Chinese ner hybrid pattern based on multi-feature fusion

This paper focuses on the Chinese Named Entity Recognition (NER) hybrid pattern, and emphasizes particularly on the fusion mechanism of multiple features for NE acquisition. It differentiates from most of previous methods mainly as that Local Features and Global Features are integrated to get higher...

Full description

Saved in:
Bibliographic Details
Main Authors: Yue-Jie Zhang, Wei Wu, Cheng Jin, Tao Zhang
Format: Conference Proceeding
Language:English
Subjects:
Conditional Random Field (CRF)
Data models
Educational institutions
Feature extraction
Humans
Machine learning
Maximum Entropy (ME)
multi-feature fusion
Named Entity Recognition (NER)
Reliability
reliability evaluation
Training
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:This paper focuses on the Chinese Named Entity Recognition (NER) hybrid pattern, and emphasizes particularly on the fusion mechanism of multiple features for NE acquisition. It differentiates from most of previous methods mainly as that Local Features and Global Features are integrated to get higher performance. Meanwhile, to reduce search space and improve processing efficiency, Heuristic Human Knowledge is introduced into the statistical model, which could increase the performance significantly. From the experimental results on data sets of People's Daily and NER Task in SIGHAN2008, it can be concluded that our hybrid model based on multi-feature fusion is an effective NER pattern to combine statistical model and heuristic human knowledge.
ISSN:2160-133X
DOI:10.1109/ICMLC.2011.6016675