than 20MB. Submissions that exceed the 2-page limit will be automatically rejected. The submission should include a title, a list of 5 keywords, and an extended abstract (e.g., problem statement, research gap, methods, results, etc.). The abstract should outline the main theoretical contributions, methods, datasets, results, impact of the work, etc. Submitted abstracts will undergo a double-blind review process, therefore, abstracts must be anonymized: do not include the author(s) names or affiliation(s) in the paper, and do not include funding or other acknowledgments. Submissions that violate these guidelines will be automatically rejected.
Topics relevant to this satellite session include, but are not limited to, the following:
Explaining or characterizing social phenomena with network science
Explaining or characterizing network structure with social science
Challenges and opportunities of network structure in society and algorithms
Mitigation strategies to solve societal / structural issues with network science
We welcome submissions from doctoral students, postdocs, and early-career researchers.
Please send us your abstracts via email to: netstructure.cfa@gmail.com
Important dates (AoE):
Submission deadline: May 19 (closed)
Notifications: May 26 June 2 (sent via email)
Accepted abstracts
Oral presentations
A threshold model of information cascades under uncertainty
Teruyoshi Kobayashi*A scalable gravity-based method for identifying critical road segments in global cities threatened by climate change
Sukankana Chakraborty, Yiyi He, Jun Rentschler, Paolo Avner, Neave O’Clery*Investigating polarization: cognitive and algorithmic biases and external effects on opinion formation
Valentina Pansanella*, Alina Sirbu, Janos Kertesz and Giulio RossettiThe impact of social interactions on the pace of discovery of new music in online platforms
Gabriele Di Bona*, Enrico Ubaldi, Iacopo Iacopini, Bernardo Monechi, Vito Latora, Vittorio Loreto
Poster presentations
Identifying influential nodes by leveraging redundant ties
Bitao Dai*, Shuo Qin, Suoyi Tan, Chuchu Liu, Jianhong Mou, Hongzhong Deng, Fredrik Liljeros, Xin LuThe voices of rape: Cognitive networks reconstruct emotional perspectives of rape survivors through their passive and active narratives on Reddit
Katherine Abramski*, Salvatore Citraro, Giulio Rossetti, Massimo StellaLinguistic temperature on networks
Tobias Galla*, Christopher Kitching, Henri Kauhanen, Deepthi Gopal, Ricardo Bermúdez-OteroLinking social preferences to social network structure and function
Josefine Bohr Brask*, Andreas Koher, Darren P. Croft, Sune Lehmann (withdrawn)
* presenting
Keynotes
Contributed talks
Poster presentations
Agenda
Monday, July 10th - Room SR2
09:00 - 09:05 Welcome
09:05 - 09:50 Matthias Raddant [ keynote ]
Interorganizational networks: how they influence profitability and diversity
Firms are linked through a variety of formal and informal ties that all together constitute interorganizational networks. As we will show, these ties have consequences for the behavior of the firm and its performance. They also influence the propagation of practices or norms, like the stance on diversity. Besides the empirical analysis of these networks, these ties can be modeled and networks can be simulated to gain insights into measures to improve corporate governance.
09:50 - 10:10 Gabriele Di Bona
The impact of social interactions on the pace of discovery of new music in online platforms
Understanding the balance between exploration and exploitation is crucial in our interconnected society. Recent models focus on individual discovery, but lack understanding of social interactions. This study analyzes music artist exploration using a dataset from Last.fm, revealing heterogeneous discovery rates and assortative clustering. A multi-agent data-driven model captures these findings, highlighting the role of social networks in collective exploration.
10:10 - 10:30 Neave O’Clery
A scalable gravity-based method for identifying critical road segments in global cities threatened by climate change
Rapid urbanization in developing countries necessitates effective urban planning to address challenges like congestion and inadequate infrastructure. This study focuses on improving employment accessibility by estimating the importance of road segments in home-to-work commuting flows. A scalable algorithm based on spatial interaction models and sampling techniques is proposed, which successfully identifies critical road segments in developing cities.
