Jeffrey Sardina, a PhD student at Trinity College Dublin and part of the ADAPT Centre, presented his paper, A Survey on Knowledge Graph Structure and Knowledge Graph Embeddings, at the 19th IEEE International Conference on Semantic Computing. The paper, co-authored with ADAPT Director Prof. John D. Kelleher and Prof. Declan O’Sullivan, was also nominated for Best Paper at the conference.
The study explores the relationship between Knowledge Graphs (KGs) and their machine learning counterpart, Knowledge Graph Embedding Models (KGEMs), which are widely used for predicting new facts based on existing data. While KGEMs have shown significant success, their performance and potential biases are heavily influenced by KG structure. Sardina’s research provides the first comprehensive survey on this topic, analysing existing literature to identify key patterns and relationships. The findings highlight the impact of graph structure on model performance and potential biases, aiming to inspire further research in this area.
Access the paper here.
