基于多重注意力的金融事件大数据精准画像

doi:10.3778/j.issn.1673-9418.2007001

摘要/Abstract

摘要：

随着知识图谱技术的兴起，利用金融事件大数据中的实体关系来构建金融事件的精准画像成为一个重要的研究方向。通过对金融事件大数据信息进行精准画像，人们可以详细分析金融事件大数据中的属性关系，全面了解金融事件的发展态势，从而分析金融市场发展趋势与规律。然而金融事件大数据存在文本数据噪音多、中文语义复杂以及实体关系抽取不准确等研究难点，导致金融事件大数据画像不精准。针对以上问题，提出一种基于多重注意力的金融事件大数据实体关系抽取算法（REMA）来进行实体关系的抽取，然后利用抽取的实体关系信息结合知识图谱技术进行金融事件大数据的精准画像。实验结果表明：在不使用外部资源的情况下，该算法在金融事件大数据中实体关系抽取的准确率、召回率以及F1值比其他对比算法均有所提升，其中准确率提升了5.6个百分点，召回率提升了4.6个百分点，F1值提升了5个百分点。

关键词: 金融事件大数据, 精准画像, 多重注意力机制, 实体关系抽取

Abstract:

With the rise of knowledge graph technology, the use of entity relationships in big data of financial event to construct accurate portraits of financial events has become an important research direction. By making accurate portraits of big data information on financial events, people can analyze attribute relationships in big data of financial events in detail, fully understand the development trend of financial events, and thus analyze the trends and laws of financial market development. However, there are many research difficulties in financial event big data, such as large text data noise, complex Chinese semantics, and inaccurate extraction of entity relationships, resulting in inaccurate portraits of financial events. In response to the above problems, this paper proposes a financial event big data entity relationship extraction algorithm based on multiple attention mechanism (REMA) to extract entity relationships, and then uses the extracted entity relationship information combined with knowledge graph technology to perform accurate financial event big data portrait. The experimental results show that the precision, recall and F1-score of the entity relationship extraction in the big data of financial events are improved compared with other comparison algorithms without using external resources. Among them, the improvement of precision is 5.6 percentage points, the improvement of recall is 4.6 percentage points, and the improvement of F1-score is 5 percentage points.

Key words: big data of financial event, accurate portrait, multiple attention mechanism, entity relationship extraction

陈剑南, 杜军平, 薛哲, 寇菲菲. 基于多重注意力的金融事件大数据精准画像[J]. 计算机科学与探索, 2021, 15(7): 1237-1244.

CHEN Jiannan, DU Junping, XUE Zhe, KOU Feifei. Accurate Portrait of Big Data of Financial Events Based on Multiple Attention Mechanism[J]. Journal of Frontiers of Computer Science and Technology, 2021, 15(7): 1237-1244.

参考文献

[1] E H H, ZHANG W J, XIAO S Q. Survey of entity relationship extraction based on deep learning[J]. Journal of Software, 2019, 30(6): 1793-1818.
鄂海红, 张文静, 肖思琪. 深度学习实体关系抽取研究综述[J]. 软件学报, 2019, 30(6): 1793-1818.
[2] ETZIONI O, CAFARELLA M, DOWNEY D, et al. Unsupervised named-entity extraction from the web: an experimental study[J]. Artificial Intelligence, 2005, 165(1): 91-134.
[3] BARTOLI A, DE LORENZO A, MEDVET E, et al. Active learning of regular expressions for entity extraction[J]. IEEE Transactions on Cybernetics, 2017, 48(3): 1067-1080.
[4] GREENFIELD K, CACERES R S, COURY M, et al. A reverse approach to named entity extraction and linking in microposts[C]//Proceedings of the 6th Workshop on “Making Sense of Microposts” Co-located with the 25th International World Wide Web Conference, Montréal, Apr 11, 2016: 67-69.
[5] EL ENAS M F. A framework for extracting biological relations from different resources[J]. International Journal of Computer Applications, 2015, 119(3): 21044.
[6] HUANG K Y. A research on weakly supervised relation extraction[D]. Beijing: Beijing University of Posts and Telecommunications, 2018.
黄恺瑜. 弱监督条件下的实体关系抽取探究[D]. 北京: 北京邮电大学, 2018.
[7] HUANG J M. Unsupervised relation extraction based on matrix factorization[D]. Wuhan: Wuhan University, 2018.
黄济民. 基于矩阵分解的无监督实体关系提取方法研究[D]. 武汉: 武汉大学, 2018.
[8] ZENG D J, LIU K, LAI S W, et al. Relation classification via convolutional deep neural network[C]//Proceedings of the 25th International Conference on Computational Linguistics, Dublin, Aug 23-29, 2014. Stroudsburg: ACL, 2014: 2335-2344.
[9] ZENG D J, LIU K, CHEN Y B, et al. Distant supervision for relation extraction via piecewise convolutional neural networks[C]//Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing, Lisbon, Sep 17-21, 2015. Stroudsburg: ACL, 2015: 1753-1762.
[10] HUANG Y Y, WANG W Y. Deep residual learning for weakly-supervised relation extraction[J]. arXiv:1707.08866, 2017.
[11] KATIYAR A, CARDIE C. Going out on a limb: joint extraction of entity mentions and relations without dependency trees[C]//Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics, Vancouver, Jul 30- Aug 4, 2017. Stroudsburg: ACL, 2017: 917-928.
[12] NGUYEN T H, GRISHMAN R. Relation extraction: perspective from convolutional neural networks[C]//Proceedings of the 1st Workshop on Vector Space Modeling for Natural Language Processing, Denver, Jun 5, 2015. Stroudsburg: ACL, 2015: 39-48.
[13] ZHOU P, SHI W, TIAN J, et al. Attention-based bidirectional long short-term memory networks for relation classification[C]//Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics, Berlin, Aug 7-12, 2016. Stroudsburg: ACL, 2016: 207-212.
[14] MIWA M, BANSAL M. End-to-end relation extraction using LSTMs on sequences and tree structures[J]. arXiv:1601. 00770, 2016.
[15] ZHENG S, WANG F, BAO H, et al. Joint extraction of entities and relations based on a novel tagging scheme[J]. arXiv:1706.05075, 2017.