Trust in Digital Services
Theses

Below you find our call for thesis applications, instructions for writing a thesis as well as an overview about current and previous theses.

Call for Thesis Applications

Below you will find an overview of currently offered topics for theses. If you are interested, you can apply by e-mail to the respective supervisor. Please state the topic and submit your curriculum vitae and transcript of records. Before applying, please read our instructions, especially the information regarding the publication track.

Instructions

Requirements

Prerequisites are an interest in topics related to digital platforms and online trust as well as the successful completion of at least one of our courses.

Publication Track

In the publication track, a submission of the thesis to a journal is intended. On the one hand, this means that the requirements to the thesis are more demanding than those for a usual thesis. Hence, completing the thesis in this format can take longer (6 + max. 2 months). However, you can expect more support and a closer cooperation with the supervisor. The publication track is particularly interesting for students who want to pursue a PhD and an academic career. For more information on the publication track, please contact us.

Supervision & Procedure

The supervision of your thesis by a researcher of the department Trust in Digital Services (TDS) ensures that you can adequately address a relevant (theoretical or practical) problem with the help of a suitable scientific method (usually empirical) within the given processing time. Your supervisor is the number one contact person for all questions concerning the content and form of your thesis. Together you will create the table of contents, on which the famous "red thread" of your work depends. Typically, the contact time is most intensive in the initial phase of the thesis and decreases continuously until shortly before submission.

Defense

A characteristic of the department is the mandatory participation in the final colloquium. In the course of this event, you will present and "defend" your research results after submitting your thesis. Participation in the final colloquium is intended to give you the opportunity to become acquainted with relevant topics from our field of research and to help you reflect on your research results and learn to defend them in front of a professional audience.

The final colloquium is a compulsory event - both actively in the role as a speaker and passively in the role as a listener to the presentations of your fellow students. The final colloquium will count for 10% of the overall evaluation of your thesis.

Each speaker is allowed a maximum of 30 minutes. The presentation itself comprises a maximum of 15 minutes, followed by a discussion with fellow students and the departmental team of a maximum of 15 minutes.

The presentation can be given either in German or English. If the presentation is held in German, the presentation slides have to be prepared in English.

Either the TU Berlin's own template (see on the right) or your own presentation template can be used.

Current and completed Theses

2023

Trust in AI (ChatGPT)

Tim Michaelis examines trust in advanced NLP models that generate human-like responses. For his master's thesis, he conducts an experimental study, using ChatGPT as an example. Given the rapid rise of this technology and the increasing number of users, the need for trust in these systems becomes evident. A particular focus of the study is the measurement of transparency and performance and how these reflect user-based trust.

Confirmation bias in news readers

People tend to seek out information that supports their beliefs and opinions. In his master’s thesis, Gregor Pahlitzsch develops and tests a democratic platform design for online newsreaders overcoming this conformation bias.

Authenticity of medical reviews on Jameda

Medical reviews are a valuable source of information for both patients and doctors. However, the authenticity and trustworthiness of medical reviews are often questioned, as they may be subject to bias or manipulation. In his master's thesis, Yong Guo tests the authenticity of medical reviews on Jameda, using semantic analysis by applying NLP methods on a set of data scraped from the platform.

User trust in ChatGPT

With the publication of ChatGPT, a potentially disruptive technology for various segments of society has been introduced. As an advanced conversational language model, ChatGPT resembles human-like interaction which can be a key factor in generating user trust. In her master's thesis, Maria Schmidt investigates ChatGPT's perceived interaction quality and its effect on user trust through an experimental study. 

Gender role congruity index

Role Congruity Theory by Eagly & Karau (2002) posits that certain aspects of gender roles ascribed to women clash with qualities desirable for leadership, leading to unfavorable evaluations of female leaders. In his master's thesis, Marc Riewe develops an index to quantify the level of incongruity which is then tested experimentally.

Long Tail or Long Tail Fallacy?

In his master's thesis, Marc Radau investigates consumer decisions in light of varying product variety and the device used in an experimental study. Working under the hypothesis that different devices allow for different rates of information processing, he investigates whether consumers choose best sellers or go for the long tail.

