The Recommender Systems research area focuses on the development and application of AI-powered recommendation systems that assist users in selecting products, information, or services. By analyzing large amounts of data and identifying individual preference patterns, recommender systems can generate personalized suggestions that simplify decision-making processes and reduce information overload. The focus is on the interplay between algorithmic predictive power and human judgment: While AI calculates recommendations based on past interactions, users retain control over selection, evaluation, and the final decision.

Possible Topics

• How Spotify & Co. Generate Recommendations – Implementation and Evaluation of a Recommendation System (BA/MA; Contact: Leonie Embacher) 
  #Analytics, #Programming
• Trust in AI-powered recommendation systems – How does explainability affect user acceptance? (BA/MA; Contact: Leonie Embacher) 
  #Explainability, #AI, #Experiment
• Explainability of review-based recommender systems – A systematic literature review (BA/MA; Contact: Leonie Embacher)
  #Explainability, #Literature-based
• Human Judgment and Algorithmic Recommendations – The Influence of Recommendation Systems on Human Decision-Making Processes (BA/MA; Contact: Leonie Embacher)
  #Decision Quality, #Literature-based
• Recommender Systems in (Social) Media - Approaches to Reducing Filter Bubbles and Echo Chambers (BA/MA; Contact: Kilian Züllig)
  #FilterBubbles, #Literature-Based, #Analytics

The topics will be individually coordinated and specified with the students in terms of content and methodology. The requirements for the topics vary depending on the degree level (Bachelor’s/Master’s). 

In an increasingly digital world, people are spending more and more time in digital environments and making decisions there. Digital nudging involves the strategic use of design elements to encourage users in digital decision-making environments—such as digital platforms—to make certain choices through small prompts (known as nudges), without coercion or restrictions.

Possible Topics

• Digital Green Nudging: An analysis of the effectiveness of digital nudges in promoting eco-friendly behavior. (BA & MA; Contact: Jana Ruß, Christopher Tille) 
  #Sustainability, #Experimental, or #Literature-based
• Identification of spillover effects of digital nudges on other decisions (e.g., different context, different time, …) (BA & MA; Contact: Jana Ruß, Christopher Tille)
  #Sustainability, #Experimental, or #Literature-based
• Personalized digital nudges and their potential. (BA & MA; Contact: Jana Ruß, Christopher Tille) 
  #Personalization, #Analytics, #Coding, or #Experimental
• Context-Aware Nudging: Development of situational digital nudging in real time. (BA & MA; Contact: Jana Ruß, Christopher Tille) 
  #Individualization, #Analytics, #Programming, or #Literature-Based

The content and methodology of the topics will be individually coordinated and specified with the students. The requirements for the topics vary depending on the degree (Bachelor’s/Master’s). 

Modern professional sports have evolved from purely intuitive decisions to “data-driven decision-making.” The use of modern tracking technologies generates vast amounts of data that hold untapped potential. Sports analytics focuses on the systematic analysis of this data to identify competitive advantages. The goal is to use advanced statistical methods and machine learning to gain objective insights for tactics, player recruitment, workload management, and economic decisions.

Possible Topics

• Development of a 360-degree player performance dashboard: design and implementation. (BA & MA; Contact: Kilian Züllig)
  #Basketball, #Application-oriented, #Analytics, #Coding
• Beyond Goals & Assists: Evaluating the quality of individual player actions in Bundesliga soccer using advanced methods. (BA & MA; Contact: Kilian Züllig)
  #Soccer, #Literature-based, #Analytics, #Coding
• Injury Prevention through Machine Learning: Analysis of training data (GPS, heart rate) to predict injury risks in competitive sports. (BA & MA; Contact: Kilian Züllig) 
  #TeamSports, #Analytics, #Coding, or #Literature-Based
• Tactical Pattern Recognition: Using positional data for the automated recognition of plays and formations in team sports. (BA & MA; Contact: Kilian Züllig)
  #TeamSports, #Analytics, #Coding, or #Literature-Based 

The content and methodology of the topics are determined and finalized in consultation with each student. The requirements for the topics vary depending on the degree program (Bachelor’s/Master’s). 

