Open Topics
The Workflow Systems and Technologies Group offers topics for bachelor and master theses as well as Master Praktikum. The following list contains a number of current suggestions for topics.
To discuss one of those or any other topic in this research field, please contact the particulary named supervisor.
» Jump to Bachelor thesis topics
» Jump to Master thesis topics
This list will be updated continuously!
Master thesis topics
Supervision Univ.-Prof. Han van der Aa, Ph.D.
Application instructions: When applying for projects with Prof. Van der Aa, please send an e-mail including a transcript of records and short CV in your e-mail. Beyond the listed topics, you are also welcome to propose your own topic or indicate general directions that interests you when applying. Note that, unless stated otherwise, all listed research directions can be scoped to be suitable for P1, P2, and master projects. Ideally you work on the same direction from P1, through P2, up to your master project.
P1-P2-Master topic
Currently there are no projects available for the 2025 Winter Semester.
Bachelor thesis topics
Supervision Univ.-Prof. Dr. Erich Schikuta
Modelling Football Possessions and Tactics with Machine Learning
Goal
Football matches involve complex factors such as player interactions, team formations, offensive and defensive performance, and overall game dynamics, all of which can influence the match situation and result. The aim of this study is to use machine learning methods (e.g., neural networks) to integrate these different aspects into one representation model to predict the Expected Possession Value (EPV)/ team tactical patterns. EPV is a metric that estimates the likelihood of a team scoring from its current possession. The necessary tracking data (all positions of the players and the ball during a match) and event data (passes, shots, tackles, ...) from football matches will be provided.
Requirements
- Python
- Basic knowledge of machine learning
- Interest and basic knowledge of football (soccer) is a plus
Number of Students: 1-2
Co-Supervision:
- Jonas Bischofberger jonas.bischofberger@univie.ac.at,
- Runqing Ma runqing.ma@univie.ac.at
Resources
Sahasrabudhe, A., & Bekkers, J. (2023). A graph neural network deep-dive into successful counterattacks. In Proceedings of the 17th Annual MIT SLOAN Sports Analytics Conference.
Anzer, G., Bauer, P., Brefeld, U., & Faßmeyer, D. (2022, March). Detection of tactical patterns using semi-supervised graph neural networks. In 16th MIT sloan sports analytics conference (pp. 1-15).
Lee, J., Park, E., & del Pobil, A. P. (2025). We know who wins: graph-oriented approaches of passing networks for predictive football match outcomes. Journal of Big Data, 12(1), 147.
WineCellar As A Service (cross-platform) App
Goal:
The goal of this project is to design and develop an information system for the administration and the management of a wine cellar.
Hereby, a thorough market analysis and comparison of available competing systems (also old Bachelor's theses) have to be done. Based on these results and customer interviews, a list of specific requirements has to be defined. Specific emphasis has to be laid on the system architecture allowing for a cloud-based approach. The functionality of the application to be developed has to cover all important workflows for wine cellar management, as wine inventory and cellar location management, wine purchase, personal wine information handling, wine web resource (information and purchase options) crawling, etc. Specific focus has to be laid on usability and responsive design.
IMPORTANT: A functional requirement is the development of a (possible cross-platform) app running on PC (Windows 11) and Smartphone (Android).
Specific functionality will be part of the requirements analysis.
Required Competences (mandatory technology stack):
- SQLite
- Cloud service (e.g. u:cloud)
- The technology stack (specifically the cross platform development framework, e.g. Flutter, Kotlin, etc.) has to be decided and is up to the student. Personal experience in an existing framework is a plus.
Number of Students: 1
Resources:
Old Bachelor theses on the same topic with different technology stacks.
Extension of Smart Media Presenter 2.0
Goal:
The Smart Media Presenter 2.0 is a specialized tool for the presentation of media resources. It allows for the creation, management, and running of presentations of images and videos. The tool was developed as part of two Bachelor's theses.
