Please wait a minute...
Advanced Search
数据分析与知识发现  2021, Vol. 5 Issue (9): 75-84
     研究论文 本期目录 | 过刊浏览 | 高级检索 |
1中国医学科学院/北京协和医学院医学信息研究所/图书馆 北京 100020
2中央财经大学金融学院 北京 102206
3厦门大学信息学院 厦门 361005
Classification Model for Medical Entity Relations with Convolutional Neural Network
Fan Shaoping1,Zhao Yuxuan2,An Xinying1,Wu Qingqiang3()
1Institute of Medical Information / Medical Library, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing 100020, China
2School of Finance, Central University of Finance and Economics, Beijing 102206, China
3School of Informatics, Xiamen University, Xiamen 361005, China
全文: PDF (1073 KB)   HTML ( 20
输出: BibTeX | EndNote (RIS)      

【目的】 为提升关系分类模型性能,降低特征计算复杂性,提出一种融合多特征嵌入的卷积神经网络实体关系分类模型。【方法】 参考已有研究的主要嵌入特征,提出融合位置和词汇级特征嵌入的卷积神经网络实体关系分类模型,并给出特征的计算表示方法,上述特征无需复杂计算算法,提高了模型性能。【结果】 所提模型在生物医学领域语料库AIMed、GENIA和ChemProt上F1值分别为0.734 2、0.976 4和0.890 0,在GENIA和ChemProt上实现了当前最佳性能。【局限】 尚未融入生物医学领域先验知识等领域特色的特征。【结论】 融合多特征嵌入的卷积神经网络实体关系分类模型具有良好的分类效果,可为生物医学领域关系抽取和知识库研究提供参考。

E-mail Alert
关键词 关系分类卷积神经网络位置特征词汇级特征    

[Objective] This paper proposes a new classification model for entity relationship based on the Convolutional Neural Network (CNN) with multi-features embedding, aiming to improve the classification results and simplify feature calculation. [Objective] Based on the existing algorithms of embedded features, our CNN model integrated word positions and lexical features, as well as demonstrated the representation methods for the features. These features did not require complex algorithm calculation, which improved the model's performance. [Results] We examined the proposed model with the Bio-Medical corpus of AIMed, GENIA and ChemProt. The F1 scores were 0.7342, 0.9764 and 0.8900, respectively. This model yielded the best results with the GENIA and ChemProt datasets. [Limitations] Our model did not include the prior domain knowledge from biomedical field. [Conclusions] The proposed model could effectively conduct entity relationship classification, which also help the research on relation extraction and knowledgebase construction in bio-medical field.

Key wordsRelation Classification    CNN    Position Features    Lexical Features
收稿日期: 2021-01-07      出版日期: 2021-06-29
ZTFLH:  分类号: G350  
通讯作者: 吴清强     E-mail:
范少萍,赵雨宣,安新颖,吴清强. 基于卷积神经网络的医学实体关系分类模型研究*[J]. 数据分析与知识发现, 2021, 5(9): 75-84.
Fan Shaoping,Zhao Yuxuan,An Xinying,Wu Qingqiang. Classification Model for Medical Entity Relations with Convolutional Neural Network. Data Analysis and Knowledge Discovery, 2021, 5(9): 75-84.
链接本文:      或
句子 实体e1 实体e2 关系:
<e1>1,25D</e1> inhibited <e2>MYC gene</e2> expression and accelerated its protein turnover 1,25D MYC gene inhibit (e1, e2
Table 1  实体关系分类示例
句子 实体e1 实体e2
Demethylation experiments further confirmed that loss of <e1>ALX4</e1> expression was regulated by <e2>CpG island</e2> hypermethylation. ALX4 CpG
Table 2  词汇级特征示例
Fig.1  卷积神经网络结构层次示意图
Fig.2  仅词表示的CNN结构
Fig.3  加入位置特征的CNN结构
Fig.4  加入词汇级特征的CNN结构
语料库 关系名称 关系语句数量 训练集 测试集
AIMed[35] False 4 834 4 861 973
True 1 000
GENIA[36] Protein-Component 1 302 1 547 310
Subunit-Complex 555
ChemProt[37] Activator 2 571 5 363 1 073
Indirect-Downregulator 446
Indirect-Upregulator 3 225
Inhibitor 194
Table 3  语料库规模与数量分布
语料库 模型结构 准确率 F1值
AIMed CNN + Word Representation + Position Features + Lexical Features 0.856 1 0.734 2
GENIA CNN + Word Representation + Position Features + Lexical Features 0.980 6 0.976 4
ChemProt CNN + Word Representation + Position Features + Lexical Features 0.923 6 0.890 0
Table 4  本文模型在AIMed、GENIA和ChemProt语料库上进行语义关系分类的准确率与F1值
Fig.5  AIMed、GENIA和ChemProt语料库在不同CNN模型关系分类效果
语料库 模型 F1值
AIMed 本文模型 0.734 2
Zhang 等[22] (Word, Position, SDP) 0.617 0
Peng等[19] (Word, Position, POS, Chunk, Dependency Information) 0.635 0
Chang等[38](Convolution Tree Kernel) 0.567 0
Hsieh等[41] (LSTMpre 0.769 0
Yadav等[42] (Att-sdpLSTM) 0.932 9
GENIA 本文模型 0.976 4
Ramesh等[40] (SVM + CFR) 0.761 0
ChemProt 本文模型 0.890 0
Corbett等[13] (RNNs + Word) 0.615 1
Lim等[43] (Tree-LSTM: Position +
Syntactic Parse Tree)
0.641 0
Beltagy等[44] (SciBERT) 0.836 4
Table 5  本文模型与其他关系抽取/分类模型性能对比
[1] The Precision Medicine Initiative[EB/OL].[2019-12-01]. .
