We propose a method to adapt an existing relation extraction system to extract new relation types with minimum supervision. Our proposed method comprises two stages: learning a lo...
—This paper presents a new framework for the completion of missing information based on local structures. It poses the task of completion as a global optimization problem with a ...