arXiv subjects

Markus Strohmaier

Publications and source records attributed to Markus Strohmaier.

At least 37 records · Page 2Linked to original sources

arXiv2024

Recommendation Fairness in Social Networks Over Time

In social recommender systems, it is crucial that the recommendation models provide equitable visibility for different demographic groups, such as gender or race. Most existing research has addressed this problem by only studying individual static snapshots of networks that typically change over time. To address this gap, we study the evolution of recommendation fairness over time and its relation to dynamic network properties. We examine three real-world dynamic networks by evaluating the fairness of six recommendation algorithms and analyzing the association between fairness and network properties over time. We further study how interventions on network properties influence fairness by examining counterfactual scenarios with alternative evolution outcomes and differing network properties. Our results on empirical datasets suggest that recommendation fairness improves over time, regardless of the recommendation method. We also find that two network properties, minority ratio, and homophily ratio, exhibit stable correlations with fairness over time. Our counterfactual study further suggests that an extreme homophily ratio potentially contributes to unfair recommendations even with a balanced minority ratio. Our work provides insights into the evolution of fairness within dynamic networks in social science. We believe that our findings will help system operators and policymakers to better comprehend the implications of temporal changes and interventions targeting fairness in social networks.

cs.SI↗

arXiv2023

The Governance of Decentralized Autonomous Organizations: A Study of Contributors' Influence, Networks, and Shifts in Voting Power

We present a study analyzing the voting behavior of contributors, or vested users, in Decentralized Autonomous Organizations (DAOs). We evaluate their involvement in decision-making processes, discovering that in at least 7.54% of all DAOs, contributors, on average, held the necessary majority to control governance decisions. Furthermore, contributors have singularly decided at least one proposal in 20.41% of DAOs. Notably, contributors tend to be centrally positioned within the DAO governance ecosystem, suggesting the presence of inner power circles. Additionally, we observed a tendency for shifts in governance token ownership shortly before governance polls take place in 1202 (14.81%) of 8116 evaluated proposals. Our findings highlight the central role of contributors across a spectrum of DAOs, including Decentralized Finance protocols. Our research also offers important empirical insights pertinent to ongoing regulatory activities aimed at increasing transparency to DAO governance frameworks.

cs.SI↗

arXiv2023

Similarity of Neural Network Models: A Survey of Functional and Representational Measures

Measuring similarity of neural networks to understand and improve their behavior has become an issue of great importance and research interest. In this survey, we provide a comprehensive overview of two complementary perspectives of measuring neural network similarity: (i) representational similarity, which considers how activations of intermediate layers differ, and (ii) functional similarity, which considers how models differ in their outputs. In addition to providing detailed descriptions of existing measures, we summarize and discuss results on the properties of and relationships between these measures, and point to open research problems. We hope our work lays a foundation for more systematic research on the properties and applicability of similarity measures for neural network models.

cs.LG↗

arXiv2023

Toxic comments reduce the activity of volunteer editors on Wikipedia

Wikipedia is one of the most successful collaborative projects in history. It is the largest encyclopedia ever created, with millions of users worldwide relying on it as the first source of information as well as for fact-checking and in-depth research. As Wikipedia relies solely on the efforts of its volunteer-editors, its success might be particularly affected by toxic speech. In this paper, we analyze all 57 million comments made on user talk pages of 8.5 million editors across the six most active language editions of Wikipedia to study the potential impact of toxicity on editors' behaviour. We find that toxic comments consistently reduce the activity of editors, leading to an estimated loss of 0.5-2 active days per user in the short term. This amounts to multiple human-years of lost productivity when considering the number of active contributors to Wikipedia. The effects of toxic comments are even greater in the long term, as they significantly increase the risk of editors leaving the project altogether. Using an agent-based model, we demonstrate that toxicity attacks on Wikipedia have the potential to impede the progress of the entire project. Our results underscore the importance of mitigating toxic speech on collaborative platforms such as Wikipedia to ensure their continued success.

cs.CY↗

arXiv2023

SensePOLAR: Word sense aware interpretability for pre-trained contextual word embeddings

