Short intro

Hi! My name is Roxana Pop and I am a PhD student at the University of Oslo under the supervision of Egor Kostylev.

Research Interests

I’m interested in:

• Neurosymbolic AI
• Temporal Knowledge Graphs
• (Temporal) Graph Learning
• DatalogMTL

My PhD objective is to extract temporal logical rules (DatalogMTL) from trained neural networks operating on Interval-based Temporal Knowledge Graphs (ITKGs). For more information, please see my recent short papers where I:

• introduce the inductive future time prediction task on ITKGs and draft a neurosymbolic architecture for it (paper)
• design a simple neural architecture for time prediction on ITKGs and describe how benchmarks can be constructed for the task (paper)

Prior to the start of my current position I did research in the area of Natural Language Processing. I worked on AMR parsing and CFQ parsing.