DBIS

We are looking for a highly motivated master student to work on an innovative project for their master’s thesis as soon as possible. The project involves the implementation and evaluation of a programmable money approach utilizing verifiable credentials and zero knowledge proofs.

The exponential growth of research data across diverse scientific disciplines necessitates the development of efficient mechanisms for organizing, discovering, and sharing data. Metadata standards play a crucial role in this process by providing structured information that facilitates data interoperability and reusability. However, reasons such as lack of training on research data management, the inherent interdisciplinarity of the use case, and the dynamic developments all lead to a fragmented landscape for researchers to navigate, making it challenging for them to identify the most appropriate standard for their specific data. This thesis addresses this issue by implementing a Metadata Standard Recommender (MSR) utilizing Large Language Models (LLMs) with Retrieval-Augmented Generation (RAG).

Our approach leverages the capabilities of LLMs to understand and process natural language descriptions of research data, combined with RAG to enhance the accuracy and relevance of recommendations by integrating domain-specific knowledge. The MSR is designed to assist researchers in selecting metadata standards that best align with their data characteristics and research objectives, thus improving data management practices and fostering data sharing and collaboration.

The relevance of this work lies in its potential to streamline the metadata standard selection process, reducing the cognitive load on researchers and promoting the adoption of best practices in data documentation. By providing personalized recommendations, the MSR enhances the accessibility and usability of research data, contributing to more efficient data-driven discovery and innovation.

This expected outcome of this thesis is an open-source MSR, and to present with it a detailed analysis of the LLM and RAG methodologies employed, an evaluation of the performance of the recommender system through user studies and machine-actionable evaluation methods, and a discussion of its implications for the broader research community.