Semantic analogy-based compound splitter

A high-quality compound splitter based on the semantic regularities in the vector space of word embeddings. Initially, we provide models for German, but broader language support will be added in the coming releases.

More details can be found in: 

  • Joachim Daiber, Lautaro Quiroz, Roger Wechsler and Stella Frank (2015) Splitting Compounds by Semantic Analogy. In Proceedings of the 1st Deep Machine Translation Workshop, Prague, Czech Republic, pp. 20 - 28