Masterarbeit
Automated Extraction of Reproducibility Information for Climate Action Research
Research Area
Intelligent Information Management
Advisers
Dr. Sheeba Samuel
Description
Climate action research is a rapidly growing field within Climate Science, producing a large volume of publications addressing environmental monitoring, sustainability, and mitigation strategies. However, important information about methodologies, datasets, and computational workflows is often scattered across publications and supplementary materials, making it difficult to assess and reproduce scientific results. Ensuring reproducibility is particularly critical in climate research, where policy decisions and long-term environmental strategies rely on reliable and transparent findings. The objective of this thesis is to design and implement a system for constructing a Knowledge Graph that captures key entities and relationships in climate action research, with a particular focus on reproducibility-related information. Using techniques from AI and Natural Language Processing, the system will automatically extract metadata from scientific publications, such as datasets used, models applied, code availability, and experimental setups. This information will be structured and linked within a knowledge graph to enable querying, analysis, and identification of reproducible research practices. The thesis will result in a prototype framework that integrates publication mining, AI-based metadata extraction, and knowledge graph construction. The system will be evaluated on a selected corpus of climate research papers to demonstrate its ability to support reproducibility analysis and improve the discoverability of research artifacts. This work aims to contribute to open science by enabling better transparency, reuse, and validation of climate-related research.
