Katharina Kann Wins SIGMORPHON 2016 Shared Task
Katharina Kann has won the SIGMORPHON 2016 shared task http://ryancotterell.github.io/sigmorphon2016/. Other participants in the shared task were MIT, NYT and University of Colorado. The task consists of generating morphologically inflected forms in 10 languages. An English example would be to generate the output "broken" for the input "to-break-PAST-PARTICIPLE". Her system applies neural machine translation methods to this task and placed first in all 9 subevaluations for all of the 10 languages that were considered. Katharina is a PhD student in the Data Science Lab at LMU Munich. She is part of the project "Solutions for High-Throughput Data", a joint project of the Database Systems Group and the Center for Information and Language Processing that is financed by Siemens.