Online rating behavior

Online reviews have gained more and more impact on consumer buying decisions. In her bachelor's thesis, Trang Nguyen conducts a systematic literature review to assess how people give ratings and reviews, with a focus on underlying psychological mechanisms.

Regionality cues on platforms

Individuals perceive their own (regional) group as the "center of the universe" and thus avoid purchasing non-regional products. In his master’s thesis, Eric Schrappe investigates the impact of embedded regional cues in the UI through a large-scale field experiment. An A/B test was designed to observe how users behave when encountering these cues as well as how these affect their trust and purchase intention.

Digital identity in trust-intensive services

In his master's thesis, Jan Kursawe addresses the development of a taxonomy for digital identity. Through a three-rounded Delphi study, the research examines the landscape, actors, data collection, and usage of digital identity. The study aims to shed light on the sustainability and potential implications of digital identity on the future of digital services.

2022

Review Authenticity on Airbnb

Online reviews have a strong influence on the decisions of customers, and therefore many fake reviews exist, written to push the own sales or weaken competitors. In this master’s thesis, Florian Schuler analyses the authenticity of reviews on Airbnb.com, based on semantic and linguistic features.

Employing low-code to accelerate digitalization

Digitalization is far from a new phenomenon and a steady companion in our daily lives, but especially in the manufacturing industry, digitalization is slow to catch on. In his master's thesis, Lukas Brauner investigates the acceleration of digitalization through the use of low-code development and the rise of the citizen developer.

Convolutional Neural Networks

Kai Fröhner's bachelor's thesis is centered around the optimisation of a classification of a small dataset of images of couches on their architectural style using Convolutional Neural Networks (CNN). It relies on fine-tuning and modifications of the datasets for the CNN.

Opinion leaders and communication on Telegram

With the rise of online social media use during the COVID-19 pandemic, new communities and communication patterns emerged. In his bachelor's thesis, Béla Padberg collects and analyses public German-language Telegram messages, identifying opinion leaders and evaluating discussed topics.

Fake reviews in the platform economy

Due to the significant impact of user reviews on the financial success on platforms, sellers have started using deceptive strategies to gain an advantage over their competition. In this master's thesis, Aleksandra Panek investigated the prevalence of fake reviews on Airbnb by applying various statistical methods on a set of data scraped from the platform.

Value of skills on freelancing platforms

In his master’s thesis, Mats Jäger investigates profiles on online freelancing platforms with respect to the effect of skills on hourly wages. One of his findings is that niche skills which increase specialization may add a premium to asking wages.

Advancement of reputation systems

In her bachelor's thesis, Rebecca Schönknecht investigates the influence of various features, extensions and modifications of reputation systems on user trust. Her experimental online survey revealed that an introduction video, link to social media account or information regarding repeated transactions can increase trust.

Online political microtargeting

Using the example of the 2019 european election in Germany, Leonard Kinzinger analyses microtargeting as a politicial communication strategy in election campaign. In his bachelor's thesis, he revealed that all major parties used the methods for digital marketing campaigns that are available on Facebook and implemented regional microtargeting.

Civic Participation in Data Analysis

Data literacy is an indispensable prerequisite for participation in today's digitized society. However, current citizen science initiatives face the challenge of engaging citizens in data analysis and interpretation to create learning opportunities. In this master's thesis, Carolin Stein explored the design of a conversational agent that supports data exploration and lowers the barriers to citizen participation in research projects.

Label for regional and green electricity

In her bachelor's thesis, Catayoun Azarm empirically investigates the application of a regional ecolabel on user interfaces in terms of user preferences and behaviour. By means of a multi-method approach, combining the eye-tracking technology with an online survey, gaze and survey data are captured and analysed.

Trust cues on energy comparison platforms

The aim of Elif Senol's master's thesis is to see how and to what extent trust mechanisms are used in energy price comparison platforms. In addition, the correlation between various trust-building attributes and offer prices is investigated. For this purpose, besides a literature review, data from a real platform is analyzed based on hedonic price modeling.

