BSc in Computer Science, 110/110 with Honours

My final thesis focused on the critical issue of reproducibility in recommender systems. The core of the project was an in-depth study and alignment of evaluation metrics to ensure that experiments could be reliably replicated and compared.

This research was applied directly to ClayRS, an open-source Python framework designed specifically for the reproducible offline evaluation of recommender systems.

Key contributions of the thesis include:

• A thorough analysis of discrepancies in metric implementations between different popular frameworks, such as ClayRS and Elliot.
• The development and implementation of new features within ClayRS to import and directly compare recommendations from external frameworks.
• This work ultimately enhanced the framework’s interoperability and provided a more robust methodology for conducting fair and reproducible experiments in the field.