Juan Carlos ist einer Data Enthusiast (data scientist * data engineer * machine learning engineer). Er wurde in Mexiko geboren und studierte Technische Physik an der Universität Tecnológico de Monterrey. Anschließend absolvierte er zwei Masterprogramme in München: Computational Science und Data Engineering. Juan Carlos arbeitete als Data Scientist bei Firmen wie BMW, Bosch und Siemens. Er hat an der Technischen Universität München promoviert, wo er die Auswirkungen von Soziale Netzwerken und Misinformation auf die Deutsche Politik erforscht hat. Derzeit arbeitet er als Head of Data von der CSU und ist für die digitale Strategie zuständig.
Orestis Papakyriakopoulos ist Postdoc am Center for Information Technology Policy an Princeton University. Er forscht Sozio-algorithmische Ökosysteme durch die Anwendung Daten-intensiven Modellen und sozialen Theorien. Orestis studierte Bauingenieurwesen (Dipl. Ing) an der TU Athen und Wissenschaft- und Technikphilosophie an der TU München. Er promovierte in Informatik an der TU München. In 2019-2020 war er Visiting Scholar am MIT Center for civic media. In der Vergangenheit war er sowohl als Ingenieur, als auch in der Wissenschafts- und Technikkommunikation tätig.
Morteza ist Postdoc am Corporate Communications research an der Universität St. Gallen. Er hat seine Promotion an der TU München abgeschlossen. In seiner Forschung wendet er Graph-theoretischen Methoden um Online Social Media zu analysieren. Morteza ist auch als Datenwissenschaftler am AI+Automation Lab an der Bayerischer Rundfunk tätig. Da entwickelt neue Methoden der Künstlichen Intelligenz, um Journalistische Arbeit zu unterstutzen und vereinfachen.
Dieses Projekt wäre ohne und die Unterstützung von Simon Hegelich, Professor für Political Data Science an der TU München, nicht möglich gewesen.
The Media During the Rise of Trump: Identity Politics, Immigration, "Mexican" Demonization and Hate-Crime
In this study, we investigate the role of the US online media ecosystem in Donald Trump’s rise and consolidation to power (2013-2019). We analyze over 54 million articles from online U.S. media and locate a media narrative shift related to three issues that Trump focused on during his 2016 presidential campaign: immigration, Latinx people, and identity politics. Given this, we develop Natural Language Processing techniques based on word embeddings to quantify biased representations of minorities in the media across time. We locate an increase in biased speech that parallels Trump’s rise to power, and a clear partisan pattern to this bias. https://ojs.aaai.org/index.php/ICWSM/article/view/18076
Politische Werbung und Microtargeting auf Facebook
Political advertisement in online social networks (OSN) has started to play a major role: By targeting, analysing and evaluating personal user data within OSN’s (so called microtargeting) it is possible to tailor campaigning to an individual user or spread political ads and publicity in contradictive user groups. While microtargeting is already employed by parties and politicians during the US presidential election campaigns for quite some time, in Germany this trend is just about to begin. Beside politicians and political parties, in Germany there is a variety of political actors, institutions and organisations with their own political beliefs. Among those, the german labour unions and the employer’s federations pay and post continuously political ads. Are they already targeting users in OSN’s to individualize their publicity? In the following paper we would like to enter into that question by analysing a data set from the Facebook Ad Library.
Political machines: a framework for studying politics in social machines
In the age of ubiquitous computing and artificially intelligent applications, social machines serves as a powerful framework for understanding and interpreting interactions in socio-algorithmic ecosystems. Although researchers have largely used it to analyze the interactions of individuals and algorithms, limited attempts have been made to investigate the politics in social machines. This study claims that social machines are per se political machines, and introduces a five-point framework for classifying influence processes in socio-algorithmic ecosystems. https://link.springer.com/article/10.1007/s00146-021-01180-6/
The spread of COVID-19 conspiracy theories on social media and the effect of content moderation
This study investigates the possibilities and limits presented by the newly created ad libraries from Facebook and Google to analyze online political campaigns. We selected Germany as a case study and focused on the months leading up to the 2019 elections to the European Parliament. We found that even though all the major German political parties engaged in online ad campaigns, they kept their attempts at microtargeting to a minimum. Although their Facebook-sponsored posts were more successful than normal posts, we did not find statistical significance for all the political parties. https://misinforeview.hks.harvard.edu/article/the-spread-of-covid-19-conspiracy-theories-on-social-media-and-the-effect-of-content-moderation/
NLP-based Feature Extraction for the Detection of COVID-19 Misinformation Videos on YouTube
This study esents a simple NLP methodology for detecting COVID-19 misinformation videos on YouTube by leveraging user comments. It uses transfer-learning pre-trained models to generate a multi-label classifier that can categorize conspiratorial content with up to than 90% accuracy.
