上下位关系抽取方法及其在金融市场的应用<sup>*</sup>

doi:10.11925/infotech.2096-3467.2020.1261

数据分析与知识发现

2021, Vol. 5

Issue (10): 60-70 https://doi.org/10.11925/infotech.2096-3467.2020.1261

研究论文

本期目录 | 过刊浏览 | 高级检索

上下位关系抽取方法及其在金融市场的应用^*

戴志宏¹(

),郝晓玲^1,²

¹上海财经大学信息管理与工程学院上海 200433
²上海财经大学上海市金融信息技术研究重点实验室上海 200433

Extracting Hypernym-Hyponym Relationship for Financial Market Applications

Dai Zhihong¹(

),Hao Xiaoling^1,²

¹School of Information Management & Engineering, Shanghai University of Finance and Economics, Shanghai 200433, China
²Shanghai Key Laboratory of Financial Information Technology, Shanghai University of Finance and Economics, Shanghai 200433, China

摘要
图/表
参考文献
相关文章
Metrics

全文: PDF (908 KB) HTML ( 14 )
输出: BibTeX | EndNote (RIS)

摘要

【目的】 针对知识图谱的上下位关系抽取技术,提出基于映射矩阵和词向量相似度相结合的抽取方法,并探索在实际应用中的有效性。【方法】 构建对应不同类别上下位词对的多个映射矩阵,充分运用实体词汇及其上下文包含的语义信息,进而利用映射矩阵识别实体词对间是否存在上下位关系。【结果】 该方法在测试集上的综合表现比已有方法的F1值高出0.032 1。以上市公司为数据集的实验结果表明,该方法有助于研究公司相似度和股票联动的相关性。【局限】 可尝试对更多的上下位词对聚类,以更精准地区分上下位关系的类别;可引入模式匹配方法,对该方法进行补充,提升抽取效果。【结论】 所提方法能够有效进行实体上下位关系识别,并能有效帮助挖掘金融市场中存在关联的上市公司、股票板块等,在金融领域的知识图谱构建中有推广和应用前景。

	服务

	把本文推荐给朋友
	加入引用管理器
	E-mail Alert
	RSS
	作者相关文章
	戴志宏
	郝晓玲

关键词 ：上下位关系, 关系抽取, 词向量, 股票联动

Abstract：

[Objective] This paper proposes a new method to extract superior-inferior relationship from knowledge graph, and then explores its effectiveness with practical application. [Methods] First, we constructed the mapping matrix for hypernym-hyponym words and their context semantics. Then, we combined word vector similarity with the matrix to extract the relation. [Results] We examined our method with datasets of listed companies and found its F1 value was more than 3% higher than those of the existing methods. The new model could help us study the association between company similarity and stock performance. [Limitations] More research is needed to improve relationship extraction with the help of clustering technique and pattern matching method. [Conclusions] The proposed method can effectively identify the relationship between entities, and study the related listed companies and stocks. It also helps us construct better knowledge graph in the financial field.

Key words： Hypernym-Hyponym Relationship Relationship Extraction Word Vector Stock Linkage

收稿日期: 2020-12-16 出版日期: 2021-08-27

ZTFLH:

TP391

基金资助:*国家社会科学基金项目(20BGL287);国家自然科学基金项目(71401096)

通讯作者: 戴志宏, ORCID：0000-0002-3890-115X E-mail: daizhihong@189.cn

引用本文:

戴志宏, 郝晓玲. 上下位关系抽取方法及其在金融市场的应用^*[J]. 数据分析与知识发现, 2021, 5(10): 60-70.
Dai Zhihong, Hao Xiaoling. Extracting Hypernym-Hyponym Relationship for Financial Market Applications. Data Analysis and Knowledge Discovery, 2021, 5(10): 60-70.

链接本文:

https://manu44.magtech.com.cn/Jwk_infotech_wk3/CN/10.11925/infotech.2096-3467.2020.1261 或 https://manu44.magtech.com.cn/Jwk_infotech_wk3/CN/Y2021/V5/I10/60

序号	实例
1	$v 虾 - v 对虾 ≈ v 鱼 - v 金鱼$
2	$v 工人 - v 木匠 ≈ v 演员 - v 小丑$
3	$v 工人 - v 木匠 ≠ v 鱼 - v 金鱼$

