[1]李彪,陈燕红*,温明,等.TResX-BiGRU:科技人才履历实体识别模型[J].计算机技术与发展,2025,(02):159-165.[doi:10.20165/j.cnki.ISSN1673-629X.2024.0315]
　LI Biao,CHEN Yan-hong*,WEN Ming,et al.TResX-BiGRU:Entity Recognition Model for Technology Talent Resumes[J].,2025,(02):159-165.[doi:10.20165/j.cnki.ISSN1673-629X.2024.0315]
点击复制

TResX-BiGRU:科技人才履历实体识别模型()

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:
期数:
2025年02期
页码:
159-165
栏目:
人工智能
出版日期:
2025-02-10

文章信息/Info

Title:
TResX-BiGRU:Entity Recognition Model for Technology Talent Resumes
文章编号:
1673-629X(2025)02-0159-07
作者:
李彪1; 2; 3; 4; 陈燕红1; 2; 3*; 温明4; 肖天赐1; 2; 3; 韩博1; 2; 3
1. 新疆农业大学 计算机与信息工程学院,新疆 乌鲁木齐 830052;
2. 智能农业教育部工程研究中心,新疆 乌鲁木齐 830052;
3. 新疆农业信息化工程技术研究中心,新疆 乌鲁木齐 830052;
4. 新疆电子研究所,新疆 乌鲁木齐 830013
Author(s):
LI Biao1; 2; 3; 4; CHEN Yan-hong1; 2; 3*; WEN Ming4; XIAO Tian-ci1; 2; 3; HAN Bo1; 2; 3
1. School of Computer and Information Engineering,Xinjiang Agricultural University,Urumqi 830052,China;
2. Engineering Research Center of Intelligent Agriculture of Ministry of Education,Urumqi 830052,China;
3. Xinjiang Engineering and Technology Research Center of Agricultural Informatization,Urumqi 830052,China;
4. Xinjiang Institute of Electronics,Urumqi 830013,China
关键词:
XLNet模型; BiGRU; 科技人才履历; 残差连接; 命名实体识别
Keywords:
XLNet model; BiGRU; technology talent resumes; residual connection; named entity recognition
分类号:
TP391.1
DOI:
10.20165/j.cnki.ISSN1673-629X.2024.0315
摘要:
科技人才履历实体识别是人才知识图谱构建、人才推荐等应用的重要基础。 针对传统语言模型在处理长文本时效果不佳,难以有效识别文本中的实体关键信息,导致实体识别准确率低等问题,该文设计了一种科技人才履历命名实体识别模型 TResX-BiGRU。 文本特征通过 XLNet 模型进行编码,捕捉其丰富的上下文语义信息;使用 BiGRU 模型对文本的时序特征进行有效建模,进一步挖掘深层次文本语义关系;引入残差连接充分融合文本的不同层次的特征信息,将 XLNet 与 BiGRU 的输出经过线性变换后进行累加,最后输出最佳的标签序列。 在自建的科技人才履历 Talents 语料库与 Resume 数据集上进行了多层次的对比分析,实验结果表明,TResX-BiGRU 模型达到了较好的性能,分别获得了 82.36% 与 97.54% 的 F1 值,在科技人才履历命名实体识别任务中是一种先进的模型。
Abstract:
In the context of constructing talent knowledge graphs and talent recommendation systems, recognizing entities in technology talent resumes is crucial. Traditional language models often perform poorly with long texts,failing to effectively identify key entity infor-mation,which results in low entity recognition accuracy. We introduce the TResX - BiGRU model for named entity recognition in technology talent resumes. The model encodes text features using the XLNet model to capture rich contextual semantic information. It employs a BiGRU model to effectively model the sequential features of the text,further exploring deep semantic relationships. A residual connection is introduced to fully integrate features from different text layers,combining the outputs of XLNet and BiGRU through linear transformation and accumulation,ultimately yielding the optimal label sequence. We conduct multi-level comparative analyses on the self-constructed Talents corpus and the Resume dataset. Experimental results demonstrate that the proposed TResX-BiGRU model achieves superior performance,with F1 scores of 82. 36% and 97. 54% , respectively, establishing it as an advanced model for named entity recognition in technology talent resumes.

相似文献/References:

[1]冼广铭,王鲁栋,曾碧卿,等.基于 LDA 和 BiGRU 的文本分类[J].计算机技术与发展,2022,32(04):15.[doi:10. 3969 / j. issn. 1673-629X. 2022. 04. 003]
　XIAN Guang-ming,WANG Lu-dong,ZENG Bi-qing,et al.Text Classification Based on LDA and BiGRU[J].,2022,32(02):15.[doi:10. 3969 / j. issn. 1673-629X. 2022. 04. 003]
[2]吴加辉,加云岗,王志晓,等.基于深度学习的微博疫情舆情文本情感分析[J].计算机技术与发展,2024,34(07):175.[doi:10.20165/j.cnki.ISSN1673-629X.2024.0087]
　WU Jia-hui,JIA Yun-gang,WANG Zhi-xiao,et al.Sentiment Analysis of Weibo Epidemic Public Opinion Text Based on Deep Learning[J].,2024,34(02):175.[doi:10.20165/j.cnki.ISSN1673-629X.2024.0087]

更新日期/Last Update: 2025-02-10