Prof. Dr. Bogdan Babych

Professor of Translation Studies, Department of Translation, Communication and Technology

Institute for Translation and Interpreting, Universität Heidelberg

Home

Research

Teaching

Collaboration

Techologies

Development of translation technologies and applications

Research and teaching experiments

1. cgiBLEU: On-line interface to BLEU / NIST evaluation toolkit: cgiBLEU
2. colloc4nlg Experiments on using collocations for Natural Language Generation: colloc4nlg; web; code
3. Kyiv2020 Kyiv Translation Summer School 2020 at Igor Sikorsky Kyiv Polytechnic Institute presentation and materials
4. smallSMT A prototype SMT system for Lingenio’s FlexNeuroTrans project: web interface web file interface
5. Ukrainian corpus and morphological tools:
   • Parameter files for the TNT part-of-speech tagger; research paper with evaluation.
   • Ukrainian news corpus, 250 million words, up to 2012: web interface; downloadable .vert file (1.2GB); smaller 10 million words corpus sample: downloadable files.
6. Processing CORD19 corpus
   • This notebook downloads and reads the free cord19 corpus into one file.
   • The notebook is hosted at IÜD, Heidelberg University github repository https://github.com/iued-uni-heidelberg/cord19.
   • CORD19 (covid-19) is an open-source corpus available from https://www.semanticscholar.org/cord19/download. Documentation is available at https://github.com/allenai/cord19. The original files are in json format.
   • The output file is in plain text format; documents are separated (by default) by <doc id=”doc1000001”> … </doc> tags. The purpose of the plain text file is for further processing, e.g., generating linguistic annotation using the TreeTagger or the Stanford parser for part-of-speech annotation or dependency / constituency parsing.