[2] 科技部关于发布国家重点研发计划精准医学研究等重点专项2016年度项目申报指南的通知[EB/OL]. [2019-12-01]. .
[2] (Notice of the Ministry of Science and Technology on Issuing 2016 Annual Project Application Guidelines for National Key R & D Plan, Precision Medicine Research and Other Key Special Projects [EB/OL]. [2019-12-01].
[3] 刘雷, 王星. 精准医学知识库的构建[J]. 中华医学图书情报杂志, 2018, 27(6):1-9.
[3] ( Liu Lei, Wang Xing. Development of Knowledge Base for Precision Medicine[J]. Chinese Journal of Medical Library and Information Science, 2018, 27(6):1-9.)
[4] Disease Ontology[EB/OL].[2019-12-01]. .
[5] KEGG: Kyoto Encyclopedia of Genes and Genomes[EB/OL].[2019-12-01]. .
[6] PharmGKB[EB/OL].[2019-12-01]. .
[7] Hendrickx I, Kim S N, Kozareva Z, et al. Semeval-2010 Task 8: Multi-Way Classification of Semantic Relations Between Pairs of Nominals [C]// Proceedings of the 5th International Workshop on Semantic Evaluation. 2010: 33-38.
[8] Corpus for Relation Classification in Medical Field[EB/OL].[2021-01-10]. .
[9] Afzal H, Eales J, Stevens R, et al. Mining Semantic Networks of Bioinformatics E-Resources from the Literature[J]. Journal of Biomedical Semantics, 2011, 2 (S1): Article No. S4.
[10] Segura-Bedmar I, Martínez P, de Pablo-Sanchez C. Using a Shallow Linguistic Kernel for Drug-Drug Interaction Extraction[J]. Journal of Biomedical Informatics, 2011, 44(5):789-804.
doi: 10.1016/j.jbi.2011.04.005 pmid: 21545845
[11] Zhao Z H, Yang Z H, Luo L, et al. Drug-Drug Interaction Extraction from Biomedical Literature Using Syntax Convolutional Neural Network[J]. Bioinformatics, 2016, 32(22):3444-3453.
[12] Zhao Z H, Yang Z H, Sun C, et al. A Hybrid Protein-Protein Interaction Triple Extraction Method for Biomedical Literature [C]//Proceedings of 2017 IEEE International Conference on Bioinformatics and Biomedicine (BIBM). IEEE, 2017.
[13] Corbett P, Boyle J. Improving the Learning of Chemical-Protein Interactions from Literature Using Transfer Learning and Specialized Word Embeddings[J]. Database, 2018. DOI: 10.1093/database/bay066.
doi: 10.1093/database/bay066
[14] Yan X, Mou L L, Li G, et al. Classifying Relations via Long Short Term Memory Networks along Shortest Dependency Paths [C]//Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing. 2015: 1785-1794.
[15] 王天时. 基于特征嵌入表示的文本分类方法研究[D]. 济南: 山东师范大学, 2020.
[15] ( Wang Tianshi. Research on Text Classification Method Based on Feature Embedding Representation[D]. Ji'nan: Shandong Normal University, 2020.)
[16] Lee J, Seo S, Choi Y S. Semantic Relation Classification via Bidirectional LSTM Networks with Entity-Aware Attention Using Latent Entity Typing[J]. Symmetry, 2019, 11(6):785.
doi: 10.3390/sym11060785
[17] Sahu S K, Anand A, Oruganty K, et al. Relation Extraction from Clinical Texts Using Domain Invariant Convolutional Neural Network [C]∥Proceedings of the 15th Workshop on Biomedical Natural Language Processing. Stroudsburg: Association for Computational Linguistics, 2016:206-215.
