| 融合多特征的中文产品命名实体识别 |
| |
| 引用本文： | 谷川. 融合多特征的中文产品命名实体识别[J]. 科学技术与工程, 2013, 13(31) |
| |
| 作者姓名： | 谷川 |
| |
| 作者单位： | 安阳师范学院 |
| |
| 基金项目： | 国家自然科学基金(60875081)、河南省基础与前沿技术研究计划项目(112300410182)、河南省软科学研究项目(132400410248) |
| |
| 摘 要： | 针对中文产品命名实体，提出了一种基于多种特征融合的识别方法。该方法以词为标注粒度，将多种特征融合到条件随机场模型中，采用递增式学习策略选取最优的特征模板，实现了从中文自由文本中识别产品命名实体。实验表明，该方法获得了令人满意的实验效果，准确率、召回率和F值分别达到94.87%、92.50%和93.67%。
|
| 关 键 词： | 多特征产品命名实体识别条件随机场 |
| 收稿时间： | 2013-06-20 |
| 修稿时间： | 2013-06-21 |
Fusion of Multiple Features for Chinese Product Named Entity Recognition |
| |
| Gu Chuan. Fusion of Multiple Features for Chinese Product Named Entity Recognition[J]. Science Technology and Engineering, 2013, 13(31) |
| |
| Authors: | Gu Chuan |
| |
| Abstract: | This paper presents an approach for Chinese product named entity recognition based on fusion of multiple features. The approach fuses multiple features to conditional random fields model, using Chinese word as processing unit, which adopts incremental learning strategy to optimize feature templates and recognizes product named entity from Chinese free text. The experimental results show that the approach performs quite well, the precision, recall and F value can reach 96.87%, 94.50% and 95.67% respectively. |
| |
| Keywords: | multiple features product named entity recognitionconditional random fields feature template |
|
| 点击此处可从《科学技术与工程》浏览原始摘要信息 |
|
点击此处可从《科学技术与工程》下载全文 |
