Knowledge transfer across multilingual corpora via latent topics
Wim De Smet, Jie Tang, Marie‐Francine Moens · 2011
Abstract. This paper explores bridging the content of two different languages via latent topics. Specifically, we propose a unified probabilistic model to si-multaneously model latent topics from bilingual corpora that discuss comparable content and use the topics as features in a cross-lingual, dictionary-less text cate-gorization ask. Experimental results on multilingual Wikipedia data show that the proposed topic model effectively discover the topic information from the bilin-gual corpora, and the learned topics successfully transfer classification knowl-edge to other languages, for which no labeled training data are available. 1