Entity Extraction is a Boring Solved Problem - Or is it?

This paper presents empirical results that contradict the prevailing opinion that entity extraction is a boring solved problem. In particular, we consider data sets that resemble familiar MUC/ACE data, and report surprisingly poor performance for both commercial and research systems. We then give an error analysis that suggests research challenges for entity extraction that are neither boring nor solved. 1 Background Entity extraction or named entity recognition, as it is sometimes called, is a known and familiar problem. Named entity (NE) tagging has been the subject of numerous shared-task evaluations, including the seminal MUC 6, MUC 7 and MET evaluations, the CoNLL shared task, the SIGHAN bake-offs, and the ACE evaluations. With this track record, and with commercial vendors now selling named-entity tagging for a fee, many naturally consider entity extraction to be an essentially solved problem. The present paper challenges this view. The main issue, as we see it, is transfer: NE ...

• Type: Conference
• Year: 2007
• Where: NAACL
• Authors: Marc Vilain, Jennifer Su, Suzi Lubar
• Keywords: Background Entity Extraction, Computational Linguistics, Data Sets, Entity Extraction, NAACL 2007