Bogdan Babych

Universität Heidelberg, Institute for Translating and Interpreting, Faculty Member

University of Leeds, Centre for Translation Studies, Faculty Member

Universität Heidelberg, Institut für Übersetzen und Dolmetschen, Faculty Member

Followers

150

Following

168

Co-authors

Public Views

Bogdan Babych is Professor of Translation Studies at Universität Heidelberg, Institute for Translation and Interpreting (IÜD), webpage: http://corpus.leeds.ac.uk/bogdan/ GitHub: https://github.com/bogdanbabych/

He works in the area of Computational Linguistics and Natural Language Processing and has published papers on evaluating and improving the quality of Machine Translation (MT) with Information Extraction techniques, extracting translation equivalents from large non-parallel corpora, developing MT for under-resourced languages, hybrid MT, on computational models for Ukrainian and Russian morphosyntax, machine translation to and from Slavonic languages, using Machine Translation for supporting language learning and authoring in non-native language. He previously worked as an Associate Professor in Translation Studies within the School of Modern Languages and Cultures a the University of Leeds and a computational linguist at L&H Speech Products, Belgium. He holds a PhD in Machine Translation from the University of Leeds, and a degree in Ukrainian Linguistics from Ukrainian National Academy of Sciences. He now coordinates an FP7 Marie Curie project HyghTra on developing a new hybrid MT architecture. He also worked in other FP7 projects -- ACCURAT (enhancing MT using comparable corpora for under-resourced languages) and TTC (mining translation terminology from comparable corpora). In 2007-2009 he worked on his project Translation Strategies in Comparable Corpora supported by the Leverhulme Early Career Research Fellowship.
Supervisors: Anthony Hartley and Yorick Wilks
Address: Institute for Translation and Interpreting (IÜD)
Heidelberg University
Plöck 57a, 69117 Heidelberg

less

InterestsView All (18)

Uploads

Papers by Bogdan Babych

Meta-evaluation of comparability metrics using parallel corpora

arXiv (Cornell University), Apr 14, 2014

Metrics for measuring the comparability of corpora or texts need to be developed and evaluated sy... more Metrics for measuring the comparability of corpora or texts need to be developed and evaluated systematically. Applications based on a corpus, such as training Statistical MT systems in specialised narrow domains, require finding a reasonable balance between the size of the corpus and its consistency, with controlled and benchmarked levels of comparability for any newly added sections. In this article we propose a method that can meta-evaluate comparability metrics by calculating monolingual comparability scores separately on the "source" and "target" sides of parallel corpora. The range of scores on the source side is then correlated (using Pearson's r coefficient) with the range of "target" scores; the higher the correlation-the more reliable is the metric. The intuition is that a good metric should yield the same distance between different domains in different languages. Our method gives consistent results for the same metrics on different data sets, which indicates that it is reliable and can be used for metric comparison or for optimising settings of parametrised metrics.

Bogdan Babych

Uploads

Papers by Bogdan Babych

Log In