10:30 - 10:45 Coffee Break
11:00 - 11:45 Gerardo Iñiguez [ keynote ]
Universal patterns in egocentric communication networks
Social networked phenomena, like group formation and communication, are hard to quantify and predict. Here I’ll show that simple mathematical models, despite lacking details, are good at uncovering underlying mechanisms and how their dynamical interplay leads to segregation and heterogeneity in social networks. I’ll focus on the universalities these models reveal and what they tell us about human behaviour off- and online.
11:45 - 12:05 Teruyoshi Kobayashi
A threshold model of information cascades under uncertainty
Social contagion, the spread of information and behaviors through social ties, is often modeled using threshold models. However, real-world contexts involve uncertainty in observing others' states. This paper introduces a threshold model with uncertainty, where using Bayesian inference we estimate the probability distribution of active neighbors for each individual in a network. The dynamics become non-monotonic, allowing for self-fulfilling events and explaining the spread of misinformation. Analytical cascade conditions are provided.
12:05 - 12:25 Valentina Pansanella
Investigating polarization: cognitive and algorithmic biases and external effects on opinion formation
The process of opinion formation in online social networks is influenced by biases, algorithmic filtering, and social and mass media. Models show that algorithmic bias increases fragmentation, while open-mindedness helps maintain moderate clusters. A methodology is introduced to estimate open-mindedness from online discussions, and a modified model replicates real-world opinion evolution.
12:25 - 13:10 Ágnes Horvát [ keynote ]
The contingent benefits of networks for collective intelligence
This talk explains contradictions in previous research about the accuracy of decision-making in networks that emerge from different forms of social exchange. We use experiments, simulations, and analyses of large-scale observational data to determine whether collective intelligence arises in centralized influence networks from open discussion, decentralized networks from controlled information exchange, and herding networks from social signaling.
13:10 - 13:15 Closing remarks
13:00 - 14:00 Lunch break + Poster session
15:30 - 16:00 Coffee break + Poster session
Poster Session (Monday, July 10 - Tuesday, July 11)
Posters will be displayed throughout these days, discussions during coffee and lunch breaks.
Poster #20 Katherine Abramski
The voices of rape: Cognitive networks reconstruct emotional perspectives of rape survivors through their passive and active narratives on Reddit
Survivors of sexual assault often feel guilt and self-blame, impacting their mental health. Online platforms like Reddit offer a space for survivors to share their experiences. This study explores linguistic choices in rape narratives, finding that the passive voice is associated with greater psychological distress and feelings of self-blame. The research highlights the connection between language and mental health and provides support for survivors.
Poster #21 Tobias Galla
Linguistic temperature on networks
Language evolution and change are influenced by social and cultural processes. This study focuses on the stability of binary language features using geospatial data and agent-based models. The concept of "linguistic temperature" is introduced to characterize a feature's propensity to change. Results align with phylogenetic approaches, offering insights into the physics of social systems and network dynamics.
Poster #22 Bitao Dai
Identifying influential nodes by leveraging redundant ties
Identifying influential nodes in complex networks is challenging. Global centrality-based methods are accurate but computationally demanding. Redundant ties in cyclic structures impact node importance. This study introduces a new centrality metric, MSTDC, which leverages spanning trees to quantify node importance with linear complexity. Experimental results demonstrate the effectiveness of MSTDC in identifying influential nodes.
PC members
Adriana Manna, CEU
Elsa Andres, CEU
Eric Dignum, UvA & CSH
Felipe Vaca, CEU
Gabriela Juncosa, CEU
Hao Cui, University College Dublin
Jan Bachmann, CEU & CSH
Júlia Számely, CEU
Leonardo Di Gaetano, CEU
Leonardo Rizzo, CEU
Leonie Neuhäuser, RWTH Aachen
Lisette Espín-Noboa, CEU & CSH
Luca Gallo, CEU
Ludovico Napoli, CEU
Manran Zhu, CEU
Martí Medina, CEU
Mauritz Cartier, CSH
Onkar Sadekar, CEU
Rafael Prieto-Curiel, CSH
Sina Sajjadi, CEU & CSH
Timur Naushirvanov, CEU
Yijing Chen, CEU