The Societal AI research area explores how artificial intelligence can help address societal challenges. In a societal context, the focus is on ethical, fair, and transparent AI applications that promote social inclusion, responsible decision-making support, and societal benefits.

Possible Topics

• How does effective collaboration between humans and GenAI work? What are the societal consequences of blind trust in GenAI? (BA & MA; Contact: Christopher Tille, Kirsten Pitz)
  #GenAI, #Trust, #Literature-based, #Experiment
• What impact does GenAI have on the education system? Which skills are becoming increasingly important with the comprehensive integration of GenAI into our daily lives? (BA & MA; Contact: Christopher Tille, Kirsten Pitz)
  #GenAI, #Education, #Literature-based
• Disinformation with GenAI – What are the effects on political opinion-forming? How is the internet changing due to the flood of AI-generated content? (BA & MA; Contact: Christopher Tille, Kirsten Pitz)
  #GenAI, #PoliticalOpinionFormation, #LiteratureBased
• How should AI-supported educational programs be designed to effectively promote democracy and prevent extremism? (BA & MA; Contact: Jana Ruß, Kirsten Pitz)
  #GenAILiteracy, #Democracy, #LiteratureBased
• Ethical Challenges of AI in a Social Context (e.g., greenwashing, extremism, transparency, etc.) (BA & MA; Contact: Christopher Tille, Kilian Züllig)
  #Ethics, #Literature-based
• Social Recommendation – Opportunities and Risks of Recommendation Systems in Social Networks (BA & MA; Contact: Jana Ruß, Kilian Züllig)
  #RecommenderSystems, #SocialMedia, #Literature-based

The content and methodology of the topics will be individually coordinated and specified with the students. The requirements for the topics vary depending on the degree program (Bachelor’s/Master’s). 

More and more people are taking charge of their own wealth accumulation. Yet despite this growing interest, young investors in particular often lack the financial literacy needed to build sustainable wealth. This trend is reflected on social media: so-called finfluencers now reach thousands of people with their content. On the one hand, they raise awareness of financial topics; on the other, they pose risks: They often oversimplify risks or provide insufficient information due to conflicts of interest. Since the phenomenon of finfluencers is still relatively new, it is important to understand and quantify the potential opportunities and dangers.

Possible Topics

• How do videos by finfluencers affect consumers’ financial literacy? (BA & MA; Contact: Jonathan Ibele)
  #Education, #SocialMedia, #Experiment
• Do viewers recognize manipulative elements (financial cues, social pressure) in finfluencer videos? (BA & MA; Contact: Jonathan Ibele)
  #Manipulation, #SocialMedia, #Experiment
• What effect does clearly labeling advertising have on the credibility of finfluencers’ messages? (BA & MA; Contact: Jonathan Ibele)
  #Marketing, #SocialMedia, #Experiment
• How do finfluencers address the risks of financial products in their videos? (BA & MA; Contact: Jonathan Ibele)
  #Risks, #GenAI, #Analytics, #Coding
• How can videos by finfluencers be automatically fact-checked? (BA & MA; Contact: Jonathan Ibele)
  #FactCheck, #GenAI, #Coding
• How can promotional content be identified in videos by finfluencers, and how do the characteristics of self-promotion differ from those of third-party advertising? (BA & MA; Contact: Jonathan Ibele) 
  #Marketing, #GenAI, #Coding

The topics will be individually coordinated and specified with the students in terms of content and methodology. The requirements for the topics vary depending on the degree level (Bachelor’s/Master’s). 

The Green AI research area explores how artificial intelligence can help address environmental challenges. In an environmental context, this involves optimizing energy and resource use, preventing environmental disasters, reducing greenhouse gas emissions, and promoting sustainable processes.

