Named Entity Recognition in Vietnamese Text Using Label Propagation

This paper presents our named entity recognition system for Vietnamese text using labeled propagation. In here we propose: (i) a method of choosing noun phrases as the named entity candidates; (ii) a method to measure the word similarity; and (iii) a method of decreasing the effect of high frequency labels in labeled documents. Experimental results show that our labeled propagate method achieves higher accuracy than the old one [12]. In addition, when the number of the labeled data is small, its accuracy is higher than when using conditional random fields.