Cross-Lingual Word Embeddings for Morphologically Rich Languages

Cross-lingual word embedding models learna shared vector space for two or more lan-guages so that words with similar meaningare represented by similar vectors regardlessof their language. Although the existing mod-els achieve high performance on pairs of mor-phologically simple languages, they performvery poorly on morphologically rich languagessuch as Turkish and Finnish. In this pa-per, we propose a morpheme-based model inorder to increase the performance of cross-lingual word embeddings on morphologicallyrich languages. Our model includes a sim-ple extension which enables us to exploit mor-phemes for cross-lingual mapping. We ap-plied our model for the Turkish-Finnish lan-guage pair on the bilingual word translationtask. Results show that our model outper-forms the baseline models by 2% in the nearestneighbour ranking