Possible Topics

• AI systems for reducing emissions or electricity consumption in the global supply chain, businesses, and residential buildings (BA & MA; Contact: Christopher Tille, Kilian Züllig)
  #ElectricityConsumption, #MachineLearning, #Coding, or #Literature-Based
• Use of AI to detect natural disasters and environmentally harmful activities (e.g., using satellite images) (BA & MA; Contact: Christopher Tille, Kilian Züllig)
  #NaturalDisasters, #MachineLearning, #Coding
• AI-supported sustainability reporting to improve preparation and auditing (BA & MA; Contact: Christopher Tille, Kilian Züllig)
  #SustainabilityReporting, #Analytics, #Coding, or #Literature-based
• Ethical dilemmas of AI in an ecological context (e.g., greenwashing, transparency, etc.) (BA & MA; Contact: Christopher Tille, Kilian Züllig)
  #Ethics, #Literature-based
• Energy Communities and Tenant-Generated Electricity. The Role of AI in Decentralized Energy Supply. (BA & MA; Contact: Kilian Züllig) 
  #P2P-Electricity-Trading, #Machine-Learning, #Coding or #Literature-based 

The content and methodology of the topics are determined and finalized in consultation with each student. The requirements for the topics vary depending on the degree program (Bachelor’s/Master’s). 

As a result of digitalization, organizations today have access to vast and ever-growing amounts of data (keyword: “Big Data”). However, empirical evidence shows that the data analyzed and used is often characterized by poor data quality—even in internal corporate customer databases, an average of approximately 30% of stored data values are incorrect. This results, for example, in annual additional costs of $15 million for the average American company. However, poor data quality is not only a major problem for businesses—in the age of “fake news”, the need for reliable information is also growing in politics and society. Therefore, quantitative methods are needed to measure, control, and improve data quality.

Data quality is also becoming increasingly important in the context of generative systems. In particular, Retrieval-Augmented Generation (RAG) systems, such as those underlying ChatGPT, place high demands on the quality of the underlying data. Data quality defects can arise both in the document corpus and in the retrieval process—for example, due to incomplete documents or imprecise user queries—and directly impact the quality of the generated responses. Consequently, there is a growing need for methods to detect and address data quality issues in RAG systems.

Possible topics:

Measuring and Improving Data Quality

• What isn’t measured isn’t managed: Developing data quality metrics (BA/MA; Contacts: Andreas Obermeier; Anna-Lena Kubillus)
• Data quality and its impact on AI-powered systems (BA/MA; Contacts: Andreas Obermeier; Hannah Knehr; Anna-Lena Kubillus)
• Data Quality in Wikis & Unstructured Data: Overview & Metrics of Individual Data Quality Dimensions (BA/MA; Contacts: Andreas Obermeier)
• Data Timeliness as a Key Factor in Data-Driven Decision Making (BA/MA; Contacts: Hannah Knehr; Anna-Lena Kubillus)

Data Uncertainty

• Data Uncertainty in Machine Learning: Incorporating Uncertainty from Data Quality Defects into Artificial Intelligence Methods (BA/MA; Contacts: Andreas Obermeier; Hannah Knehr; Anna-Lena Kubillus)

DQ in GenAI Applications

• Data Quality in GenAI Applications (e.g., RAG systems) (BA/MA; Contacts: Andreas Obermeier; Anna-Lena Kubillus; Mike Rothenhäusler)
• Impact of data quality defects on the performance of GenAI (e.g., RAG systems) and strategies for targeted improvement (BA/MA; Contacts: Andreas Obermeier; Anna-Lena Kubillus; Mike Rothenhäusler)
• Detection of data quality defects in retrieval-augmented generation systems (BA/MA; Contacts: Mike Rothenhäusler)
• Strategies for resolving data quality defects in retrieval-augmented generation systems (BA/MA; Contacts: Mike Rothenhäusler)