[18] Quan C Q, Hua L, Sun X, et al. Multichannel Convolutional Neural Network for Biological Relation Extraction[J]. BioMed Research International, 2016: 1-10.
[19] Peng Y F, Lu Z Y. Deep Learning for Extracting Protein-Protein Interactions from Biomedical Literature [C]//Proceedings of the BioNLP 2017 Workshop. 2017: 29-38.
[20] Sahu S K, Anand A. Drug-Drug Interaction Extraction from Biomedical Texts Using Long Short Term Memory Network[J]. Journal of Biomedical Informatics, 2018, 86:15-24.
doi: 10.1016/j.jbi.2018.08.005
[21] Peng Y F, Rios A, Kavuluru R, et al. Extracting Chemical-Protein Relations with Ensembles of SVM and Deep Learning Models[J]. Database the Journal of Biological Database & Curation, DOI: 10.1093/database/bay073.
doi: 10.1093/database/bay073
[22] Zhang Y J, Lin H F, Yang Z H, et al. A Hybrid Model Based on Neural Networks for Biomedical Relation Extraction[J]. Journal of Biomedical Informatics, 2018, 81:83-92.
doi: 10.1016/j.jbi.2018.03.011
[23] 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: Technical Papers. 2014: 2335-2344.
[24] Socher R, Huval B, Manning C D, et al. Semantic Compositionality Through Recursive Matrix-Vector Spaces [C]// Proceedings of the 2012 Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning. 2012:1201-1211.
[25] 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. 2015: 39-48.
[26] Nguyen T H, Grishman R. Employing Word Representations and Regularization for Domain Adaptation of Relation Extraction [C]//Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers). 2014:68-74.
[27] Choi S P. Extraction of Protein-Protein Interactions (PPIs) from the Literature by Deep Convolutional Neural Networks with Various Feature Embeddings[J]. Journal of Information Science, 2018, 44(1):60-73.
doi: 10.1177/0165551516673485
[28] Porumb M, Barbantan I, Lemnaru C, et al. REMed: Automatic Relation Extraction from Medical Documents [C]//Proceedings of the 17th International Conference on Information Integration and Web-based Applications & Services. ACM, 2015: 19.
[29] Mikolov T, Chen K, Corrado G, et al. Efficient Estimation of Word Representations in Vector Space[OL]. arXiv Preprint, arXiv: 1301.3781.
[30] Krizhevsky A, Sutskever I, Hinton G. ImageNet Classification with Deep Convolutional Neural Networks[J]. Communications of the ACM, 2017, 60(6):84-90.
doi: 10.1145/3065386
[31] Yang Y M. An Evaluation of Statistical Approaches to Text Categorization[J]. Information Retrieval, 1999, 1(1-2):69-90.
doi: 10.1023/A:1009982220290
[32] Bunescu R, Ge R F, Kate R J, et al. Comparative Experiments on Learning Information Extractors for Proteins and Their Interactions[J]. Artificial Intelligence in Medicine, 2005, 33(2):139-155.
pmid: 15811782
[33] Ohta T, Pyysalo S, Kim J D, et al. A Re-evaluation of Biomedical Named Entity-Term Relations[J]. Journal of Bioinformatics and Computational Biology, 2010, 8(5):917-928.
doi: 10.1142/S0219720010005014
[34] Taboureau O, Nielsen S K, Audouze K, et al. ChemProt: A Disease Chemical Biology Database[J]. Nucleic Acids Research, 2011, 39(S1):D367-D372.
doi: 10.1093/nar/gkq906
[35] AIMed [DB/OL].[2021-01-12]. .
[36] Relation Annotation [EB/OL].[2021-01-12]. .
[37] BioCreative VII [EB/OL]. [2021-01-12]. .
[38] Chang Y C, Chu C H, Su Y C, et al. PIPE: A Protein-Protein Interaction Passage Extraction Module for BioCreative Challenge[J]. Database, DOI: 10.1093/database/baw101.
doi: 10.1093/database/baw101
[39] Björne J, Salakoski T. Generalizing Biomedical Event Extraction [C]//Proceedings of the 2011 BioNLP Shared Task Workshop. 2011: 183-191.
