Towards Entity Correctness, Completeness and Emergence for Entity Recognition
Published: 2015 Mai
Buchtitel: Proceedings of the 24rd International Conference on World Wide Web (WWW 2015) Poster Track
Verlag: to appear
Linking unstructured text to knowledge bases (KBs) by mapping words or phrases to the corresponding entities in KBs, is the problem of entity recognition and disambiguation. In this paper, we focus on the task of entity recognition in Web text to address the challenges of entity correctness, completeness and emergence that exiting approaches mainly suffer from. Experimental results show that the proposed approach significantly outperforms the state-of-the-art approaches in terms of precision, F-measure, micro-accuracy and macro-accuracy, while still preserving high recall.
Download: Media:WWW 2015 submission 1741.pdf