Roxana Pop

Education

PhD (Artificial Intelligence), University of Oslo 2021 -         
      Researching hybrid AI methods for temporal prediction

Master (AI and Computer Vision), Technical University of Cluj-Napoca 2018 - 2020
      Thesis on graph-based AMR parsing

Bachelor (Computer Science), Technical University of Cluj-Napoca 2014 - 2018
      Thesis on transition-based AMR parsing

Teaching

Teaching Assistant (Technical University of Cluj-Napoca) 2018 - 2021
   ○ Fundamental Algorithms laboratory (2018-2019 and 2019-2020 academic years)
   ○ Computer Programming laboratory (2019-2020 academic year)
   ○ Logic Programming laboratory (2020-2021 academic year)

Industry experience

Google (Software Engineering Intern) 2020
       Worked on semantic parsing on CFQ in Flax
         ○ implemented two parsing systems on CFQ (Natural Language ➔ SPARQL)
         ○ code available on Github

Cloudflight (Software Developer) 2018 - 2020
       Worked on a large scale project for the automotive industry

Selected publications

See my publications on dblp .

Master Projects

• Dynamic link prediction and time prediction on temporal Knowledge Graphs
• Comparing trainability of Graph Neural Networks for logic-expressible functions

Research Interests

Currently in my PhD I am investigating neural methods for temporal prediction on structured data, with an emphasis on graph neural networks extensions. My overarching objective is to combine neural methods with symbolic AI approaches, and extract logical rules from the trained neural networks for making the predictions. In this way, the predictions can be understood by human experts, but the extraction is data-driven, making the best of the two AI paradigms.

Before my PhD I did research in Natural Language Processing, mostly in AMR (Abstract Meaning Representation) parsing - a type of semantic parsing.