Rapid Ukrainian-English Dictionary Creation Using Post-Edited Corpus Data

Abstract

This paper describes the development of a new corpus-based Ukrainian-English dictionary. The dictionary was built from scratch, we used no pre-existing dictionary data. A rapid dictionary development method was used which consists of generating dictionary parts directly from a large corpus, and of post-editing the automatically generated data by native speakers of Ukrainian (not professional lexicographers). The method builds on Baisa et al. (2019) which was improved and updated, and we used a different data management model. As the data source, a 3-billion-word Ukrainian web corpus from the TenTen series (Jakubíček et al., 2013) was used.

The paper briefly describes the corpus, then we thoroughly explain the individual steps of the automatic generation—post-editing workflow, including the volume of the manual work needed for the particular phases in terms of person-days. We also present details about the newly created dictionary and discuss directions for its further development.