FATE: Fairness, Accountability, Transparency & Ethics of Machine Learning

Nowadays algorithms are everywhere and exert a huge influence on our lives. For example, algorithms are used to evaluate the performance of employees and students, to estimate the recidivism rate of criminals, to make medical diagnoses, to recommend products and to compile news feeds for our social media profiles. However, with such great power comes responsibility. The aim of this topic is to more deeply understand how algorithms influence individuals and society.
This topic consists of several thesis topics including:
Conducting a systematic literature survey on ...
- ... algorithms & fairness.
- ... algorithms & accountability.
- ... algorithms & transparency.
- ... algorithms & ethics.

Designing and conducting an experiment on ...
- ... the influence of transparency on algorithm trust.
- ... the influence of having control on algorithm trust.

Contact person: Annika Baumann, annika.baumannuuni-potsdampde

Perceptions of Autonomous Driving: Perceived Benefits and Risks

Recent technological advances might in the near future lead to more and more autonomous vehicles being present on the streets all over the world. However, with automation, there might be specific benefits and risks associated which are so far not well understood. The aim of the thesis will be to identify relevant discussions/forums to apply content analysis (and/or methods of text analytics such as topic modeling) to derive a deeper understanding on associated perception towards risks and benefits of society related to autonomous vehicles.

Contact person: Annika Baumann, annika.baumannuuni-potsdampde

The Influence of Smartwatch Usage on Psychological Well-Being

The influence of self-tracking device usage on physical well-being has been a topic of research for quite a while. However, how the usage of smartwatches and similar devices does influence the psychological well-being of individuals has not been understood so far. The goal of this master thesis is to design an experiment or to construct a survey which enables to answer this research question. The overall goal of the thesis should be to conduct the experiment or survey and derive results and provide answers towards the topic of interest. 

Contact person: Annika Baumann, annika.baumannuuni-potsdampde

The Rise of Chief Ethics Officer

Ethical issues are one of the top three risks of Artificial Intelligence (AI). Unlike many CXO roles, the chief ethics officer (including chief trust officer and chief ethics and compliance officer) doesn’t have a consistent job description. Some companies create their own code of conduct and rules for ethical behavior, even in the absence of outside guidance. For example, Salesforce hired Paula Goldman as chief ethical and humane use officer. Her work will likely focus on the intersection of ethics and AI.

The Master Thesis can cover different topics from a systematic literature review, to classifying companies’ ethical guidelines to conducting case studies with AI companies.
 
Contact person: Antonia Köster, antonia.koesteruuni-potsdampde

Current State of Research on Social Comparison and Status

The goal of this bachelor or master thesis is to present the current state of research on social comparison in the offline and offline context. We constantly compare ourselves to other people. For instance, we feel rich or poor based on the comparisons we make. In the end, the author of the thesis should be able to summarize the current knowledge on that topic. The method of this thesis is a systematic literature review.

Contact person: Antonia Köster, antonia.koesteruuni-potsdampde

Digitalpakt Deutschland 

Mit dem DigitalPakt Schule wollen Bund und Länder für eine bessere Ausstattung der Schulen mit digitaler Technik sorgen. Ziel der Bachelor oder Masterarbeit ist es, einen Datensatz mit Twitter Nachrichten der letzten 90 Tage zu dem Thema DigitalPakt auszuwerten. Der Datensatz mit ungefähr 800 Nachrichten wird von Frau Köster zur Verfügung gestellt.

Kontaktperson: Antonia Köster, antonia.koesteruuni-potsdampde

Uncertainty-Mitigating Signals on Sharing Economy Platforms

The advent of peer-to-peer sharing platforms, like Airbnb and Blablacar, has ushered in a new era in travel worldwide. However, to ensure sustainability in the long term, information asymmetry inherent to such platforms has to be tackled. Currently, sharing economy platforms offer a multitude of in-built trust-enhancing cues that may reduce information asymmetry, signal trust and aid potential guests in their decision making. Nevertheless, little is known about the effectiveness of these cues in shaping online consumption behavior. A thesis is expected to explore the effectiveness of trust-enhancing cues commonly deployed by service providers to promote trust in the sharing economy. Different methodologies are possible to apply depending on the skills of the student: eye-tracking tool, experiments, surveys, focus groups, interviews or data mining.