Adding interpretability to word embeddings represents an area of active research in text representation. Recent work has explored thepotential of embedding words via so-called polar dimensions (e.g. good vs. bad, correct vs. wrong). Examples of such recent approaches include SemAxis, POLAR, FrameAxis, and BiImp. Although these approaches provide interpretable dimensions for words, they have not been designed to deal with polysemy, i.e. they can not easily distinguish between different senses of words. To address this limitation, we present SensePOLAR, an extension of the original POLAR framework that enables word-sense aware interpretability for pre-trained contextual word embeddings. The resulting interpretable word embeddings achieve a level of performance that is comparable to original contextual word embeddings across a variety of natural language processing tasks including the GLUE and SQuAD benchmarks. Our work removes a fundamental limitation of existing approaches by offering users sense aware interpretations for contextual word embeddings.

cs.CL↗

arXiv2022

Properties of Group Fairness Metrics for Rankings

In recent years, several metrics have been developed for evaluating group fairness of rankings. Given that these metrics were developed with different application contexts and ranking algorithms in mind, it is not straightforward which metric to choose for a given scenario. In this paper, we perform a comprehensive comparative analysis of existing group fairness metrics developed in the context of fair ranking. By virtue of their diverse application contexts, we argue that such a comparative analysis is not straightforward. Hence, we take an axiomatic approach whereby we design a set of thirteen properties for group fairness metrics that consider different ranking settings. A metric can then be selected depending on whether it satisfies all or a subset of these properties. We apply these properties on eleven existing group fairness metrics, and through both empirical and theoretical results we demonstrate that most of these metrics only satisfy a small subset of the proposed properties. These findings highlight limitations of existing metrics, and provide insights into how to evaluate and interpret different fairness metrics in practical deployment. The proposed properties can also assist practitioners in selecting appropriate metrics for evaluating fairness in a specific application.

cs.LG↗

arXiv2022

Neighborhood Structure Configuration Models

We develop a new method to efficiently sample synthetic networks that preserve the d-hop neighborhood structure of a given network for any given d. The proposed algorithm trades off the diversity in network samples against the depth of the neighborhood structure that is preserved. Our key innovation is to employ a colored Configuration Model with colors derived from iterations of the so-called Color Refinement algorithm. We prove that with increasing iterations the preserved structural information increases: the generated synthetic networks and the original network become more and more similar, and are eventually indistinguishable in terms of centrality measures such as PageRank, HITS, Katz centrality and eigenvector centrality. Our work enables to efficiently generate samples with a precisely controlled similarity to the original network, especially for large networks.

cs.SI↗

arXiv2022

Adversarial Inter-Group Link Injection Degrades the Fairness of Graph Neural Networks

We present evidence for the existence and effectiveness of adversarial attacks on graph neural networks (GNNs) that aim to degrade fairness. These attacks can disadvantage a particular subgroup of nodes in GNN-based node classification, where nodes of the underlying network have sensitive attributes, such as race or gender. We conduct qualitative and experimental analyses explaining how adversarial link injection impairs the fairness of GNN predictions. For example, an attacker can compromise the fairness of GNN-based node classification by injecting adversarial links between nodes belonging to opposite subgroups and opposite class labels. Our experiments on empirical datasets demonstrate that adversarial fairness attacks can significantly degrade the fairness of GNN predictions (attacks are effective) with a low perturbation rate (attacks are efficient) and without a significant drop in accuracy (attacks are deceptive). This work demonstrates the vulnerability of GNN models to adversarial fairness attacks. We hope our findings raise awareness about this issue in our community and lay a foundation for the future development of GNN models that are more robust to such attacks.

cs.LG↗

arXiv2022

Improving the visibility of minorities through network growth interventions

Improving the position of minorities in networks via interventions is a challenge of high theoretical and societal importance. In this work, we examine how different network growth interventions impact the position of minority nodes in degree rankings over time. We distinguish between two kinds of interventions: (i) group size interventions, such as introducing quotas, that regulate the ratio of incoming minority and majority nodes; and (ii) behavioural interventions, such as homophily, i.e. varying how groups interact and connect to each other. We find that even extreme group size interventions do not have a strong effect on the position of minorities in rankings if certain behavioural changes do not manifest at the same time. For example, minority representation in rankings is not increased by high quotas if the actors in the network do not adopt homophilic behaviour. As a result, a key finding of our research is that in order for the visibility of minorities to improve, group size and behavioural interventions need to be coordinated. Moreover, their potential benefit is highly dependent on pre-intervention conditions in social networks. In a real-world case study, we explore the effectiveness of interventions to reach gender parity in academia. Our work lays a theoretical and computational foundation for further studies aiming to explore the effectiveness of interventions in growing networks.

physics.soc-ph↗

arXiv2022

Minorities in networks and algorithms

In this chapter, we provide an overview of recent advances in data-driven and theory-informed complex models of social networks and their potential in understanding societal inequalities and marginalization. We focus on inequalities arising from networks and network-based algorithms and how they affect minorities. In particular, we examine how homophily and mixing biases shape large and small social networks, influence perception of minorities, and affect collaboration patterns. We also discuss dynamical processes on and of networks and the formation of norms and health inequalities. Additionally, we argue that network modeling is paramount for unveiling the effect of ranking and social recommendation algorithms on the visibility of minorities. Finally, we highlight the key challenges and future opportunities in this emerging research topic.

physics.soc-ph↗

arXiv2022

Combining sensors and surveys to study social contexts: Case of scientific conferences

In this paper, we present a unique collection of four data sets to study social behaviour. The data were collected at four international scientific conferences, during which we measured face-to-face contacts along with additional information about individuals. Building on innovative methods developed in the last decade to study human social behaviour, interactions between participants were monitored using the SocioPatterns platform, which allows collecting face-to-face physical proximity events every 20 seconds in a well-defined social context. Through accompanying surveys, we gathered extensive information about the participants, including sociodemographic characteristics, Big Five personality traits, DIAMONDS situational perceptions, measure of scientific attractiveness, motivations for attending the conferences, and perceptions of the crowd (e.g., in terms of gender distribution). Linking the sensor and survey data provides a rich window into social behaviour: At the individual level, the data sets allow personality scientists to investigate individual differences in social behaviour and pinpoint which individual characteristics (e.g., social roles, personality traits, situational perceptions) drive these individual differences. At the group level, the data also allow studying the mechanisms responsible for interacting patterns within a scientific crowd during a social, networking and idea-sharing event. The data are available for secondary analysis.

physics.soc-ph↗

arXiv2022

Characterizing the country-wide adoption and evolution of the Jodel messaging app in Saudi Arabia

Social media is subject to constant growth and evolution, yet little is known about their early phases of adoption. To shed light on this aspect, this paper empirically characterizes the initial and country-wide adoption of a new type of social media in Saudi Arabia that happened in 2017. Unlike established social media, the studied network Jodel is anonymous and location-based to form hundreds of independent communities country-wide whose adoption pattern we compare. We take a detailed and full view from the operators perspective on the temporal and geographical dimension on the evolution of these different communities -- from their very first the first months of establishment to saturation. This way, we make the early adoption of a new type of social media visible, a process that is often invisible due to the lack of data covering the first days of a new network.

cs.SI↗

arXiv2021

Inequality and Inequity in Network-based Ranking and Recommendation Algorithms

Though algorithms promise many benefits including efficiency, objectivity and accuracy, they may also introduce or amplify biases. Here we study two well-known algorithms, namely PageRank and Who-to-Follow (WTF), and show to what extent their ranks produce inequality and inequity when applied to directed social networks. To this end, we propose a directed network model with preferential attachment and homophily (DPAH) and demonstrate the influence of network structure on the rank distributions of these algorithms. Our main findings suggest that (i) inequality is positively correlated with inequity, (ii) inequality is driven by the interplay between preferential attachment, homophily, node activity and edge density, and (iii) inequity is driven by the interplay between homophily and minority size. In particular, these two algorithms reduce, replicate and amplify the representation of minorities in top ranks when majorities are homophilic, neutral and heterophilic, respectively. Moreover, when this representation is reduced, minorities may improve their visibility in the rank by connecting strategically in the network. For instance, by increasing their out-degree or homophily when majorities are also homophilic. These findings shed light on the social and algorithmic mechanisms that hinder equality and equity in network-based ranking and recommendation algorithms.

