Cross-Lingual Cross-Document Coreference with Entity Linking

This paper describes our approach to the 2011 Text Analysis Conference (TAC) Knowledge Base Population (KBP) cross-lingual entity linking problem. We recast the problem of entity linking as one of cross-document entity coreference. We compare an approach where deductive entity linking informs crossdocument coreference to an inductive approach where coreference and linking judgements are mutually beneficial. We also describe our approach to cross-lingual entity linking comparing a native linking approach with an approach utilizing machine translation. Our results show that inductive linking to a native language knowledge base offers the best performance.