CAJ | 학술논문

提出一种基于条件随机域模型的生物命名实体识别方法,结合单词构词特性以及距离依赖特性,在JNLPBA的GENIAV3.02数据上进行实验,测试结果表明,引入距离依赖后,系统的识别性能比只利用单特性的条件随机域方法提高2.54%,可获得较好的识别效果,提高了系统的识别效率.
제출일충기우조건수궤역모형적생물명명실체식별방법,결합단사구사특성이급거리의뢰특성,재JNLPBA적GENIAV3.02수거상진행실험,측시결과표명,인입거리의뢰후,계통적식별성능비지이용단특성적조건수궤역방법제고2.54%,가획득교호적식별효과,제고료계통적식별효솔.
A biological named entity recognition method based on Conditional Random Fields(CRF) is proposed, which combines the word characteristics and the distance between words. Experiments are carried out with GENIAV3.02 datasets given by JNLPBA. Experimental results show that, after introducing words distance characteristics, the proposed method can achieve a performance improvement of 2.54% compared to simple conditional random fields, therefore achieving a better recognition result and improve the efficiency of systems.