Andreas Kaltenbrunner

Jacopo D'Ignazi, Emma Fraxanet Morales, Andreas Kaltenbrunner et al.

Despite extensive research, the mechanisms through which online platforms shape extremism and polarization remain poorly understood. We identify and test a mechanism, grounded in empirical evidence, that explains how ranking algorithms can amplify both phenomena. This mechanism is based on well-documented assumptions: (i) users exhibit position bias and tend to prefer items displayed higher in the ranking, (ii) users prefer like-minded content, (iii) users with more extreme views are more likely to engage actively, and (iv) ranking algorithms are popularity-based, assigning higher positions to items that attract more clicks. Under these conditions, when platforms additionally reward \emph{active} engagement and implement \emph{personalized} rankings, users are inevitably driven toward more extremist and polarized news consumption. We formalize this mechanism in a dynamical model, which we evaluate by means of simulations and interactive experiments with hundreds of human participants, where the rankings are updated dynamically in response to user activity.

Lorenzo Betti, Carlo Abrate, Andreas Kaltenbrunner

We employ Natural Language Processing techniques to analyse 377808 English song lyrics from the "Two Million Song Database" corpus, focusing on the expression of sexism across five decades (1960-2010) and the measurement of gender biases. Using a sexism classifier, we identify sexist lyrics at a larger scale than previous studies using small samples of manually annotated popular songs. Furthermore, we reveal gender biases by measuring associations in word embeddings learned on song lyrics. We find sexist content to increase across time, especially from male artists and for popular songs appearing in Billboard charts. Songs are also shown to contain different language biases depending on the gender of the performer, with male solo artist songs containing more and stronger biases. This is the first large scale analysis of this type, giving insights into language usage in such an influential part of popular culture.

Nurudin Alvarez-Gonzalez, Andreas Kaltenbrunner, Vicenç Gómez

Identifying similar network structures is key to capture graph isomorphisms and learn representations that exploit structural information encoded in graph data. This work shows that ego-networks can produce a structural encoding scheme for arbitrary graphs with greater expressivity than the Weisfeiler-Lehman (1-WL) test. We introduce IGEL, a preprocessing step to produce features that augment node representations by encoding ego-networks into sparse vectors that enrich Message Passing (MP) Graph Neural Networks (GNNs) beyond 1-WL expressivity. We describe formally the relation between IGEL and 1-WL, and characterize its expressive power and limitations. Experiments show that IGEL matches the empirical expressivity of state-of-the-art methods on isomorphism detection while improving performance on seven GNN architectures.

Ishari Amarasinghe, Salvatore Romano, Jacopo Amidei et al.

Estimation of mis/disinformation prevalence in social media is crucial for designing mitigation strategies to limit its impact. Yet, such estimations are subject to several uncertainties that are rarely quantified jointly. In this study, we present a methodological contribution in which confidence intervals were used to quantify uncertainties related to mis/disinformation prevalence. The analysis draws on a multi-platform, multilingual dataset annotated by professional fact-checkers. Data were collected between March and April 2025 from Facebook, Instagram, LinkedIn, TikTok, X/Twitter, and YouTube across four EU Member States (France, Poland, Slovakia, and Spain). We account for different causes of uncertainty: (i) sample uncertainty, (ii) annotation uncertainty arising from human disagreement and misclassification, and (iii) data retrieval uncertainty induced by keyword-based data collection. First, we estimate the uncertainty arising from the different causes separately using confidence intervals, simulation-based methods, and bootstrapping. Finally, we combined multinomial simulations of annotator behaviour with keyword and post-resampling to capture the joint impact of measurement uncertainty on mis/disinformation prevalence estimates. The proposed methodological approach highlights the importance of uncertainty-aware estimation of mis/disinformation prevalence for robust analysis. The empirical results of this study show that keyword-based data retrieval can exceed baseline variability, leading to wider confidence intervals around prevalence estimates.

Jacopo D'Ignazi, Andreas Kaltenbrunner, Yelena Mejova et al.

Over the last few years, verifying the credibility of information sources has become a fundamental need to combat disinformation. Here, we present a language-agnostic model designed to assess the reliability of web domains as sources in references across multiple language editions of Wikipedia. Utilizing editing activity data, the model evaluates domain reliability within different articles of varying controversiality, such as Climate Change, COVID-19, History, Media, and Biology topics. Crafting features that express domain usage across articles, the model effectively predicts domain reliability, achieving an F1 Macro score of approximately 0.80 for English and other high-resource languages. For mid-resource languages, we achieve 0.65, while the performance of low-resource languages varies. In all cases, the time the domain remains present in the articles (which we dub as permanence) is one of the most predictive features. We highlight the challenge of maintaining consistent model performance across languages of varying resource levels and demonstrate that adapting models from higher-resource languages can improve performance. We believe these findings can assist Wikipedia editors in their ongoing efforts to verify citations and may offer useful insights for other user-generated content communities.

