Motivation 

As a media organization committed to public-service innovation, ZDF is continuously exploring new methods to evaluate and improve its recommendation system. Traditional A/B testing can be costly and limited in scope, especially in real-world environments where control over conditions is restricted. In collaboration with TUM-DI-LAB, we aim to develop a simulation-based evaluation framework that enables fast and cost-effective testing of recommendation strategies under controlled assumptions. Goals This project seeks to build a lightweight, modular simulation environment where students can model user behavior and test various recommendation algorithms. The goal is to enable rapid experimentation without affecting live systems. 

As part of the project team, you will: 

1. Data Familiarization: Gain an understanding of our data and infrastructure.
2. Model User Behavior: Simulate typical interaction patterns. You will work with Python or browser-based frameworks to create the simulation backend.
3. Test Recommendation Strategies: Evaluate the impact of different algorithms in a sandbox setup. Track key metrics such as click-through rate, diversity, and viewing volume. 

The simulation playground will offer ZDF editors and product teams a new way to experiment with recommendation logics—quickly, safely, and transparently. 

Requirements 

We are looking for students with the following skills and interests: 

• Problem-Solving & Critical Thinking: Ability to deconstruct complex problems and develop creative solutions. Ideally with basic knowledge of recommendation algorithms or user behavior modeling.
• Programming & Simulation: Proficiency in Python and interest in modeling complex systems.
• Curiosity & Enthusiasm: A passion for learning, innovation, and contributing to public-service media. Join us to shape the future of public-service media and contribute to a more informed world. 

Apply to this project here