基于多任务学习的短文本实体链接方法

doi:10.19678/j.issn.1000-3428.0060474

计算机工程 ›› 2022, Vol. 48 ›› Issue (3): 315-320. doi: 10.19678/j.issn.1000-3428.0060474

• 开发研究与工程应用 • 上一篇

基于多任务学习的短文本实体链接方法

詹飞^1,2, 朱艳辉^1,2, 梁文桐^1,2, 张旭^1,2, 欧阳康^1,2, 孔令巍^1,2, 黄雅淋^1,2

1. 湖南工业大学计算机学院, 湖南株洲 412008;
2. 湖南省智能信息感知及处理技术重点实验室, 湖南株洲 412008

收稿日期:2021-01-04 修回日期:2021-02-23 发布日期:2021-03-02
作者简介:詹飞(1993-),男,硕士研究生,主研方向为自然语言处理、知识工程;朱艳辉(通信作者),教授;梁文桐、张旭、欧阳康、孔令巍、黄雅淋,硕士研究生。
基金资助:
国家自然科学基金（61871432）；湖南省自然科学基金（2020JJ6089）；湖南省教育厅基金资助重点项目（19A133）。

Short Text Entity Linking Method Based on Multi-Task Learning

ZHAN Fei^1,2, ZHU Yanhui^1,2, LIANG Wentong^1,2, ZHANG Xu^1,2, OUYANG Kang^1,2, KONG Lingwei^1,2, HUANG Yalin^1,2

1. School of Computer, Hunan University of Technology, Zhuzhou, Hunan 412008, China;
2. Hunan Key Laboratory of Intelligent Information Perception and Processing Technology, Zhuzhou, Hunan 412008, China

Received:2021-01-04 Revised:2021-02-23 Published:2021-03-02

摘要/Abstract

摘要： 实体链接是明确文本中实体指称的重要手段，也是构建知识图谱的关键技术，在智能问答、信息检索等领域中具有重要作用，但由于短文本的上下文语境不丰富、表达不正式、语法结构不完整等特点，现有的短文本实体链接方法准确率较低。提出一种新的短文本实体链接方法，将多任务学习方法引入短文本实体链接过程中，从而增强短文本实体链接方法的效果。在此基础上，构建多任务学习模型，将短文本实体链接作为主任务，并引入实体分类作为辅助任务，促使模型学习到更加通用的底层表达，提高模型的泛化能力，优化模型在短文本实体链接任务中的表现。在CCKS2020测评任务2提供的数据集上的实验结果表明，辅助任务的引入能够缓解短文本实体链接过程中信息不充分的问题，且该多任务学习模型的F值为0.894 9，优于基于BERT编码器的单任务实体链接模型。

关键词: 短文本实体链接, 多任务学习, 实体分类, 辅助任务, 底层表达

Abstract: Entity linking is an important means to clarify the entity reference in the text, and the rise of the knowledge graph has prompted the development of entity linking technology, including playing an important role in question answering and information retrieval.However, the accuracy of the existing short text entity linking method is low because of the lack of rich context, informal expression and incomplete grammatical structure of short texts.A short new text entity linking method based on multi-task learning is proposed to solve this problem.Specifically, the multi-task learning method is introduced into the short text entity linking process, a multi task learning model is constructed, which takes a short text entity link as the main task and introduces entity classification as an auxiliary task-this alleviates the problem of insufficient information in the process of a short text entity link.On this basis, the model is promoted to learn more general underlying expressions, improve the generalization ability of the model, and optimize the performance of the model in the short text entity link task.The experimental results on the data set provided by the National Knowledge Mapping and Semantic Computing Conference (CCKS 2020) Evaluation Task 2 show that the introduction of auxiliary tasks can alleviate the problem of insufficient information in the process of short text entity linking and the F value of this proposed model reached 0.894 9, which is better than the single task entity linking model based on Bidirectional Encoder Representations from Transformers(BERT).

