Oscar Täckström

I'm a research scientist at SICS and a PhD student at the Dept. of Linguistics and Philology at Uppsala University. My research interests are in machine learning, with application to problems in natural language processing, especially in structured latent variable models and semi-supervised models. Applications of recent interest include sentiment analysis, weakly supervised multilingual dependency parsing, semantic role labeling and hedge detection. For my Master's thesis I studied distributed representations of lexical semantics as a means for semi-supervised learning in the context of document classification. I have also dabbled in speaker identification, machine translation and active learning with application to dialogue act classification.

I spent the summers of 2010 and 2011 in New York City as an intern in the research group at Google. Together with my host Ryan McDonald I've worked on fine-grained sentiment analysis from coarse-grained labels, huge scale label propagation for lexicon construction, and most recently cross language model projection for dependency parsing and other tasks.

Funding for my PhD research is granted by GSLT. My main supervisor is Joakim Nivre and my assistant supervisor is Jussi Karlgren. I hold a MSc degree in Media Technology from the Royal Institute of Technology (KTH), majoring in machine learning and human computer interaction. Parallell to my MSc studies I studied philosophy full time for a year at Stockholm University.