Solving Heterogenous Economic Models

In her master's thesis, Yanan Zhang uses deep reinforcement learning algorithm to simulate a dynamic stochastic general equilibrium model with heterogeneous multi-agent and exogenous shocks. She uses PettingZoo API to customize the environment and Stabel-baseline3 to train the model. Eventually the program achieves the convergence and finds the optimal policy for each agent.

Preventing eutrophication and overturning of lakes

In this master's thesis, Jakob Waidner applies the Design Science Research Method to tackle the issue of eutrophication in urban freshwater lakes. Together with the municipal utilities at Frankfurt/M. a digital platform is developed that helps decision-makers implement dedicated nature-based solutions facing the impact of climate change on vulnerable ecosystems.

Experience and reputation in high-risk environments

Trust on platforms is the key to success, especially in high-risks environments. Experience and reputation are important parameters in a service provider's profile and of great importance in the decision- making process of a user. In this master's thesis, Robert Taglauer analyzes the results from an online experiment to investigate how experience and reputation of a profile offering a high-risk service (like a tandem flight) engender trust.

The Economics of Multi-Hop Ride Sharing

Despite the economic and environmental benefits of ride sharing, it remains less popular compared to other modes of transport. One way to increase its attractivity is to allow multi-hop ride sharing with transfers in-between. In her master's thesis, Sarah Gia Nastiti examines the perception of ride-sharing platform customers on the concept of multi-hop ride sharing through the lens of a qualitative research.

Impact of online rating discrepancies

Consumers rely on reputations such as star ratings to assess a product/profile offered online. Reputation portability makes it possible to import ratings from other platforms so that several ratings from different providers are displayed on one platform. In this master's thesis, Pavel Dykmann uses an eye tracking experiment to investigate the effect of deviating ratings on consumer behaviour and trust.

Bibliometric network analysis

In his bachelor's thesis, Edison von Matt deals with applied bibliometric network analysis for the evaluation of literature corpora.

Digital Apps for University Education

Against the backdrop of the Corona pandemic, Imam Ihsan's master's thesis explores the role of digital apps as learning tools in higher education through a survey he developed and conducted.

Platform Capitalism vs. Cooperativism

Digital platforms are forced to reposition themselves on a socio-economic level in order to survive. In his bachelor's thesis, Luis Fischer examines the transformation from platform capitalism to platform cooperativism. How sustainable and future-proof is the platform cooperative model?

White Label Platforms in Crowdfunding

Increasingly, large players such as banks are entering the growing crowdfunding market with their own platforms. In her master's thesis, Lixuan Tu uses expert interviews to examine the implications of the white label model for crowdfunding stakeholders.

Herd Behavior in Equity Crowdfunding

In her bachelor's thesis, Nhu Ngoc Nguyen analyzes field data from an experiment on the visibility of investors on two crowdinvesting platforms. The data comes from her supervisor's cooperation with the white-label provider Portagon (http://portagon.com).

Filter Bubbles in Political Decision-Making

In her master's thesis, Hannah Carstens experimentally investigates so-called "information bubbles" as triggers for political opinion formation on meta-news sites.
 

Wage Gaps in the Platform Economy

In his master's thesis, Jonathan Bruns examines gender wage gaps in the platform economy using the example of Helpling (www.helpling.de), a platform for arranging household-related services, including cleaning in private households and offices. 

Identifying phishing with natural language processing

Phishing emails are a major threat to cybercitizens. To be able to take advantage of phishing emails, attackers are trying to manipulate victims' emotions. In this master's thesis, Atacan Korkmaz uses two 2 fine-tuned BERT base models, aimed at detecting phishing emails as well as the emotional weight of phishing emails, which affects the recipient's decision-making.

Popularity and online activity

In his master’s thesis, Jonas Nowottnick investigates whether relatively unknown politicians can increase their public presence with the use of social media. His analysis revealed that activity on social media does not affect the level of attention that a politician receives.

Strategies for future mobility

Joshua Mania investigated how multi-hop ride sharing can contribute to more environmentally friendly mobility. The study for his master's thesis confirmed that multi-hop ride sharing increases the absolute number of trips between any cities and the supply of trips that were not offered before.