Key words: short text entity linking, multi-task learning, entity classification, auxiliary task, underlying expression

中图分类号:

TP391.4

詹飞, 朱艳辉, 梁文桐, 张旭, 欧阳康, 孔令巍, 黄雅淋. 基于多任务学习的短文本实体链接方法[J]. 计算机工程, 2022, 48(3): 315-320.

ZHAN Fei, ZHU Yanhui, LIANG Wentong, ZHANG Xu, OUYANG Kang, KONG Lingwei, HUANG Yalin. Short Text Entity Linking Method Based on Multi-Task Learning[J]. Computer Engineering, 2022, 48(3): 315-320.

http://www.ecice06.com/CN/Y2022/V48/I3/315

图/表 8

20220401124115

20220401124118

20220401124121

20220401124124

20220401124127

20220401124131

20220401124134

20220401124137

参考文献

[1] 刘峤, 李杨, 段宏, 等.知识图谱构建技术综述[J].计算机研究与发展, 2016, 53(3):582-600. LIU Q, LI Y, DUAN H, et al.Knowledge graph construction techniques[J].Journal of Computer Research and Development, 2016, 53(3):582-600.(in Chinese)
[2] WANG F, WU W, LI Z, et al.Named entity disambiguation for questions in community question answering[J].Knowledge-Based Systems, 2017, 126(5):68-77.
[3] SHEN W, WANG J, HAN J.Entity linking with a knowledge base:issues, techniques, and solutions[J].IEEE Transactions on Knowledge and Data Engineering, 2014, 27(2):443-460.
[4] ZENG W, ZHAO X, TANG J, et al.CLEEK:a Chinese long-text corpus for entity linking[C]//Proceedings of the 12th Language Resources and Evaluation Conference.Marseille, France:European Language Resources Association, 2020:2026-2035.
[5] ZHANG Y, YANG Q.An overview of multi-task learning[J].National Science Review, 2018, 5(1):30-43.
[6] RUDER S.An overview of multi-task learning in deep neural networks[J].National Science Review, 2018:5(1), 14-19.
[7] DOERSCH C, ZISSERMAN A.Multi-task self-supervised visual learning[C]//Proceedings of IEEE International Conference on Computer Vision.Washington D.C., USA:IEEE Press, 2017:2051-2060.
[8] MISRA I, SHRIVASTAVA A, GUPTA A, et al.Cross-stitch networks for multi-task learning[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2016:3994-4003.
[9] LIU S, JOHNS E, DAVISON A J.End-to-end multi-task learning with attention[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2019:1871-1880.
[10] DONG D, WU H, HE W, et al.Multi-task learning for multiple language translation[C]//Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics and the 7th International Joint Conference on Natural Language Processing.Beijing, China:Association for Computational Linguistics, 2015:1723-1732.
[11] ZHANG T, XU C, YANG M H.Multi-task correlation particle filter for robust object tracking[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2017:4335-4343.
[12] RAIMAN J, RAIMAN O.Deeptype:multilingual entity linking by neural type system evolution[EB/OL].[2020-12-01].https://arxiv.org/abs/1802.01021.
[13] SEVGILI O, SHELMANOV A, ARKHIPOV M, et al.Neural entity linking:a survey of models based on deep learning[EB/OL].[2020-12-01].https://arxiv.org/abs/2006.00575v3.
[14] CETOLI A, BRAGAGLIA S, O'HARNEY A D, et al.A neural approach to entity linking on wikidata[C]//Proceedings of European Conference on Information Retrieval.Berlin, Germany:Springer, 2019:78-86.
[15] TRISEDYA B D, QI J, ZHANG R.Entity alignment between knowledge graphs using attribute embeddings[EB/OL].[2020-12-01].https://people.eng.unimelb.edu.au/jianzhongq/papers/AAAI2019_EntityAlignment.pdf.
[16] PHAN M C, SUN A, TAY Y, et al.NeuPL:attention-based semantic matching and pair-linking for entity disambiguation[C]//Proceedings of Conference on Information and Knowledge Management.New York, USA:ACM Press, 2017:1667-1676.
[17] LIU C, LI F, SUN X, et al.Attention-based joint entity linking with entity embedding[J].Information, 2019, 10(2):46-54.
[18] SIL A, KUNDU G, FLORIAN R, et al.Neural cross-lingual entity linking[EB/OL].[2020-12-01].https://arxiv.org/pdf/1712.01813.pdf.
[19] GU Y, QU X, WANG Z, et al.Read, retrospect, select:an MRC framework to short text entity linking[EB/OL].[2020-12-01].https://www.researchgate.net/publication/348321250_Read_Retrospect_Select_An_MRC_Framework_to_Short_Text_Entity_Linking.
[20] 张晟旗, 王元龙, 李茹, 等.基于局部注意力机制的中文短文本实体链接[J].计算机工程, 2021, 47(11):77-83, 92. ZHANG S Q, WANG Y L, LI R, et al.Entity linking based on local attention mechanism for Chinese short text[J].Computer Engineering, 2021, 47(11):77-83, 92.(in Chinese)
[21] DEVLIN J, CHANG M W, LEE K, et al.Bert:pre-training of deep bidirectional transformers for language understanding[EB/OL].[2020-12-01].https://arxiv.org/abs/1810.04805.

