Sheeba Samuel
Senior Researcher and Lecturer · TU Chemnitz, Germany
I am a computer scientist working at the intersection of Semantic Web, Knowledge Graphs, and Data Science. My research focuses on enhancing the reproducibility, explainability, and interoperability of scientific experiments and machine learning models — leveraging provenance, linked data, and knowledge graphs across interdisciplinary domains including biomedicine, biodiversity, and data science.
Research Themes
Reproducibility
Methodologies and tools to ensure scientific studies and ML/DL models can be reliably reproduced and validated.
Knowledge Graphs
Semantic Web technologies, ontology engineering, and LLM-supported knowledge graph construction for science.
Data Provenance
Capturing and representing provenance across scientific experiments, ML pipelines, and computational notebooks.
Explainability (XAI)
Integrating domain knowledge to interpret and explain deep learning model predictions in life sciences.
News
- Apr 2026 New preprint: Containing the Reproducibility Gap: Automated Repository-Level Containerization for Scholarly Jupyter Notebooks
- Jan 2026 New preprint: Learning to be Reproducible: Custom Loss Design for Robust Neural Networks
- 2025 TU Chemnitz joined Jupyter4NFDI Integration Phase (2025–2027) as partner institution.
- 2025 Paper published in PeerJ Computer Science Multi-LLM information retrieval pipeline for extracting deep learning methodologies in biodiversity research
- 2025 Proceedings Co-Chair at Mensch und Computer 2025
- 2025 Two papers accepted at ICWE 2025: ShEx2SPARQL and ResearchFlow
- 2025 Co-organizer of the Workshop on Data Engineering for Data Science (DE4DS) at BTW 2025
- 2024 Paper published in GigaScience: Computational reproducibility of Jupyter notebooks from biomedical publications