GWU English TAC-KBP EL Diagnostic Task with Name Mention

This paper describes the entity linking system participating in the 2015 Knowledge Base Population (KBP) track at the Text Analysis Conference (TAC) by GWU’s Natural Language Processing (NLP) group (Care4Lang) in collaboration with the NLP consulting company Luki Labs. Our proposed system uses a supervised modeling approach with a feature set that targets the overlapping information between the query and the candidate entities from the KB. In addition, it uses an unsupervised approach to cluster the mentions that don’t have a reference in the KB. It is a first participation for both teams and the attained results are promising and encouraging for further research.