DescriptionOpen Information Extraction or Open IE is a paradigm which enables extraction of relational tuples from text without pre-specifying relations. Most of the work in Open IE has been done for English. In this thesis we leverage the Open IE tools present in English to generate data in a non-English language by using cross lingual projection. This data can be used to train models capable of extracting relational tuples in multiple
languages. Universal Dependencies are used to generate features for these models that can be used across multiple languages.