Master Thesis on Large Language Models for Patent Retrieval - remote possible

Patent retrieval is a highly important task for patent professionals in many stages of the patent lifecycle. Current retrieval systems use term frequencies or neural embeddings to retrieve relevant prior art from a large index. LLMs could function as powerful rerankers and improve retrieval results.

• Your Master thesis aims to assess the efficacy of LLM-based rerankers for patent retrieval.
• Furthermore, you will research state-of-the-art methods from the literature and analyze their performance in the context of patent data.
• In addition, you will implement computational experiments to identify advantages over prior methods, as well as areas for improvement and develop novel approaches to address identified limitations.
• If interested, you can participate in publishing the work at a major NLP conference or workshop.

• Education: Master studies in the field of AI, Computer Science, Software Engineering or comparable with very good grades
• Experience and Knowledge: profound knowledge of Machine Learning, Large Language Models, programming in Python and PyTorch; experience with Information Retrieval, Linux and LaTeX; prior knowledge of patents is an advantage
• Personality and Working Practice: you excel at working autonomously, systematically organizing your tasks, and applying analytical thinking to solve complex problems
• Work Routine: mobile working within Germany
• Languages: proficient in German or English