My name is Miloš and I'm a PhD candidate (finishing soon I hope :) ) at the Institute for Logic, Language and Computation at University of Amsterdam working on Statistical Machine Translation under supervision of Prof. Khalil Sima'an .
My research interests lie mostly in the area of syntactic analysis of language. I do that in two directions: monolingual parsing with rich grammar formalisms (Minimalist Grammars and CCG) and bilingual parsing with synchronous grammars for machine translation.
My thesis topic is concerned mostly with synchronous parsing. I try to induce the hierarchical structure present in word alignments in an unsupervised way; as, for example, in this paper. Using this hierarchical structure (which is similar to syntactic structure but not the same) we can improve machine translation in different aspects. The ones I concentrate on are reordering and evaluation.
In short, my view is that it is wrong to ignore the hierarchical structure present in language (this happens very often in MT and other areas of NLP) and it is fun to learn this structure in unsupervised way. I would tell you more, but it would require more words, time and beer.
|favorite beer (other than BEER)|
|favorite programming language|