Explainable AI (XAI) is playing an increasingly significant role in our lives. From chatbots, spam filters, and shopping recommendations for individuals to fraud detection at financial institutions and marketing strategies for businesses—AI is already behind much of what we encounter. The growing prevalence of AI systems is opening up unprecedented possibilities. However, areas of application where critical decisions are made—such as in corporate controlling or when it comes to the creditworthiness of individuals—are under close scrutiny. The reason is the opacity of AI. In fact, studies show that many Europeans view decisions made by AI systems with unease. And this is true even when the algorithms demonstrably deliver better results than human experts. In critical application areas, it is important to understand how AI results are generated. This is where the research area of Explainable AI (XAI) comes in: the focus of research is on methods that make the results and functioning of AI systems understandable to human users.

Possible topics:

• Explainability for GenAI (BA/MA; Contact: Hannah Knehr, Chiara Schwenke)
• Explainability for detecting digital disinformation (BA/MA; Contact: Hannah Knehr, Chiara Schwenke)
• Development and implementation of Explainable AI (MA; Contact: Maximilian Buck, Hannah Knehr)
• New Approaches in Psychotherapy – Explainable AI in Mental Health (BA/MA; Contact: Hannah Knehr)
• User-Centric Explainable AI (BA; Contact: Maximilian Buck, Hannah Knehr, Chiara Schwenke)
• EU AI Act – What Laws Will Soon Require of XAI (BA/MA; Contact: Maximilian Buck)
• Fairness and Bias in AI Systems (BA/MA; Contact: Chiara Schwenke)
• Causal AI – Causality as the Key to Explainable AI (BA/MA; Contact: Maximilian Buck)
• Explainable AI for Integrating Human Expertise into AI Systems (BA/MA; Contact: Hannah Knehr, Maximilian Buck)
• Human-Centered AI Systems with Human-in-the-Loop (BA/MA; Contact: Hannah Knehr, Maximilian Buck)
• Artificial Intelligence in Controlling – Forecasts as a Basis for Decision-Making (BA/MA; Contact: Maximilian Buck, Lara Frost)
• Interactive Explanations for AI Users (BA/MA; Contact: Chiara Schwenke, Hannah Knehr)
• Teaching AI Literacy with Explainable AI (BA/MA; Contact: Chiara Schwenke)
• Explainable AI for Time Series Forecasting (MA; Contact: Lara Frost)
• Explainable AI for Time Series Classification (BA/MA; Contact: Lara Frost)

The world of work is facing major changes. Rapid technological advancements and societal upheavals call for solutions on how to shape the future of work in Baden-Württemberg. Digitalization, automation, and artificial intelligence are considered key drivers of future economic growth and will have a significant impact on the world of work. In addition, society’s transition toward climate neutrality is bringing about far-reaching changes, including a significant reduction in greenhouse gas emissions, the shift to a resource-efficient circular economy, and the implementation of the energy transition. In coping with these changes, one thing is particularly important: people who shape economic and social progress. That is why we must now build the skills necessary to shape a positive future for everyone—so-called future skills.

Possible topics:

• Extraction of skills from job postings (BA/MA; Contact: Chiara Schwenke)
• Creation of future skills lists from job postings using machine learning methods (MA; Contact: Chiara Schwenke)
• Automated detection of emerging skills using machine learning methods (BA/MA; Contact: Chiara Schwenke)
• Time series analyses of future skills (BA/MA; Contact: Chiara Schwenke)

The ability to process large volumes of data from various sources (big data) and the use of artificial intelligence can help companies fundamentally improve their methods for recruiting, developing, and retaining employees. Despite the availability of advanced analytical methods, however, HR decisions in many companies are still based on gut feelings and intuition. This is where People Analytics comes in: As a central element of future-proof HR strategies, People Analytics enables an evidence-based approach to human resources, in which technologies are used to analyze data on current, former, and future employees across all HR areas, with the goal of improving HR and overall company performance.