cs.SI↗

arXiv2021

Structack: Structure-based Adversarial Attacks on Graph Neural Networks

Recent work has shown that graph neural networks (GNNs) are vulnerable to adversarial attacks on graph data. Common attack approaches are typically informed, i.e. they have access to information about node attributes such as labels and feature vectors. In this work, we study adversarial attacks that are uninformed, where an attacker only has access to the graph structure, but no information about node attributes. Here the attacker aims to exploit structural knowledge and assumptions, which GNN models make about graph data. In particular, literature has shown that structural node centrality and similarity have a strong influence on learning with GNNs. Therefore, we study the impact of centrality and similarity on adversarial attacks on GNNs. We demonstrate that attackers can exploit this information to decrease the performance of GNNs by focusing on injecting links between nodes of low similarity and, surprisingly, low centrality. We show that structure-based uninformed attacks can approach the performance of informed attacks, while being computationally more efficient. With our paper, we present a new attack strategy on GNNs that we refer to as Structack. Structack can successfully manipulate the performance of GNNs with very limited information while operating under tight computational constraints. Our work contributes towards building more robust machine learning approaches on graphs.

cs.LG↗

arXiv2021

Redescription Model Mining

This paper introduces Redescription Model Mining, a novel approach to identify interpretable patterns across two datasets that share only a subset of attributes and have no common instances. In particular, Redescription Model Mining aims to find pairs of describable data subsets -- one for each dataset -- that induce similar exceptional models with respect to a prespecified model class. To achieve this, we combine two previously separate research areas: Exceptional Model Mining and Redescription Mining. For this new problem setting, we develop interestingness measures to select promising patterns, propose efficient algorithms, and demonstrate their potential on synthetic and real-world data. Uncovered patterns can hint at common underlying phenomena that manifest themselves across datasets, enabling the discovery of possible associations between (combinations of) attributes that do not appear in the same dataset.

cs.DB↗

arXiv2021

Group mixing drives inequality in face-to-face gatherings

Uncovering how inequality emerges from human interaction is imperative for just societies. Here we show that the way social groups interact in face-to-face situations can enable the emergence of disparities in the visibility of social groups. These disparities translate into members of specific social groups having fewer social ties than the average (i.e., degree inequality). We characterize group degree inequality in sensor-based data sets and present a mechanism that explains these disparities as the result of group mixing and group-size imbalance. We investigate how group sizes affect this inequality, thereby uncovering the critical size and mixing conditions in that a critical minority group emerges. If a minority group is larger than this critical size, it can be a well-connected, cohesive group; if it is smaller, minority cohesion widens degree inequality. Finally, we expose the under-representation of individuals in degree rankings due to mixing dynamics and propose a way to reduce such biases.

physics.soc-ph↗

arXiv2021

A Comparative Evaluation of Quantification Methods

Quantification represents the problem of estimating the distribution of class labels on unseen data. It also represents a growing research field in supervised machine learning, for which a large variety of different algorithms has been proposed in recent years. However, a comprehensive empirical comparison of quantification methods that supports algorithm selection is not available yet. In this work, we close this research gap by conducting a thorough empirical performance comparison of 24 different quantification methods on overall more than 40 data sets, considering binary as well as multiclass quantification settings. We observe that no single algorithm generally outperforms all competitors, but identify a group of methods including the threshold selection-based Median Sweep and TSMax methods, the DyS framework including the HDy method, Forman's mixture model, and Friedman's method that performs best in the binary setting. For the multiclass setting, we observe that a different, broad group of algorithms yields good performance, including the HDx method, the Generalized Probabilistic Adjusted Count, the readme method, the energy distance minimization method, the EM algorithm for quantification, and Friedman's method. We also find that tuning the underlying classifiers has in most cases only a limited impact on the quantification performance. More generally, we find that the performance on multiclass quantification is inferior to the results obtained in the binary setting. Our results can guide practitioners who intend to apply quantification algorithms and help researchers to identify opportunities for future research.

cs.LG↗

arXiv2021

Volunteer contributions to Wikipedia increased during COVID-19 mobility restrictions

Wikipedia, the largest encyclopedia ever created, is a global initiative driven by volunteer contributions. When the COVID-19 pandemic broke out and mobility restrictions ensued across the globe, it was unclear whether Wikipedia volunteers would become less active in the face of the pandemic, or whether they would rise to meet the increased demand for high-quality information despite the added stress inflicted by this crisis. Analyzing 223 million edits contributed from 2018 to 2020 across twelve Wikipedia language editions, we find that Wikipedia's global volunteer community responded remarkably to the pandemic, substantially increasing both productivity and the number of newcomers who joined the community. For example, contributions to the English Wikipedia increased by over 20% compared to the expectation derived from pre-pandemic data. Our work sheds light on the response of a global volunteer population to the COVID-19 crisis, providing valuable insights into the behavior of critical online communities under stress.

cs.CY↗