Nurudin Alvarez-Gonzalez, Andreas Kaltenbrunner, Vicenç Gómez

We present a novel edge-level ego-network encoding for learning on graphs that can boost Message Passing Graph Neural Networks (MP-GNNs) by providing additional node and edge features or extending message-passing formats. The proposed encoding is sufficient to distinguish Strongly Regular Graphs, a family of challenging 3-WL equivalent graphs. We show theoretically that such encoding is more expressive than node-based sub-graph MP-GNNs. In an empirical evaluation on four benchmarks with 10 graph datasets, our results match or improve previous baselines on expressivity, graph classification, graph regression, and proximity tasks -- while reducing memory usage by 18.1x in certain real-world settings.

Nurudin Alvarez-Gonzalez, Andreas Kaltenbrunner, Vicenç Gómez

Identifying emotions from text is crucial for a variety of real world tasks. We consider the two largest now-available corpora for emotion classification: GoEmotions, with 58k messages labelled by readers, and Vent, with 33M writer-labelled messages. We design a benchmark and evaluate several feature spaces and learning algorithms, including two simple yet novel models on top of BERT that outperform previous strong baselines on GoEmotions. Through an experiment with human participants, we also analyze the differences between how writers express emotions and how readers perceive them. Our results suggest that emotions expressed by writers are harder to identify than emotions that readers perceive. We share a public web interface for researchers to explore our models.

Nurudin Alvarez-Gonzalez, Andreas Kaltenbrunner, Vicenç Gómez

Learning embeddings from large-scale networks is an open challenge. Despite the overwhelming number of existing methods, is is unclear how to exploit network structure in a way that generalizes easily to unseen nodes, edges or graphs. In this work, we look at the problem of finding inductive network embeddings in large networks without domain-dependent node/edge attributes. We propose to use a set of basic predefined local encodings as the basis of a learning algorithm. In particular, we consider the degree frequencies at different distances from a node, which can be computed efficiently for relatively short distances and a large number of nodes. Interestingly, the resulting embeddings generalize well across unseen or distant regions in the network, both in unsupervised settings, when combined with language model learning, as well as in supervised tasks, when used as additional features in a neural network. Despite its simplicity, this method achieves state-of-the-art performance in tasks such as role detection, link prediction and node classification, and represents an inductive network embedding method directly applicable to large unattributed networks.

Erik Borra, Andreas Kaltenbrunner, Michele Mauri et al.

Collaborative content creation inevitably reaches situations where different points of view lead to conflict. We focus on Wikipedia, the free encyclopedia anyone may edit, where disputes about content in controversial articles often reflect larger societal debates. While Wikipedia has a public edit history and discussion section for every article, the substance of these sections is difficult to phantom for Wikipedia users interested in the development of an article and in locating which topics were most controversial. In this paper we present Contropedia, a tool that augments Wikipedia articles and gives insight into the development of controversial topics. Contropedia uses an efficient language agnostic measure based on the edit history that focuses on wiki links to easily identify which topics within a Wikipedia article have been most controversial and when.

Nikolaos Lykousas, Costantinos Patsakis, Andreas Kaltenbrunner et al.

The continuous and increasing use of social media has enabled the expression of human thoughts, opinions, and everyday actions publicly at an unprecedented scale. We present the Vent dataset, the largest annotated dataset of text, emotions, and social connections to date. It comprises more than 33 millions of posts by nearly a million of users together with their social connections. Each post has an associated emotion. There are 705 different emotions, organized in 63 "emotion categories", forming a two-level taxonomy of affects. Our initial statistical analysis describes the global patterns of activity in the Vent platform, revealing large heterogenities and certain remarkable regularities regarding the use of the different emotions. We focus on the aggregated use of emotions, the temporal activity, and the social network of users, and outline possible methods to infer emotion networks based on the user activity. We also analyze the text and describe the affective landscape of Vent, finding agreements with existing (small scale) annotated corpus in terms of emotion categories and positive/negative valences. Finally, we discuss possible research questions that can be addressed from this unique dataset.