The goal of this project is to extend the existing tool with an AI-based open-source photo enhancement module and improve some basic functionalities.
The AI-photo enhancer codebase is free to choose; however, a possibility could be Upscaly (https://upscayl.org/) with its codebase (https://github.com/upscayl/upscayl)
Any further improvement of the code base is welcomed.
Required Competences:
- Typescript/Javascript, Electron, React.js, MaterialUI and Firebase
Number of Students: 1
Resources:
Version 1.0 (original): Lukas Jäger
Thesis: https://ucloud.univie.ac.at/index.php/s/4xylY3NL38ibQMp),
Codebase https://github.com/lksjgr/SmartMediaPresenter
Version 2.0 (actual): Wsewolod Golubkow
Thesis und Codebase: https://github.com/blubcow/SmartMediaPresenter
Video analysis of Dancesport clips
Goal
Competitive dancesport is an acyclic team sport (couple, formation teams) with a strong technical focus besides physical demands. The winner of a dancesport competition (tournament) is decided by agglomeration of the subjective opinion of a group of human judges. Goal of this work is the first step towards a more objective decision process by introducing an AI-based judge. Hence, this work is strongly research orientated. Starting point of this research endeavour is the identification of sportive dance objects/activities/characteristics by video analysis.
The research approach of this work uses YOLO Real-Time Object Detection and training of dance sport actions using existing dance videos/photos.
As a motivation a first and very simple detection approach is illustrated by two video clips, on the one hand an original TV video scene https://ucloud.univie.ac.at/index.php/s/z9jFL98QmoCzkBg, and on the other hand the same video scene after a detection process performed by YOLO11 https://ucloud.univie.ac.at/index.php/s/2H9WjPqCnGWBimx.
The following tasks have to be performed as part of the thesis:
- Creation of a collection of dance video data by categorization (couple, formation, latin, ballroom, video position).
- Development/Adaptation of a GUI for specific YOLO dancesport training/detection tasks (e.g. pure Python, YOLOSHOW, OpenCV GUI, …).
- Training of YOLO modules for object identification of “female dancer”, “male dancer” and “couples”.
Further some questions have to be answered or at least recommendation have to be given:
- Which YOLO version and which model to be used
- Which quality of video necessary/recommended
- Which supported YOLO computer vision Tasks useful
- Analysis of the best video position
- Center from above
- Corner from angle (common for TV broadcasting)
- Center from angle (common for TV broadcasting)
- Identification of ballroom vs latin dancing
- Identification of dances (couple/team)
Required Competences:
- Python and its ecosystem
- Tie to competitive dancesport is surely a plus.
Number of Students: 1
Resources:
- YOLO Info Ultralytics https://www.ultralytics.com/
- YOLO11 Software https://github.com/ultralytics/ultralytics
Topic Individual
It is also possible to submit individual topics. Doing so, please submit an abstract to your supervisor via email. If the topic fits the requirements for a bachelor thesis, the scope of the work is agreed with the supervisor at the beginning.
Supervisor: Erich SCHIKUTA, erich.schikuta@univie.ac.at
Supervision and thesis in German or English
Supervision Marian Lux
Topic: AI 119251– Improvement of a Multi-Agent approach which uses Large Language Models (LLMs) for responses by incorporating Retrieval-Augmented Generation (RAG) to consider own data in queries
Goals:
- Improve an agentic architecture where each agent has a particular role to improve answers on questions by using one of the latest local LLMs together with the RAG approach [1].
- Improve the existing pipeline and workflow of the chatbot [2]
- For questions and answers a simple chatbot is already implemented by using the Telegram UI [3] where you will improve the user interface of the Telegram bot (buttons, structured output etc.) or even implement a new UI, e.g., Streamlit [4]
- Data for RAG with the agentic architecture (i.e., multiple ReAct[5] agents) is extracted from web URLs (defined during runtime) or documents (uploaded during runtime). The use case of the multi agent approach, using CrewAI [6] is currently developed for a blogger:
- Research alternative state-of-the-art approaches and evaluate if the current implementation is already the best
- Make the code flexible that it works with different configs and questions on different use cases/domains, e.g., a lawyer.
