Recognition of chinese personal names based on CRFs and law of names

Authors:
Zheng Lvexing;Lv Xueqiang;Liu Kun;Du Yuncheng
Affiliations:
Chinese Information Processing Research Center, Beijing Information Science & Technology University, Beijing, China;Chinese Information Processing Research Center, Beijing Information Science & Technology University, Beijing, China and Beijing TRS Information Technology Co., Ltd., Beijing, China;Chinese Information Processing Research Center, Beijing Information Science & Technology University, Beijing, China and Beijing TRS Information Technology Co., Ltd., Beijing, China;Chinese Information Processing Research Center, Beijing Information Science & Technology University, Beijing, China and Beijing TRS Information Technology Co., Ltd., Beijing, China
Venue:
APWeb'12 Proceedings of the 14th international conference on Web Technologies and Applications
Year:
2012

Citing 3
Cited 0

Conditional Random Fields: Probabilistic Models for Segmenting and Labeling Sequence Data

ICML '01 Proceedings of the Eighteenth International Conference on Machine Learning
Named entity recognition using an HMM-based chunk tagger

ACL '02 Proceedings of the 40th Annual Meeting on Association for Computational Linguistics
Maximum entropy models for named entity recognition

CONLL '03 Proceedings of the seventh conference on Natural language learning at HLT-NAACL 2003 - Volume 4

Quantified Score

Hi-index	0.00

Visualization

Abstract

Recognition of chinese personal names becomes a difficult and key point in chinese unknown word recognition. This paper explored the context boundary of names and the law of names. The context boundary of names is concentrated, which can reduce recognition errors brought up by Forward Maximum Matching Segmentation; from real text corpus, we discover that names begin with surname, dislocation characters of names reach 70.83%, and rare characters of names reach 9.42%. This paper improved the Forward Maximum Matching Segmentation and implemented a name recognition test based on CRFs, which was combined with the surname, the context boundary, the dislocation character and the rare character. The open test shows that recall reaches 91.24% from BakeOff-2005.