From Babylon

BABYLON

Natural Language Processing for Languages with Scarce Resources

According to recent studies there are more than 7,000 languages spoken worldwide. From these, only about 15-20 languages can currently take advantage of the benefits provided by machine translation and other language processing tools. The goal of the Babylon project is to overcome this problem by developing natural language resources and tools for languages with scarce resources. In particular, our goal is to develop efficient methods for generating large amounts of word and phrase aligned multilingual parallel corpora especially for languages where currently there is little or no parallel text available. Parallel aligned texts represent a rich resource that can be used for creating supervised models of machine translation, as well as for building multilingual lexicons, or for annotation transfers from a well-studied language to a language studied to a lesser degree.