Table 1 上下位词对向量偏移计算结果

Fig.1 基于矩阵映射+词向量的上下位抽取方法

Table 2 词向量空间中最近的5个词

Fig.2 相似度下限及聚类数目对F1值的影响
注：图例中的三个数字分别是指相似度下限的三个不同取值

参数	取值
$δ 1$	0.5
$δ 2$	6
细分区间及相似度下限	$1.5 < d ≤ 2$ ,0.15/0.2
	$2 < d ≤ 4$ ,0.25/0.3
	$4 < d ≤ 6$ ,0.35/0.4
聚类类别数K	60

Table 3 相关参数

指标	$δ + Simi$	映射偏离 $δ$	词向量相似度 $Simi$
准确率	0.835 4	0.643 6	0.716 7
召回率	0.834 7	0.815 2	0.805 4
F1值	0.835 0	0.719 3	0.758 4

Table 4 测试集判断结果

Table 5 本文方法与前人研究方法对比

Table 6 中国神华按行业和产品划分的主营构成

Table 7 抽取的行业和产品词列表

Table 8 抽取的互联网领域实体词

Table 9 抽取的上下位词对

Fig.3 互联网产品与服务层级结构

Table 10 三组股票日收益率及日超额收益率相关系数

[1]	陈金栋, 肖仰华. 一种基于语义的上下位关系抽取方法[J]. 计算机应用与软件, 2019, 36(2): 216-221.
[1]	(Chen Jindong, Xiao Yanghua. Hypernymy Relation Extraction Based on Semantics[J]. Computer Applications and Software, 2019, 36(2): 216-221.)
[2]	钟茂生, 刘慧, 刘磊. 词汇间语义相关关系量化计算方法[J]. 中文信息学报, 2009, 32(2): 115-122.
[2]	(Zhong Maosheng, Liu Hui, Liu Lei. Method of Semantic Relevance Relation Measurement Between Words[J]. Journal of Chinese Information Processing, 2009, 32(2): 115-122.)
[3]	邱科达, 马建玲. 基于文本语料的上下位关系识别研究综述[J]. 情报科学, 2020, 38(7): 162-172.
[3]	(Qiu Keda, Ma Jianling. A Review of Hypernym Relation Recognition from Text Corpora[J]. Information Science, 2020, 38(7): 162-172.)
[4]	Hearst M A. Automatic Acquisition of Hyponyms from Large Text Corpora [C]//Proceedings of the 14th Conference on Computational Linguistics - Volume 2. 1992: 539-545.
[5]	汤青, 吕学强, 李卓. 本体概念间上下位关系抽取研究[J]. 微电子学与计算机, 2014, 31(6): 68-71.
[5]	(Tang Qing, Lv Xueqiang, Li Zhuo. Research on Domain Ontology Concept Hyponymy Relation Extraction[J]. Microelectronics & Computer, 2014, 31(6): 68-71.)
[6]	李军锋, 吕学强, 李卓. 专利领域本体概念语义层次获取[J]. 情报学报, 2014, 33(9): 986-993.
[6]	(Li Junfeng, Lv Xueqiang, Li Zhuo. Deriving Concept Semantic Hierarchy of Ontology in Patents[J]. Journal of the China Society for Scientific and Technical Information, 2014, 33(9): 986-993.)
[7]	张晨童, 张佳影, 张知行, 等. 融合常用语的大规模疾病术语图谱构建[J]. 计算机研究与发展, 2020, 57(11): 2467-2477.
[7]	(Zhang Chentong, Zhang Jiaying, Zhang Zhixing, et al. Construction of Large-Scale Disease Terminology Graph with Common Terms[J]. Journal of Computer Research and Development, 2020, 57(11): 2467-2477.)
[8]	Snow R. Learning Syntactic Patterns for Automatic Hypernym Discovery [C]//Proceedings of the 17th International Conference on Neural Information Processing Systems. 2005: 1297-1304.
[9]	Suchanek F M, Kasneci G, Weikum G. YAGO: A Large Ontology from Wikipedia and WordNet[J]. Journal of Web Semantics, 2008, 6(3): 203-217. doi: 10.1016/j.websem.2008.06.001
[10]	陆凯华, 李正华, 张民. 汉语上下位关系分类数据集构建和基准方法比较[J]. 厦门大学学报(自然科学版), 2020, 59(6): 1004-1010.
[10]	(Lu Kaihua, Li Zhenghua, Zhang Min. Data Construction and Benchmark Method Comparison for Chinese Hypernym-Hyponym Relation Classification[J]. Journal of Xiamen University (Natural Science Edition), 2020, 59(6): 1004-1010.)
[11]	丁晟春, 侯琳琳, 王颖. 基于电商数据的产品知识图谱构建研究[J]. 数据分析与知识发现, 2019, 3(3): 45-56.
[11]	(Ding Shengchun, Hou Linlin, Wang Ying. Product Knowledge Map Construction Based on the E-Commerce Data[J]. Data Analysis and Knowledge Discovery, 2019, 3(3): 45-56.)
[12]	黄毅, 王庆林, 刘禹. 一种基于条件随机场的领域术语上下位关系获取方法[J]. 中南大学学报(自然科学版), 2013, 44(S2): 355-359.
[12]	(Huang Yi, Wang Qinglin, Liu Yu. An Acquisition Method of Domain-Specific Terminological Hyponymy Based on CRF[J]. Journal of Central South University (Science and Technology), 2013, 44(S2): 355-359.)
[13]	马晓军, 郭剑毅, 线岩团, 等. 结合词向量和Bootstrapping的领域实体上下位关系获取与组织[J]. 计算机科学, 2018, 45(1): 67-72.
[13]	(Ma Xiaojun, Guo Jianyi, Xian Yantuan, et al. Entity Hyponymy Acquisition and Organization Combining Word Embedding and Bootstrapping in Special Domain[J]. Computer Science, 2018, 45(1): 67-72.)
