Marvelous Agglutinative Language Effect on Cross Lingual Transfer Learning

Wooyoung Kim, Chaerin Jo, Minjung Kim, Wooju Kim

Abstract: E-Commerce Services such as Amazon and Alibaba have international consumers. The languages used in those services are very diverse. In order to provide a product search system to international consumers, it is essential to develop artificial intelligence that captures the semantic similarities among different languages. In this study, we propose a model that captures similarities between various languages as one artificial intelligence model. As for multilingual language models, it is important to select languages for training because of the curse of multilinguality. It is known that using languages with similar language structures is effective for cross lingual transfer learning. However, we demonstrate that using agglutinative languages such as Korean is more effective in cross lingual transfer learning. This is a great discovery that will change the training strategy of cross lingual transfer learning.