- Implement function calls
- Bugfixes
The code is hosted public on GitHub.
Recommended requirements:
Implementation in Python
Access to computer hardware which can run local 8B LLMs
Interest in building LLM AI solutions
Supervisor:
Dr. Marian LUX - marian.lux@univie.ac.at
Supervision and thesis in German or English
References:
[1] Lewis, P., Perez, E., Piktus, A., Petroni, F., Karpukhin, V., Goyal, N., ... & Kiela, D. (2020). Retrieval-augmented generation for knowledge-intensive nlp tasks. Advances in Neural Information Processing Systems, 33, 9459-9474.
[2] https://github.com/MLUX-University-of-Vienna/CrewAI-Multi-Agent-Blog-Chatbot_WS2024
[3] https://core.telegram.org/bots/api
[5] Yao, S., Zhao, J., Yu, D., Du, N., Shafran, I., Narasimhan, K., & Cao, Y. (2022). React: Synergizing reasoning and acting in language models. arXiv preprint arXiv:2210.03629.
[6] https://docs.crewai.com/introduction
Topic: AI 119252 – Using a Multimodal Large Language Model (LMM) to improve Retrieval-Augmented Generation (RAG) by answering questions on custom data
Goal:
Develop an agentic architecture which enables to ask questions about PDFs or from web content, containing text, images and tables. The agentic architecture uses modern local hosted LLMs/LMMs like, Llama [1]/Quen [4]/ LLaVA[6-7] and the RAG approach [2].
For questions and answers a simple chatbot should be implemented by using a UI framework (e.g., Streamlit [3]) or an existing chat UI (e.g., Telegram UI [5])
The data for RAG, including images like diagrams or even tables, is extracted from web URLs or documents.
Automated tests with LLM inferencing evaluate the approach.
The code and whole documentation will be hosted public on GitHub as open source.
Recommended requirements:
Implementation in Python
Access to computer hardware which can run local 8B LLMs
Interest in building LLM AI solutions
Supervisor:
Dr. Marian LUX - marian.lux@univie.ac.at
Supervision and thesis in German or English
References:
[1] https://ai.meta.com/llama/
[2] Lewis, P., Perez, E., Piktus, A., Petroni, F., Karpukhin, V., Goyal, N., ... & Kiela, D. (2020). Retrieval-augmented generation for knowledge-intensive nlp tasks. Advances in Neural Information Processing Systems, 33, 9459-9474.
[4] https://qwenlm.github.io/blog/qwen2.5-vl/
[5] https://core.telegram.org/bots/api
[6] https://llava-vl.github.io/
[7] Liu, H., Li, C., Li, Y., & Lee, Y. J. (2024). Improved baselines with visual instruction tuning. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 26296-26306).
Topic: AI 119253 – Developing a ReAct inspired Agent from scratch for utilizing on Large Language Models (LLMs).
Goal:
Develop a ReAct (Reason + Act) inspired Agent [1],[2] from scratch by considering splitting a query into sub tasks and by providing a pool of tools to choose from to answer the query in a minimum set of iterations. (cf., image “How do autonomous agents work?” in [7]
The agent works with local open source LLMs (max. 8b) where it is recommended to use the Ollama [3][6]. Furthermore, the implementation must be easy to extend and maintain.
For evaluation of the agent, provide at least the following tools
- Vector search for documents of a particular topic
- Web search for current data
- Wikipedia Search for Encyclopedia
- Calculator for mathematical expressions
on a simple chatbot implementation (where an existing one can be used as well).