Exploring Political Ad Libraries for Online Advertising Transparency: Lessons from Germany and the 2019 European Elections
This study investigates the possibilities and limits presented by the newly created ad libraries from Facebook and Google to analyze online political campaigns. We selected Germany as a case study and focused on the months leading up to the 2019 elections to the European Parliament. We found that even though all the major German political parties engaged in online ad campaigns, they kept their attempts at microtargeting to a minimum. Although their Facebook-sponsored posts were more successful than normal posts, we did not find statistical significance for all the political parties. https://socialmediaandsociety.org/ publications/
Dancing to the Partisan Beat: A First Analysis of Political Communication on TikTok
This study aims to perform a primary evaluation of political communication on TikTok. We collect a set of US partisan Republican and Democratic videos to investigate how users communicated with each other about political issues. We illustrate that political communication on TikTok is much more interactive in comparison to other social media platforms. Republican users generated more political content and their videos received more responses; on the other hand, Democratic users engaged significantly more in cross-partisan discussions. https://arxiv.org/ pdf/2004.05478.pdf
Bias in word embeddings
Recent studies demonstrate that word embeddings contain and amplify biases present in data, such as stereotypes and prejudice. In this study, we provide a complete overview of bias in word embeddings. We develop a new technique for bias detection for gendered languages and use it to compare bias in embeddings trained on Wikipedia and on political social media data. We investigate bias diffusion and prove that existing biases are transferred to further machine learning models. We test two techniques for bias mitigation and show that the generally proposed methodology for debiasing models at the embeddings level is insufficient. Finally,we employ biased word embeddings and illustrate that they can be used for the detection of similar biases in new data.
Political communication on social media: A tale of hyperactive users and bias in recommender systems
A segment of the political discussions on Online Social Networks (OSNs) is shaped by hyperactive users. These are users that are over-proportionally active in relation to the mean. By training collaborative filtering and deep learning recommendation algorithms on simulated political networks, we illustrate that models provide different suggestions to users, when accounting for or ignoring hyperactive behavior both in the input dataset and in the methodology applied. We attack the trained models with adversarial examples by strategically placing hyperactive users in the network and manipulating the recommender systems’ results. Given that OSNs are not per se designed to foster political discussions, we discuss the implications for the political discourse and the danger of algorithmic manipulation of political communication.
The Rise of Germany’s AfD: A Social Media Analysis
This paper seeks to understand the AfD’s social media strategy over the last years on the full gamut of social media platforms and to verify the effectiveness of the party’s online messaging strategy. For this purpose, we collected data related to Germany’s main political parties from Facebook, Twitter, YouTube, and Instagram. This analysis proves the AfD’s superior online popularity relative to the rest of Germany’s political parties. The evidence also indicates that automated accounts contributed to this online superiority.
Measuring the Ease of Communication in Bipartite Social Endorsement Networks: A Proxy to Study the Dynamics of Political Polarization
In this work, complex weighted bipartite social networks are developed to efficiently analyze, project and extract network knowledge. Specifically, to assess the overall ease of communication between the different network sub-clusters, a proper projection and measurement method is developed in which the defined measurement is a function of the network structure and preserves maximum relevant information. Using simulations, it is shown how the introduced measurement correlates with the concept of political polarization. The method successfully captured the increasing political polarization between the Alternative für Deutschland’s (AfD) supporters and the supporters of other political parties.
Distorting Political Communication: The Effect Of Hyperactive Users In Online Social Networks
Online Social Networks (OSNs) are used increasingly for political purposes. Among others, politicians externalize their views on issues, and users respond to them, initiating political discussions. Part of the discussions are shaped by hyperactive users. These are users that are over-proportionally active in relation to the mean. In this paper, we define the hyperactive user on the social media platform Facebook, both theoretically and mathematically. We apply a topic modelling algorithm on German political parties' posts and user comments to identify the topics discussed. We prove that hyperactive users have a significant role in the political discourse: They become opinion leaders, as well as set the content of discussions, thus creating an alternate picture of the public opinion.
Estimating the Political Orientation of Twitter Users in Homophilic Networks
There have been many efforts to estimate the political orientation of citizens and political actors. With the burst of online social media use in the last two decades, this topic has undergone major changes. Many researchers and political cam-paigns have attempted to measure and estimate the political orientation of online social media users. In this paper, we use a combination of metric learning algorithms and label propa-gation methods to estimate the political orientation of Twitter users.
https://pdfs.semanticscholar.org/ f1ea/5c8a0abddee04db40347 d31fda70b2506c29.pdf
Social media and microtargeting: Political data processing and consequences for Germany
This study investigates the actual possibilities and legal limits of performing data-driven microtargeting in Germany. It takes into consideration GDPR, the role of social media companies and the epistemological and ethical issues in
evaluating the political opinion of potential voters.
Social media report: the 2017 German federal elections
This report presents thorough research on social media platforms during the months before the 2017 German federal election. The focus is on Facebook and Twitter given their increasing role in online political communication. Over 350
million tweets and 37 thousand Facebook posts related to German politics were collected and analyzed. This work takes an overlook at the online interaction between users and political parties. Moreover, it tries to identify
disinformation and manipulation techniques.
Social Media und Microtargeting in Deutschland
We investigate the possibilities and limits of microtargeting based on social-media data. We evaluate politically and ethically the consequences of the specific campaigning technique for the political system.