[14]	孙佳伟, 李正华, 陈文亮, 等. 基于词模式嵌入的词语上下位关系分类[J]. 北京大学学报(自然科学版), 2019, 55(1): 1-7.
[14]	(Sun Jiawei, Li Zhenghua, Chen Wenliang, et al. Hypernym Relation Classification Based on Word Pattern[J]. Acta Scientiarum Naturalium Universitatis Pekinensis, 2019, 55(1): 1-7.)
[15]	吴婷, 李明扬, 孔芳. 基于同义推理的篇章级实体上下位关系语料库构建[J]. 中文信息学报, 2020, 34(4): 38-46.
[15]	(Wu Ting, Li Mingyang, Kong Fang. Construction of Textual Entity Hypernymy Corpus Based on Synonymy Reasoning[J]. Journal of Chinese Information Processing, 2020, 34(4): 38-46.)
[16]	汪诚愚, 何晓丰, 宫学庆, 等. 面向上下位关系预测的词嵌入投影模型[J]. 计算机学报, 2020, 43(5): 868-883.
[16]	(Wang Chengyu, He Xiaofeng, Gong Xueqing, et al. Word Embedding Projection Models for Hypernymy Relation Prediction[J]. Chinese Journal of Computers, 2020, 43(5): 868-883.)
[17]	Kotlerman L, Dagan I, Szpektor I, et al. Directional Distributional Similarity for Lexical Expansion [C]//Proceedings of ACL-IJCNLP 2009 Conference (Short Papers). 2009: 69-72.
[18]	王思丽, 祝忠明, 杨恒, 等. 基于模式和投影学习的领域概念上下位关系自动识别研究[J]. 数据分析与知识发现, 2020, 4(11): 15-25.
[18]	(Wang Sili, Zhu Zhongming, Yang Heng, et al. Automatically Identifying Hypernym-Hyponym Relations of Domain Concepts with Patterns and Projection Learning[J]. Data Analysis and Knowledge Discovery, 2020, 4(11): 15-25.)
[19]	吴志祥, 王昊, 王雪颖, 等. 基于奇异值分解的专利术语层次关系解析研究[J]. 情报学报, 2017, 36(5): 473-483.
[19]	(Wu Zhixiang, Wang Hao, Wang Xueying, et al. Study on Chinese Patent Terms Hierarchy Parse Based on Singular Value Decomposition[J]. Journal of the China Society for Scientific and Technical Information, 2017, 36(5): 473-483.)
[20]	Kozareva Z, Hovy E. A Semi-Supervised Method to Learn and Construct Taxonomies Using the Web [C]//Proceedings of the 2010 Conference on Empirical Methods in Natural Language Processing. 2010: 1110-1118.
[21]	Fu R J, Qin B, Liu T. Exploiting Multiple Sources for Open-Domain Hypernym Discovery [C]//Proceedings of the 2013 Conference on Empirical Methods in Natural Language Processing. 2013: 1224-1234.
[22]	刘琦, 肖仰华, 汪卫. 一种面向海量中文文本的典型类属关系识别方法[J]. 计算机工程, 2015, 41(2): 26-30.
[22]	(Liu Qi, Xiao Yanghua, Wang Wei. A Recognition Approach of Typical Generic Relationship for Massive Chinese Text[J]. Computer Engineering [J]. Computer Engineering, 2015, 41(2): 26-30.)
[23]	甘丽新, 万常选, 刘德喜, 等. 基于句法语义特征的中文实体关系抽取[J]. 计算机研究与发展, 2016, 53(2): 284-302.
[23]	(Gan Lixin, Wan Changxuan, Liu Dexi, et al. Chinese Named Entity Relation Extraction Based on Syntactic and Semantic Features[J]. Computer Research and Development, 2016, 53(2): 284-302.)
[24]	Bansal M, Burkett D, de Melo G, et al. Structured Learning for Taxonomy Induction with Belief Propagation [C]// Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2014: 1041-1051.
[25]	段利国, 徐庆, 李爱萍, 等. 实体词语义信息对中文实体关系抽取的作用研究[J]. 计算机应用研究, 2017, 34(1): 141-146.
[25]	(Duan Liguo, Xu Qing, Li Aiping, et al. Research on Effect of Entities Semantic Information on Chinese Entity Relation Extraction[J]. Computer Application Research, 2017, 34(1): 141-146.)
[26]	Mikolov T, Yih S W T, Zweig G. Linguistic Regularities in Continuous Space Word Representations [C]//Proceedings of the 2013 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2013.
[27]	付瑞吉. 开放域命名实体识别及其层次化类别获取[D]. 哈尔滨: 哈尔滨工业大学, 2014.
[27]	(Fu Ruiji. Open-Domain Named Entity Recognition and Hierarchical Category Acquisition[D]. Harbin: Harbin Institute of Technology, 2014.)
[28]	Lai S W, Liu K, He S Z, et al. How to Generate a Good Word Embedding[J]. IEEE Intelligent Systems, 2016: 31(6): 5-14.

