1st ERCIS Master Thesis published!

We are delighted to have Marten Jostmann as our first author publishing his Master Thesis in our “ERCIS Master Theses Series” 🥳

Marten wrote his thesis at the Department of Information Systems of the University of Münster, supervised by Prof. Dr. Herbert Kuchen and co-supervised by Dr. Hendrik Winkelmann.

The thesis titled “Evaluation of Hypothetical Document and Query Embeddings for Information Retrieval Enhancements in the Context of Diverse User Queries” has been laudated by Herbert Kuchen as follows:

“Marten Jostmann has developed two approaches to search for unstructured documents in company-owned information repositories, based on the use of Large Language Models. He evaluated them intensively in terms of effectiveness, robustness against query reformulations, efficiency, and costs. He used the well-known and popular classic method BM25 as the basis for a comparison. In experiments with two extensive sample repositories, the two methods proved to be clearly superior. Mr. Jostmann has excellently solved the challenging task mentioned above and demonstrated a further, initially surprising application of the currently hotly debated large language models.”

If you would like to deep-dive into the thesis, check it out here:

https://doi.org/10.17879/25968508220 – the full text is available there!