Ralph Abboud

Bio. I completed a D.Phil in Computer Science from the University of Oxford under the supervision of Dr. İsmail İlkan Ceylan and Prof. Thomas Lukasiewicz in 2022. My main research area is graph representation learning (GRL). In particular, I am interested in studying the strengths and limitations of GRL models, such as shallow embedding models and graph neural networks (GNNs), and proposing machine learning approaches with improved inductive capacity, interpretability, and representation power over relational data.

Current Work. I am currently a Science Associate within the Learning Engineering program at Schmidt Futures, working on applying large language models (LLMs) and structure-oriented machine learning approaches (e.g., GRL techniques) to improve middle school math education.

Contact. You can reach out to me via my work e-mail, rabboud {at}, or via ralph {at} You can also find my old Oxford departmental webpage here.


Selected Publications

Shortest Path Networks for Graph Property Prediction, LoG 2022.
Ralph Abboud , Radoslav Dimitrov , İsmail İlkan Ceylan
Approximate Weighted Model Integration on DNF Structures, AIJ 2022.
Ralph Abboud , İsmail İlkan Ceylan , Radoslav Dimitrov
The Surprising Power of Graph Neural Networks with Random Node Initialization, IJCAI 2021.
Ralph Abboud , İsmail İlkan Ceylan , Martin Grohe , Thomas Lukasiewicz
BoxE: A Box Embedding Model for Knowledge Base Completion, NeurIPS 2020.
Ralph Abboud , İsmail İlkan Ceylan , Thomas Lukasiewicz , Tommaso Salvatori
For a complete list of my publications, please check my DBLP and Google Scholar profiles.


D.Phil in Computer Science
2018 - 2022
University of Oxford
Thesis Title: Learning and Inference over Relational Data
M.Sc. in Computer Science
2017 - 2018
University of Oxford
B.E. in Computer Engineering
2013 - 2017
Lebanese American University (LAU) , Minor: Mathematics