Possible topics:

• Innovative Applications of People Analytics: Case Studies and Best Practices (BA/MA; Contact: Chiara Schwenke)
• Artificial Intelligence in the Workplace: Shaping the Transformation (BA/MA; Contact: Chiara Schwenke)
• Quality of GenAI Use in the Workplace (BA/MA; Contact: Chiara Schwenke)
• Empirical Study of the Influence of People Analytics on Business Problems and Organizational Performance (BA/MA; Contact: Chiara Schwenke)
• AI Ethics in People Analytics (BA/MA; Contact: Chiara Schwenke)
• Organizational Prerequisites for the Successful Implementation of People Analytics (BA/MA; Contact: Chiara Schwenke)

Traditional sales processes often reach their limits when it comes to data availability and the ability to analyze and process data. In brick-and-mortar retail, employees generally cannot memorize all product and customer details. Even if this information were available to service experts (e.g., via a traditional IT system), processing it manually would be nearly impossible due to the large volume and complexity of the relevant data, or it would be distorted by subjective personal opinions. Recommender systems enable objective product recommendations to users based on large amounts of data. Robots like Pepper have the ability to perceive customers in the physical world, interact with them, and also possess the computing power to process the data obtained in this way more objectively, quickly, and accurately than humans.

Possible topics:

• Use of the humanoid robot Pepper in customer service (BA/MA; Contact: Andreas Obermeier)
• Explainable & Conversational Recommender Systems – Overview (BA; Contact: Andreas Obermeier)
• Natural Language Generation – Systems for automating customer service (BA/MA; Contact: Andreas Obermeier)

Modern information systems can not only generate economic value but also make a significant contribution to addressing social and societal problems and challenges. Our research on the social impact of information systems focuses on the pressing societal issues of our time, such as unemployment, the shortage of skilled workers, integration, and strengthening democracy. For example, our studies have shown that digital applications enhance the job-search efforts of young people, and that online peer groups (i.e., digital self-help groups) in particular prove to be valuable in numerous social contexts, such as unemployment under difficult conditions, unemployment among older adults, career guidance for young people, or the integration of refugees. The digital nature of modern information systems brings particular advantages, such as flexibility in terms of time and location, as well as the possibility of anonymity and thus a protected exchange. Artificial intelligence can serve as an additional catalyst for social innovation, for example, by enabling counseling to be scaled and personalized.

Possible topics:

• AI as an assistant and advisor for people (BA/MA; Contact: Maximilian Buck, Chiara Schwenke)
• Explanatory AI assistant to support job searches for people over 50 (in cooperation with a startup) (BA/MA; Contacts: Chiara Schwenke)
• Explainable AI assistant to support job searches for people with refugee backgrounds (in collaboration with an NGO) (BA/MA; Contacts: Chiara Schwenke)
• Explainable AI for learning about digital disinformation (BA/MA; Contacts: Hannah Knehr; Chiara Schwenke)

Competitive sports are undergoing a profound digital transformation. Today, training, competition, and movement data are being generated on a scale never seen before—supplemented by video analysis, sensor data, and textual reports. Despite the ever-increasing availability of data, its true potential for practical application remains largely untapped in many sports. Simply collecting data does not automatically lead to better decisions or improved performance—rather, there is often a lack of systematic analytical approaches that generate actionable insights from the flood of data. This is precisely where Sports Analytics comes in. Sports Analytics is dedicated to the systematic evaluation of these diverse data sources to make athletic and organizational decisions more informed. Through the use of modern methods from statistics, data science, and machine learning, patterns are revealed, performance is evaluated more objectively, and potential is identified early on. The focus is on developing data-driven models and intelligent decision-support systems that measurably improve athletic performance, strategic planning, and sustainable success.

