| 基于Bi-LSTM-CRF的商业领域命名实体识别 |
| |
| 引用本文： | 丁晟春,方振,王楠.基于Bi-LSTM-CRF的商业领域命名实体识别[J].现代情报,2009,40(3):103-110. |
| |
| 作者姓名： | 丁晟春 方振 王楠 |
| |
| 作者单位： | 1. 南京理工大学经济管理学院, 江苏 南京 210094;2. 江苏省社会公共安全科技协同创新中心, 江苏 南京 210094 |
| |
| 基金项目： | 国家社会科学基金项目"基于社会网络分析的网络舆情主题发现研究"（项目编号：15BTQ063）。 |
| |
| 摘 要： | 目的/意义] 为解决目前网络公开平台的多源异构的企业数据的散乱、无序、碎片化问题，提出Bi-LSTM-CRF深度学习模型进行商业领域中的命名实体识别工作。方法/过程] 该方法包括对企业全称实体、企业简称实体与人名实体3类命名实体识别。结果/结论] 实验结果显示对企业全称实体、企业简称实体与人名实体3类命名实体识别的识别率平均F值为90.85%，验证了所提方法的有效性，证明了本研究有效地改善了商业领域中的命名实体识别效率。
|
| 关 键 词： | 商业领域 命名实体识别 深度学习 Bi-LSTM-CRF |
Business Domain Named Entity Recognition Based on Bi-LSTM-CRF |
| |
| Authors: | Ding Shengchun Fang Zhen Wang Nan |
| |
| Institution: | 1. School of Economics and Management, Nanjing University of Science and Technology, Nanjing 210094, China;2. Jiangsu Social Public Security Science and Technology Collaborative Innovation Center, Nanjing 210094, China |
| |
| Abstract: | Purpose/Significance] In order to solve the problem of scattered,disordered and fragmented multi-source heterogeneous enterprise data of the current network open platform,the Bi-LSTM-CRF deep learning model was proposed for the named entities recognition in the business field.Method/Process] This method included three kinds of named entities:enterprise full name entity,enterprise short name entity and personal name entity.Result/Conclusion] The experimental results showed that the average F value of the recognition rate of the three types of named entities,namely enterprise full entity,enterprise abbreviation entity and person name entity,was 90.85%,which verified the effectiveness of the proposed method.It was proved that this study effectively improved the efficiency of named entity recognition in the commercial field. |
| |
| Keywords: | business domain named entity recognition deep learning Bi-LSTM-CRF |
|
| 点击此处可从《现代情报》浏览原始摘要信息 |
| 点击此处可从《现代情报》下载免费的PDF全文 |
