I am currently based in the ADAPT Centre in Trinity College Dublin, in the final year of my PhD programme. My research work centres on Knowledge Graphs (see a Wikiepdia summary
here) and knowledge graph embeddings (see a Wikiepdia summary
here). The core of my work is in determining how knowledge graph structure (measured in terms of the degrees and frequency of nodes and edges) affects knowledge graph embedding models and link prediction.
You can find my research publications and blogs here:
- Peer-Reviewed Research: here
- Research Blogs: here
As a part of my research, I have created two open-source, user-centric libraries to allow others to reproduce and build upon my work. These are:
- TWIG, a system that allow pre-hoc prediction of the performance of knoweldge graph embeddigs models, as well as the optimal hyperparameters to use with them, based on graph structural features. You can find the TWIG repo here: https://github.com/Jeffrey-Sardina/TWIG-TWM-dev
- TWIG-I, a novel link prediction paradigm that uses graph structural features alone (in the absence of embeddings) to perform link prediction. It further allows for cross-graph and cross-domain transfer learning of link preciction, and to my knowledge if the first link prediction model to be able to do so. You can find the TWIG-I repo here: https://github.com/Jeffrey-Sardina/TWIG-I
In addition, here are a few general links that you may fine to be of use: