We currently offer bachelor and master theses in the following research domains:

LLM and Explainable Smart Environment

We are currently mainly focusing on leveraging LLMs within our Explainable Smart Environment research. 

The possible thesis projects include, but are not limited to, the following topics:

• Leveraging LLMs for Explainable Smart Environments: How effectively can LLMs generate context-aware, personalized explanations for smart-home behaviors, and how does this impact user understanding, trust, and task performance?
• LM-Driven Detection of Explanation Needs in Smart Homes: Can LLMs infer when a user is confused, uncertain, or surprised, and proactively trigger explanations? 

Hallucination in LLMs

We explore the conceptual foundations and empirical behavior of hallucinations in Large Language Models, focusing on how human factors, prompt characteristics, and interface design can systematically influence the reliability and perceived credibility of AI-generated content. This research direction examines hallucination not just as an accuracy problem, but as a socio-technical phenomenon shaped by user psychology, framing, and system design.

Possible research questions include, but are not limited to:

• Studying how different prompt styles (e.g., assertive vs. neutral, supportive vs. confrontational, or framed through different social roles such as parent, government, spouse) influence the way LLMs generate information, justify claims, or unintentionally drift into hallucinations
• Investigating how user mood, emotional framing, or affective cues shape the type, style, and confidence of LLM responses, including conditions under which hallucinations become more likely or more persuasive.
• Do transparency features such as source linking or chain-of-thought reasoning actually reduce hallucination—or do they increase the illusion of factuality for users, making hallucinations more convincing?

Empirical SE

We are currently focusing on the empirical evaluation of explanation types in interactive smart environments.

A possible thesis project can include investigating how different explanation styles, such as counterfactual, contrastive, and causal explanations, affect user understanding, trust, satisfaction, and task performance in smart environments. The goal is to identify which explanation type is best suited for everyday users interacting with intelligent systems and to understand how individual differences (e.g., cognitive style, background, demographics) influence explanation preference and effectiveness.

LLM and SE

We are currently investigating the role of LLM as tools and their potential misconceptions in modern software engineering practice.

A possible thesis project can examine whether Large Language Models can meaningfully support or even replace aspects of software engineering, particularly for domain experts who lack formal programming training. Many experts (e.g., physicists, clinicians) rely increasingly on LLMs to generate complex data-analysis code, but their ability to test, validate, and assess the reliability of such code remains uncertain. We investigate how well LLM-generated code aligns with established software engineering processes, requirements elicitation, systematic testing, verification, debugging, and where illusions of competence or hidden risks emerge in real-world use.

How to apply?

If you are interested in writing a thesis, please consider the following points:

1. If you already have an idea for a topic of your interest: Note that we prefer theses that are closely related to our research areas (see above). In any case, you are welcome to send an e-mail (German or English) to us with
   1. a brief overview of the specialization you are interested in,
   2. a reason why you consider mentoring by SSE to be particularly promising, and
   3. a short curriculum vitae, including a recent overview of your grades.
2. If you have already participated in courses of our chair, but have not yet found a specific topic that interests you: Please do not hesitate to send an e-mail (German or English) to us with
   1. A list of the three courses you have enjoyed the most in your course of study so far,
   2. a scientific publication that you find interesting and that has something to do with our research interests, and
   3. a short curriculum vitae, including a recent overview of your grades.