Hello, world!

I am a PhD candidate at the University of Edinburgh and a member of the CDT in Data Science. I am supervised by Walid Magdy, Bonnie Webber and Maria Wolters.

I am investigating how the usage of sarcasm in natural language varies across cultures. Using insight from linguistics and anthropology, I build computational models that learn to detect, generate and interpret sarcasm. I work with terabytes of data collected from social media platforms to train and test my models.

I am a (Catholic) Christian. I am co-organising a discussion group on faith and science at the University of Edinburgh. I am very interested in fitness and sports nutrition, I play the piano, sing and ride horses.

Interests

  • Computational sarcasm detection, generation and interpretation
  • Knowledge representation and computational semantics
  • Research ethics and data privacy
  • Deep language modelling
  • Christian theology

History

I completed my MSc in Computer Science at the University of Oxford in the UK. In Oxford I built neural language models that focused on capturing morphological variation in Romanian language.

I completed my BSc in Computer Science at Jacobs University in Germany. At Jacobs I used topic models to detecting semantic similarity between scientific documents across the arXiv corpus.

I was a researcher at the National Institute for Standards and Technology in the USA. I worked on extending the core of LaTeXML – a TeX parser written in Perl, which produces XML-based, content-oriented marked-up documents. My contribution was providing an extensible graphics library, to allow the conversion of TeX graphics to SVG. This paper expands on the topic.

I did an internship at the German Research Centre for Artificial Intelligence, where I extended LaTeXML with modules for parsing documents from the arXiv corpus.

I worked as a software engineer at VisualDNA, where I developed a data integration platform, and at TheySay, where I developed evamind.

Contact

silviu@oprea.ch