选择文件类型/文献管理软件名称

选择包含的内容

基于多任务学习的短文本实体链接方法

Short Text Entity Linking Method Based on Multi-Task Learning

RichHTML

PDF

可视化

被引次数

摘要/Abstract

引用本文

使用本文

图/表 8

参考文献

相关文章 11

编辑推荐

Metrics

本文评价

[1]	李晓腾, 张盼盼, 勾智楠, 高凯. 基于多任务学习的多模态命名实体识别方法[J]. 计算机工程, 2023, 49(4): 114-119.
[2]	饶东宁, 罗南岳. 基于多任务强化学习的堆垛机调度与库位推荐[J]. 计算机工程, 2023, 49(2): 279-287,295.
[3]	游乐圻, 裴忠民, 罗章凯. 融合实体类型信息的本体-实例联合学习方法[J]. 计算机工程, 2022, 48(7): 82-88.
[4]	王强, 林友芳, 万怀宇. 基于多任务学习的快件送达时间预测方法[J]. 计算机工程, 2022, 48(4): 314-320.
[5]	曹加旺, 田维维, 刘学玲, 李郁欣, 冯瑞. 基于改进U-Net的人脑黑质致密部分割[J]. 计算机工程, 2022, 48(11): 14-21,29.
[6]	叶剑锋, 徐轲, 熊峻峰, 王化明. 基于注意力机制和辅助任务的语义分割算法[J]. 计算机工程, 2021, 47(9): 203-209,216.
[7]	周伟枭, 蓝雯飞. 融合文本分类的多任务学习摘要模型[J]. 计算机工程, 2021, 47(4): 48-55.
[8]	李亚, 张雨楠, 彭程, 杨俊钦, 刘淼. 基于多任务学习的人脸属性识别方法[J]. 计算机工程, 2020, 46(3): 229-236.
[9]	张翔, 陈欣. 多任务自主学习的肺癌诊断方法[J]. 计算机工程, 2020, 46(2): 292-297,303.
[10]	张文金. 一种基于权重矩阵分解的在线多任务学习算法[J]. 计算机工程, 2019, 45(8): 190-197.
[11]	李薇,肖仰华,汪卫. 基于中文知识图谱的人物实体识别[J]. 计算机工程, 2017, 43(3): 225-231,240.

模态框（Modal）标题

选择文件类型/文献管理软件名称

选择包含的内容

基于多任务学习的短文本实体链接方法

Short Text Entity Linking Method Based on Multi-Task Learning

RichHTML

PDF

可视化

被引次数

摘要/Abstract

引用本文

使用本文

图/表 8

参考文献

相关文章 11

编辑推荐

Metrics

本文评价