[1]	王一钒,李博,史话,苗威,姜斌. 古汉语实体关系联合抽取的标注方法*[J]. 数据分析与知识发现, 2021, 5(9): 63-74.
[2]	马江微, 吕学强, 游新冬, 肖刚, 韩君妹. 融合BERT与关系位置特征的军事领域关系抽取方法^*[J]. 数据分析与知识发现, 2021, 5(8): 1-12.
[3]	张建东, 陈仕吉, 徐小婷, 左文革. 基于词向量的PDF表格抽取研究^*[J]. 数据分析与知识发现, 2021, 5(8): 34-44.
[4]	魏庭新,柏文雷,曲维光. 词向量和语义知识相结合的汉语未登录词语义预测研究*[J]. 数据分析与知识发现, 2020, 4(6): 109-117.
[5]	王思丽,祝忠明,杨恒,刘巍. 基于模式和投影学习的领域概念上下位关系自动识别研究*[J]. 数据分析与知识发现, 2020, 4(11): 15-25.
[6]	聂卉,何欢. 引入词向量的隐性特征识别研究*[J]. 数据分析与知识发现, 2020, 4(1): 99-110.
[7]	聂维民,陈永洲,马静. 融合多粒度信息的文本向量表示模型 ^*[J]. 数据分析与知识发现, 2019, 3(9): 45-52.
[8]	俞琰,陈磊,姜金德,赵乃瑄. 结合词向量和统计特征的专利相似度测量方法 ^*[J]. 数据分析与知识发现, 2019, 3(9): 53-59.
[9]	邵云飞,刘东苏. 基于类别特征扩展的短文本分类方法研究 ^*[J]. 数据分析与知识发现, 2019, 3(9): 60-67.
[10]	文秀贤,徐健. 基于用户评论的商品特征提取及特征价格研究 ^*[J]. 数据分析与知识发现, 2019, 3(7): 42-51.
[11]	余本功,陈杨楠,杨颖. 基于nBD-SVM模型的投诉短文本分类^*[J]. 数据分析与知识发现, 2019, 3(5): 77-85.
[12]	吴粤敏,丁港归,胡滨. 基于注意力机制的农业金融文本关系抽取研究^*[J]. 数据分析与知识发现, 2019, 3(5): 86-92.
[13]	张佩瑶,刘东苏. 基于词向量和BTM的短文本话题演化分析^*[J]. 数据分析与知识发现, 2019, 3(3): 95-101.
[14]	李慧,柴亚青. 基于卷积神经网络的细粒度情感分析方法^*[J]. 数据分析与知识发现, 2019, 3(1): 95-103.
[15]	李心蕾, 王昊, 刘小敏, 邓三鸿. 面向微博短文本分类的文本向量化方法比较研究^*[J]. 数据分析与知识发现, 2018, 2(8): 41-50.

Viewed

Full text

Abstract

Cited

Shared

Discussed