LCC Approaches to Knowledge Base Population at TAC 2010

The Knowledge Base Population (KBP) track at the Text Analysis Conference 2010 marks the second year of this important information extraction evaluation. This paper describes the design and implementation of LCC’s systems which participated in the tasks of Entity Linking, Slot Filling, and the new task of Surprise Slot Filling. For the entity linking task, our top score was achieved through a robust context modeling approach which incorporates topical evidence. For slot filling, we used the output of the entity linking system together with a combination of different types of relation extractors. For surprise slot filling, our customizable extraction system was extremely useful due to the time sensitive nature of the task.