Please wait a minute...
Advanced Search
数据分析与知识发现  2018, Vol. 2 Issue (10): 77-83    DOI: 10.11925/infotech.2096-3467.2018.0114
  研究论文 本期目录 | 过刊浏览 | 高级检索 |
基于卷积神经网络与SVM分类器的隐喻识别*
黄孝喜,李晗雨(),王荣波,王小华,谌志群
杭州电子科技大学认知与智能计算研究所 杭州 310018
Recognizing Metaphor with Convolution Neural Network and SVM
Xiaoxi Huang,Hanyu Li(),Rongbo Wang,Xiaohua Wang,Zhiqun Chen
Institute of Cognitive and Intelligent Computing, Hangzhou Dianzi University, Hangzhou 310018, China
全文: PDF(576 KB)   HTML
输出: BibTeX | EndNote (RIS)      
摘要 

【目的】针对中英文的隐喻数据集, 提出一种基于卷积神经网络与SVM分类器的隐喻识别方法。【方法】将实验数据向量化, 结合词性特征和关键词特征作为卷积神经网络的输入, 通过卷积层和池化层提取特征, 应用SVM进行分类。针对卷积神经网络的池化层中特征采样的不完全性, 提出将MaxPooling与MeanPooling组合在一起的改进方法。【结果】相对于直接使用卷积神经网络, 利用本文方法进行隐喻识别的准确率在英文动宾语料、英文形容词-名词词组语料和中文隐喻语料分别提高4.12%、0.84%和4.50%。【局限】中文分词不准确, 影响词向量模型训练; 卷积神经网络的层数过少, 影响特征的完整性。【结论】根据中英文数据集上隐喻识别的结果分析, 该方法在两个数据集上都取得了良好效果。

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
黄孝喜
李晗雨
王荣波
王小华
谌志群
关键词 隐喻识别卷积神经网络支持向量机特征提取    
Abstract

[Objective] This paper presents a new method to recognize metaphor, from the Chinese and English datasets. [Methods] First, we mapped the experimental dataset to vector space, which was also input to a convolutional neural network along with the property and keyword features. Then, we extracted the needed features with the help of convolutional and pooled layers, as well as classified them using SVM. Finally, we combined the Max-Pooling and Mean-Pooling to improve the extracted features’ accuracy. [Results] Compared with the traditional models, our method increased the accuracy of extracted features from the corpus of English verb-object, English adjective-noun and Chinese metaphor by 4.12%, 0.84% and 4.50% respectively. [Limitations] The Chinese word segmentation affects the training of word vector model. We need to add more layers to the convolutional neural networks. [Conclusions] The proposed method could effectively identify metaphor from Chinese and English corpus.