Possible topics:

• Context-dependent evaluation of passes and space creation (BA/MA; Contact: Chiara Schwenke, Mike Rothenhäusler)
• Holistic evaluation of player actions based on tracking and event data (BA/MA; Contact: Chiara Schwenke, Mike Rothenhäusler)
• AI-powered decision support in scouting (BA/MA; Contact: Chiara Schwenke, Mike Rothenhäusler)
• Modeling individual player contributions to team success probability, taking into account network and interaction effects (BA/MA; Contact: Chiara Schwenke, Mike Rothenhäusler)
• Data-driven sports organizations: The potential of agent-based AI for analysis, planning, and decision-making (BA/MA; Contact: Chiara Schwenke, Mike Rothenhäusler)
• ML in sports – Is it possible to predict the results of sporting events using machine learning? (BA/MA; Contact: Chiara Schwenke)

Background: Private natural history collections play an important role in expanding and supplementing museum holdings. However, assessing their value prior to a potential acquisition poses significant challenges for museums: the material, scientific, and cultural value of such collections is often difficult to quantify. While simplified procedures are often used in practice, these can only inadequately reflect the actual scientific and historical value of the collection. At the same time, such collections offer significant opportunities: Time and again, it has been shown that collections acquired at relatively low cost later turn out to contain valuable objects, some of which are scientifically significant. A deeper understanding of valuation processes, their criteria, and decision-making logic is therefore important not only for museums but also for cultural policy and scientific institutions. The case study of a natural history museum offers a practical starting point for systematically examining and further developing valuation mechanisms.

Possible areas of focus:

• Analysis of existing valuation practices for natural history collections (e.g., market mechanisms, valuation guidelines, museum valuation processes)
• Case study-based examination of selected acquisition processes in natural history museums
• Development of a criteria or valuation framework that integrates material, scientific, and cultural values
• Reflection on the opportunities and risks of different evaluation approaches, particularly with regard to transparency, sustainability, and scientific value

The specific design (e.g., qualitative interviews, document analysis, framework development) will be determined in consultation with the supervisor.

Background: Agentic AI refers to AI systems with a high degree of autonomy and decision-making capability. These systems can prepare or autonomously implement strategic decisions; however, their role within organizations has yet to be thoroughly researched. As autonomy increases, new risks and challenges may arise—ranging from liability issues and loss of control to ethical dilemmas. Agentic AI can be particularly problematic for companies with limited AI expertise.

Possible Focus Areas:

• Systematic literature review on Agentic AI and its fields of application, as well as the development of a framework for its use
• Analysis of current research on the risks and challenges posed by Agentic AI, and the derivation of best practices for dealing with them

The specific details will then be developed in consultation with the supervisor.

Background: Economic turbulence has increased dramatically in recent years. In addition to climate change, the COVID-19 pandemic, and geopolitical shocks such as Russia’s war in Ukraine, a long-term decoupling of the global economy—known as “decoupling”—is becoming evident. As a highly interconnected, export-oriented economy, German industry in particular faces a multitude of challenges across various sectors.

Possible Focus Areas:

• Development of a method for analyzing various stress factors and creation of a cause-and-effect model to assess the impact of crises on SMEs
• Definition of key performance and control metrics for companies under stress and creation of a KPI driver tree to identify early warning indicators
• Identification, analysis, and processing of freely accessible data sources for the early detection of external disruptive factors in corporate networks
• Evaluation and development of incentive systems to increase crisis resilience

The specific details will then be developed in consultation with the supervisor.

Introductory reading:

• Lamorgese, A., Patnaik, M., Linarello, A., & Schivardi, F. (2024). Management practices and resilience to shocks: Evidence from COVID-19. Management Science, 70(12), 9058-9072.
• Hayne, C. (2022). The effect of discontinuous and unpredictable environmental change on management accounting during organizational crisis: A field study. Contemporary Accounting Research, 39(3), 1758-1796.
• Hyun, J. H., Matejka, M., Oh, P., & Ahn, T. S. (2022). Performance targets and ex post incentive plan adjustments. Contemporary Accounting Research, 39(2), 863-892.