The code is open-source and will be hosted public on GitHub. A PyPI repository is expected. Therefore, the agent is used like SOTA ReAct agent frameworks, e.g., LlamaIndex[4] or LangChain [5].
Recommended requirements:
Implementation in Python
Interest in building LLM AI solutions
Supervisor:
Dr. Marian LUX – marian.lux@univie.ac.at
Supervision and thesis in German or English
References:
[1] Yao, S., Zhao, J., Yu, D., Du, N., Shafran, I., Narasimhan, K., & Cao, Y. (2023, January). React: Synergizing reasoning and acting in language models. In International Conference on Learning Representations (ICLR).
[2] ReAct Prompting: https://www.promptingguide.ai/techniques/react
[3] Ollama: https://ollama.com/
[4] LlamaIndex ReAct Agent: https://docs.llamaindex.ai/en/stable/examples/agent/react_agent/
[5] LangChain ReAct Agent: https://python.langchain.com/v0.1/docs/modules/agents/agent_types/react/
[6] Ollama Python Library: https://github.com/ollama/ollama-python
[7] Image: How do autonomous agents work? https://medium.com/design-bootcamp/a-comprehensive-and-hands-on-guide-to-autonomous-agents-with-gpt-b58d54724d50
Topic: Process Mining – Process Discovery Visualization Tool in Python
Goal:
Improving/maintaining and providing new features to the already existing process mining[6] tool from previous bachelor theses [1],[5].
Different topics based on this process mining tool are possible. Some of them are listed below and the scope of the work is agreed with the supervisor at the beginning. Please write an email with your preferences to the supervisor:
- Bug fixing and improvements of the existing solution [10]
- Incorporating/merging code from other code branches which were developed from prior bachelor projects
- Development of more miners and variants
- Improving the existing algorithms with additional visualization
- Colors for Nodes and Edges based on their importance
- Sophisticated graphics in detail views (edges/nodes)
- Displaying the happy path
- Filtering of nodes does not change the whole graph (arrangement of all nodes)
- Decision tool: Which process mining algorithm fits best for a particular use case or data set [2],[6],[7]
- Integration of a local open source LLM which helps with the decision.
- Questionnaire based for use case in combination with data analysis
- Development of an Object Centric Process Mining integration [8], [9]
- Improving the UI
- XES importer and exporter tool [6]
- Display (most popular) variants
- Improvement of the File importer
- Filter methods
- Search functionalities
- Save project (not as pickle because changes in classes and newer versions make them unusable, but as data transfer object - DTO)
The whole work is open source, also the contribution during the bachelor thesis. This means that the code must be easily maintainable and extendable. The written code should contain unit tests as well.
The code will be hosted public on GitHub.
Because of the different topics, multiple students may work simultaneously on the same code base. This implies that at the end of the thesis, all changes must be merged into the main project (GitHub).
Recommended requirements:
Implementation in Python
Only basic frameworks (numpy, scikit-learn etc.) can be used. For other more sophisticated frameworks, a permission from supervisor is mandatory.
Supervisor:
Dr. Marian LUX - marian.lux@univie.ac.at
Supervision and thesis in German or English
References:
[5] https://github.com/MLUX-University-of-Vienna?tab=repositories
[6] Van Der Aalst, W., & van der Aalst, W. (2016). Data science in action (pp. 3-23). Springer Berlin Heidelberg.
[7] https://research.aimultiple.com/process-mining-algorithms/
[8] https://www.ocpm.info/ocel_demo.html
[9] https://encyclopedia.pub/video/video_detail/785
[10] https://github.com/fabianf00/ProcessMiningVisualization_WS23/issues/41
Topic: Individual
It is also possible to submit individual topics. Doing so, please submit an abstract to your supervisor via email. If the topic fits the requirements for a bachelor thesis, the scope of the work is agreed with the supervisor at the beginning.
Supervisor:
Dr. Marian LUX - marian.lux@univie.ac.at
Supervision and thesis in German or English