An adaptive approach for web scale named entity recognition

Named entities are the basic components for semantic Web ontologies and social association networks. How to recognize named entities on a Web scale is challenging due to named entity disambiguation, learning and acquisition of vocabularies and patterns etc. In this paper, we propose a novel adaptive named entity recognition (NER) framework which addresses these challenges on multiple domains on the Web. We propose an approach for discovering domain hierarchies from Web link structures, and formalizing domain vocabulary and patterns as association rules on these domains for NER. These domain vocabulary and patterns are defined on the domain hierarchy for achieving effectiveness and efficiency in NER.