Machine Learning for Managing Structured and Semi-Structured Data
MCML Authors
Max Berrendorf
Dr.
* Former Member
Abstract
Max Berrendorf
Dr.
* Former Member
Abstract
As data availability grows across sectors, machine learning, especially graph neural networks, plays a crucial role in extracting insights by automating complex analysis, including relational learning. Knowledge graphs help store entity facts, though they often require automated methods like Link Prediction and Entity Alignment to fill in missing information due to the sheer volume. This thesis advances knowledge graph completion by improving Entity Alignment through active learning, refining Link Prediction with metadata, and introducing a new evaluation metric, as well as a software library to aid researchers. (Shortened).
BibTeXKey: Ber22