Contact Person: Olga Abramova, oabramov@uni-potsdam.de

Discrimination in Sharing Economy

Sharing Economy Platforms like Airbnb and Blablacar allow hosts/drivers to decide whether to accept or reject a traveler after seeing his or her name and often a picture—a market design choice that may further enable discrimination.

In this thesis, it is expected to investigate the existence and extent of discrimination (e.g. racial, gender, etc) on Sharing Economy platforms. Different methodologies are possible to apply depending on the skills of the student: experiments or data mining.

Contact Person: Olga Abramova, oabramov@uni-potsdam.de

Understanding Regret in the Context of Social Media

The proliferation and use of social media and online forums has led to a significant increase in text-based content. The simplicity of distributing content in the shortest time possible to a large network of diverse people can lead to ill-considered actions or mistakes that ultimately might lead to regret which then results into user-deleted content. The aim of the thesis is to understand how often and what kind of content gets deleted from the platform Reddit. To do so, it will be necessary to collect data from Reddit using the API and then to analyze content deletion behavior via a machine learning approach. The overall goal would be to understand which characteristics might ultimately lead to the deletion of content through analyzing the importance of employed features.

Contact person: Annika Baumann, annika.baumannuuni-potsdampde

The Anti-Vaxxer Movement: Particularities of Online Discussions

Recently the WHO stated that the anti-vaccine movement can be considered as one of the major health threats of the year 2019. Indeed, online communication related to this topic, both in the direction of proponents and opponents, has increased tremendously in recent years. The goal of this thesis would be to understand specific particularities of discussions related to the anti-vaccine movement. Main tasks of this explorative research topic would be to first identify relevant discussion forums and to then apply methods of text analytics to gain deeper insights on topics, sentiment, and emotions related to this topic.

Contact person: Annika Baumann, annika.baumannuuni-potsdampde

User Behavior in Online Social Networks: Self-presentation

Online social networks such as Instagram, Facebook, Snapchat, Twitter and LinkedIn have become extremely popular all over the world and play a significant role in people’s daily lives. A thesis is expected to explore behavioral patterns of its users (e.g. information disclosure, social connectivity and interaction, self-presentation) and their psychological consequences. Different methodologies are possible to apply depending on the skills of the student: eye-tracking tool, experiments, surveys, focus groups, interviews, data mining, systematic literature review or meta-review analysis.

Contact Person: Olga Abramova, oabramov@uni-potsdam.de

Lurking and Active Participation Online - Exploring the Role of Emotions

Online forums and social media depend on individuals’ active participation. However, the share of people who actively engage in comparison to so-called “lurkers” is considered to be small. The goal of this thesis is to quantify lurking and explore emotions expressed in language as a potential pathway using web-scraping and (basic) data science techniques.

Contact: Fenne große Deters, fenne.grosse.deters@uni-potsdam.de

Female refugees in Germany: An empirical study on the social inclusion of female refugees in Berlin/Brandenburg.

Since the refugee crisis in late 2014 and early 2015, refugees have been entering Europe to start their new, safe life. Several studies on how socially included the refugees are in Europe, especially in Germany as it attracted the highest number of refugees in the EU28, but none of these studies has investigated the effect of social inclusion on women in particular. In this master’s thesis, we will empirically (either quantitative, qualitative, or both) investigate this topic and how we can empower female refugees to feel more included in the new society.

Contact person: Safa'a AbuJarour, safaa.abujarour@uni-potsdam.de

Job opportunities for refugees: An investigation of the job market in the case of refugees.

In this thesis, we will scan the job market and investigate the offered job positions that accept refugees or are dedicated to this group. Interviews with refugees, industries, or NGOs that offer their help with respect to job offers for refugees are to be conducted to collect insights on how the refugees are accepted in the job market and how different stakeholders could contribute positively to the current situation.

Contact person: Safa'a AbuJarour, safaa.abujarour@uni-potsdam.de

 A Critical Look at the Design of Control Groups in Experiments

The goal of this bachelor or master thesis is to present how control groups should be ideally designed by drawing on literature from psychology. The author of the thesis should be able to summarize the current knowledge on control groups. In addition, this thesis should analyze how control groups are designed in experiments published in information systems journals. What can we learn? The method of this thesis is a systematic literature review.

Contact person: Antonia Köster, antonia.koesteruuni-potsdampde