AI in platform-based business models

Artificial intelligence has become increasingly important in the further development of machines and processes. In her bachelor’s thesis, Greta Mickunaite investigated the success factors of companies with platform business models in relation to artificial intelligence. The success factors were determined through a structured literature review and illustrated with the examples of Airbnb and Amazon.

2021

Discrimination by artificial intelligence

In her bachelor's thesis, Lotte Grünwald examines ethics in the context of artificial intelligence. The aim of the thesis was to "investigate moral aspects of the public perception of algorithmic fairness with a special focus on the risks for structurally discriminated minorities".

Online reviews and purchase decisions

In his master's thesis, Kaan Isik investigates the impact of product reviews on consumers' purchase decisions. He uses historical data from an online retail store and examines the conversion rates of products before and after a review.

Sustainable tourism by means of platform data

In his master's thesis, Felix Hoffmann estimates the share of sustainable tourism jobs in all tourism jobs (SDG 8.9.2) on the European market, using data from an online travel platform.

Current and future gentrification

Gentrification is a widespread socio-economic problem and at the same time an opportunity, if handled correctly. In his master's thesis, Paul Spende investigates methods of present and forecast gentrification based on platform data.

Trust signals in crowdfunding

Crowdfunding is a form of financing shaped by the internet, in which a large number of investors ("crowd") participate financially in projects, usually with small amounts, in expectation of a return. In his master's thesis, René Kyewski investigates the influence of investor visibility on the investment decisions of the crowd and the funding success of campaigns.

Technical literature review tool

In his bachelor's thesis, Luís Serrano Ramos developped a tool for technical literature reviews based on Google Scholar and selected publisher websites. Never before has it been so easy to identify patterns of authors, keywords, and their relationships to each other on a given topic.

Fake Reviews in e-commerce

In this master's thesis, Niklas Radtke investigates the phenomenon that customer reviews are systematically bought and faked for profit-increasing purposes in e-commerce. The goal is to identify such fake reviews on Amazon.com by means of computer-aided algorithms.

Online apartment descriptions

In his bachelor's thesis, Max Palupski investigates the effects of the level of detail in online apartment descriptions on customer interest and bookings. The thesis was conducted in cooperation with the Berlin-based apartment agency Wunderflats (www.wunderflats.com).

Reputation portability design

In his master's thesis, Valentin Duesterberg researches the design of interfaces for reputation portability on online marketplaces.

2020

Trust in artificial intelligence (AI)

In her master's thesis, Johanna Langfeldt uses experimental data to investigate the influence of transparency of decision support systems based on artificial intelligence (AI) on users' trust in this AI.

Reputation portability or trust "to go"

In his bachelor's thesis, Nicolas Wolski uses data from the platform Bonanza.com to investigate the influence of online feedback from customers (Star Ratings) on the economic success of sellers on the platform. The special feature of Bonanza.com is that both the platform's own ratings and those imported from eBay are displayed.

Characteristics and modes of actions of B2B platforms

In this master's thesis, Inga Stange investigates which characteristics and modes of action are responsible for the success of B2B platforms (for the mediation of services and goods between companies). Furthermore, she examines which control options operators and users have in their hands.

Trust on green power platforms

Digital transformation is a key driver of the current energy transition towards a green and sustainable future. In his master's thesis, Thanh Ngo Chi investigates trust-building signals identified in green power platforms and their applicability to power grid service platforms.

Text ratings in the platform economy

In her master's thesis, Sezin Sezgin uses large text review datasets to investigate the influence of the "sentiment" of reviews on the success of providers on peer-to-peer platforms. One specific question is whether one "bad" review is enough to render many "very good" reviews ineffective.

Skills in demand in online job ads

In her bachelor's thesis, Linda Schulz (née Färber) uses data from an online job advertising agency and Natural Language Processing to investigate which applicant skills are in demand, in which clusters they appear, and how they are related to other factors such as industry, hierarchy level, etc.

The Long-Tail-Theory and consumer behavior

According to the Long-Tail-Theory, increased variety steers customers towards niche markets and products, which may fulfill their preferences more accurately than bestseller products. However, within an experiment conducted for his bachelor’s thesis, Berk Güngörün observed that people tend to choose bestseller products with increasing variety.