[40] Ramesh B P, Prasad R, Miller T, et al. Automatic Discourse Connective Detection in Biomedical Text[J]. Journal of the American Medical Informatics Association, 2012, 19(5):800-808.
doi: 10.1136/amiajnl-2011-000775
[41] Hsieh Y L, Chang Y C, Chang N W, et al. Identifying Protein-Protein Interactions in Biomedical Literature Using Recurrent Neural Networks with Long Short-Term Memory [C]//Proceedings of the 8th International Joint Conference on Natural Language Processing. 2017.
[42] Yadav S, Ekbal A, Saha S, et al. Feature Assisted Stacked Attentive Shortest Dependency Path Based Bi-LSTM Model for Protein-Protein Interaction[J]. Knowledge-Based Systems, 2019, 166:18-29.
doi: 10.1016/j.knosys.2018.11.020
[43] Lim S, Kang J. Chemical-Gene Relation Extraction Using Recursive Neural Network[J]. Database, DOI: 10.1093/database/bay060.
doi: 10.1093/database/bay060
[44] Beltagy I, Lo K, Cohan A. SciBERT: A Pretrained Language Model for Scientific Text [C]//Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing. 2019: 3615-3620.
[45] 胡正银, 刘蕾蕾, 代冰, 等. 基于领域知识图谱的生命医学学科知识发现探析[J]. 数据分析与知识发现, 2020, 4(11):1-14.
[45] ( Hu Zhengyin, Liu Leilei, Dai Bing, et al. Discovering Subject Knowledge in Life and Medical Sciences with Knowledge Graph[J]. Data Analysis and Knowledge Discovery, 2020, 4(11):1-14.)
[46] Beltagy I, Lo K, Cohan A. SciBERT: Pretrained Contextualized Embeddings for Scientific Text[OL]. arXivPreprint,arXiv: 1903. 10676.
[47] Zhu Y, Li L S, Lu H B, et al. Extracting Drug-Drug Interactions from Texts with BioBERT and Multiple Entity-aware Attentions[J]. Journal of Biomedical Informatics, 2020, 106:103451.
doi: S1532-0464(20)30079-4 pmid: 32454243
[1] 范涛,王昊,吴鹏. 基于图卷积神经网络和依存句法分析的网民负面情感分析研究*[J]. 数据分析与知识发现, 2021, 5(9): 97-106.
[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] 孟镇,王昊,虞为,邓三鸿,张宝隆. 基于特征融合的声乐分类研究*[J]. 数据分析与知识发现, 2021, 5(5): 59-70.
[4] 韩普,张展鹏,张明淘,顾亮. 基于多特征融合的中文疾病名称归一化研究*[J]. 数据分析与知识发现, 2021, 5(5): 83-94.
[5] 董淼, 苏中琪, 周晓北, 兰雪, 崔志刚, 崔雷. 利用Text-CNN改进PubMedBERT在化学诱导性疾病实体关系分类效果的尝试[J]. 数据分析与知识发现, 2021, 5(11): 145-152.
[6] 邱尔丽,何鸿魏,易成岐,李慧颖. 基于字符级CNN技术的公共政策网民支持度研究 *[J]. 数据分析与知识发现, 2020, 4(7): 28-37.
[7] 刘伟江,魏海,运天鹤. 基于卷积神经网络的客户信用评估模型研究*[J]. 数据分析与知识发现, 2020, 4(6): 80-90.
[8] 徐月梅,刘韫文,蔡连侨. 基于深度融合特征的政务微博转发规模预测模型*[J]. 数据分析与知识发现, 2020, 4(2/3): 18-28.
[9] 向菲,谢耀谈. 基于混合采样与迁移学习的患者评论识别模型*[J]. 数据分析与知识发现, 2020, 4(2/3): 39-47.
[10] 彭郴,吕学强,孙宁,张乐,姜肇财,宋黎. 基于CNN的消费品缺陷领域词典构建方法研究*[J]. 数据分析与知识发现, 2020, 4(11): 112-120.
[11] 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.
[12] 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.
[13] 刘勘,陈露. 面向医疗分诊的深度神经网络学习*[J]. 数据分析与知识发现, 2019, 3(6): 99-108.
[14] 徐月梅, 吕思凝, 蔡连侨, 张小娅. 结合卷积神经网络和Topic2Vec的新闻主题演变分析*[J]. 数据分析与知识发现, 2018, 2(9): 31-41.
[15] 黄孝喜, 李晗雨, 王荣波, 王小华, 谌志群. 基于卷积神经网络与SVM分类器的隐喻识别*[J]. 数据分析与知识发现, 2018, 2(10): 77-83.
Full text



版权所有 © 2015 《数据分析与知识发现》编辑部
地址:北京市海淀区中关村北四环西路33号 邮编:100190