Key wordsMetaphor Recognition    Convolution Neural Network    Support Vector Machines    Feature Extraction
收稿日期: 2018-01-29     
基金资助:*本文系教育部人文社会科学研究规划基金项目“融合深度神经网络模型的汉语隐喻计算研究”(项目编号: 18YJA740016)和教育部人文社会科学研究青年基金项目“基于语义相关性的汉语组块切分模型研究”(项目编号: 12YJCZH201)的研究成果之一
引用本文:   
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.
Xiaoxi Huang,Hanyu Li,Rongbo Wang,Xiaohua Wang,Zhiqun Chen. Recognizing Metaphor with Convolution Neural Network and SVM. Data Analysis and Knowledge Discovery, DOI:10.11925/infotech.2096-3467.2018.0114.
链接本文:  
http://manu44.magtech.com.cn/Jwk_infotech_wk3/CN/10.11925/infotech.2096-3467.2018.0114
图1  基于SVM与CNN的隐喻识别的整体框架
图2  卷积神经网络结构
图3  Tanh和ReLU对比
Verb Noun Class Relation
See development
Live dream
Envy eat
Break window
Boy cry
Paint dry
Metaphorical
Metaphorical
Metaphorical
Literal
Literal
Literal
VO
VO
SV
VO
SV
SV
表1  TSV中动词隐喻的主语-动词或动词-宾语关系
Metaphorical Literal
bright smile
bushy eyebrows
cautious smile
dark history
deep faith
desolate beauty
economic battle
fading memory
faint impression
blue fence
blinding light
biting dog
bright sun
bright light
burning tree
burning arm
dark face
dirty hands
表2  TSV-TRAIN中的形容词-名词短语
实验 准确率
CNN-sentence-eng 81.80%
CNN_SVM -sentence-eng 87.23%
CNN-Word/Pos-eng 86.00%
CNN_SVM -Word/Pos-eng 90.12%
CNN-AN-eng 86.36%
CNN_SVM -AN-eng 87.20%
Rei等[13] 83.00%
表3  英文语料隐喻识别
实验 准确率
CNN-sentence-ch 72.5%
CNN_SVM -sentence-ch 77.00%
表4  中文语料隐喻识别
[1] Wilks Y.A Preferential, Pattern-seeking, Semantics for Natural Language Inference[A]// Words and Intelligence I. Text, Speech and Language Technology[M]. Springer, 2007.
[2] Fass D.Met*: A Method for Discriminating Metonymy and Metaphor by Computer[J]. Computational Linguistics, 1991, 17(1): 49-90.
[3] Neuman Y, Assaf D, Cohen Y, et al.Metaphor Identification in Large Texts Corpora[J]. PLoS One, 2013, 8(4): e62343.
doi: 10.1371/journal.pone.0062343 pmid: 3639214
[4] Shutova E, Sun L, Korgonen A.Metaphor Identification Using Verb and Nouns Clustering[C]// Proceedings of the 23rd International Conference on Computational Linguistics. 2010.
[5] Hovy D, Srivastava S, Kumar S, et al.Identifying Metaphorical Word Use with Tree Kernels[C]// Proceedings of the 1st Workshop on Metaphor in NLP. 2013.
[6] Rai S, Chakraverty S, Tayal D K.Supervised Metaphor Detection Using Conditional Random Fields[C]// Proceedings of the 4th Workshop on Metaphor in NLP. 2016.
[7] Tsvetkov Y, Boytsov L, Gershman A, et al.Metaphor Detection with Cross-Lingual Model Transfer[C]// Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics. 2014.
[8] Kalchbrenner N, Grefenstette E, Blunsom P.A Convolutional Neural Network for Modelling Sentences[OL]. arXiv Preprint, arXiv: 1404.2188.
[9] Graves A, Fernández S, Schmidhuber J.Bidirectional LSTM Networks for Improved Phoneme Classification and Recognition[A]// Artificial Neural Networks: Formal Models and Their Applications[M]. Springer, 2005.
[10] Graves A, Schmidhuber J.Framewise Phoneme Classification with Bidirectional LSTM and Other Neural Network Architectures[J]. Neural Networks, 2005, 18:602-610.
doi: 10.1016/j.neunet.2005.06.042 pmid: 16112549
[11] Dinh E L D, Gurevych I. Token-Level Metaphor Detection Using Neural Networks[C]// Proceedings of the 4th Workshop on Metaphor in NLP. 2016.
[12] Bizzoni Y, Chatzikyriakidis S, Ghanimifard M.“Deep” Learning: Detecting Metaphoricity in Adjective-Noun Pairs[C]// Proceedings of the Workshop on Stylistic Variation. 2017: 43-52.
[13] Rei M, Bulat L, Kiela D, et al.Grasping the Finer Point: A Supervised Similarity Network for Metaphor Detection[C]// Proceedings of EMNLP. 2017: 1537-1546.
[14] 王治敏, 王厚峰, 俞士汶. 基于机器学习方法的汉语名词隐喻识别[J]. 高技术通讯, 2006, 17(6): 575-580.
doi: 10.3321/j.issn:1002-0470.2007.06.005
(Wang Zhimin, Wang Houfeng, Yu Shiwen.Chinese Nominal Metaphor Recognition Based on Machine Learning[J]. Chinese High Technology Letter, 2006, 17(6): 575-580.)
[15] 徐扬. 基于最大熵模型的汉语隐喻现象识别[J]. 计算机工程和科学, 2007, 29(4): 95-103.
(Xu Yang.Recognition of the Chinese Metaphor Phenomena Based on the Maximum Entropy Model[J]. Computer Engineering and Science, 2007, 29(4): 95-103.)
[16] 李斌, 于丽丽, 石民, 等. “像”的明喻计算[J]. 中文信息学报, 2008, 22(6): 27-32.
(Li Bin, Yu Lili, Shi Min, et al.Computation of Chinese Simile with “Xiang”[J]. Journal of Chinese Information Processing, 2008, 22(6): 27-32.)
[17] 黄孝喜. 隐喻机器理解的若干关键问题研究[D]. 杭州: 浙江大学, 2009.
(Huang Xiaoxi.Research on Some Key Issues of Metaphor Computation[D]. Hangzhou: Zhejiang University, 2009.)
[18] Kim Y.Convolutional Neural Networks for Sentence Classification [OL]. arXiv Preprint, arXiv: 1408.5882.
[19] Mikolov T, Chen K, Corrado G, et al.Efficient Estimation of Word Representations in Vector Space[OL]. arXiv Preprint, arXiv: 1301.3781.
[20] Mikolov T, Sutskever I, Chen K, et al.Distributed Representations of Words and Phrases and Their Com- positionality[A]// Advances in Neural Information Processing Systems[M]. Springer, 2013.
[21] Lécun Y, Bottou L, Bengio Y, et al.Gradient-based Learning Applied to Document Recognition[J]. Proceedings of the IEEE, 1998, 86(11): 2278-2324.
doi: 10.1109/5.726791
[22] 李航. 统计学习方法[M]. 第1版. 北京: 清华大学出版社, 2016: 95-123.
(Li Hang.Statistical Learning Method[M]. The 1st Edition. Beijing: Tsinghua University Publishing House, 2016: 95-123.)
[1] 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.
[2] 侯君,刘魁,李千目. 基于ESSVM的分类推荐*[J]. 数据分析与知识发现, 2018, 2(3): 9-21.
[3] 李伟卿,王伟军. 基于大规模评论数据的产品特征词典构建方法研究*[J]. 数据分析与知识发现, 2018, 2(1): 41-50.
[4] 李昌兵,庞崇鹏,李美平. 基于权重的Apriori算法在文本统计特征提取方法中的应用*[J]. 数据分析与知识发现, 2017, 1(9): 83-89.
[5] 曾金,陆伟,丁恒,陈海华. 基于图像语义的用户兴趣建模*[J]. 数据分析与知识发现, 2017, 1(4): 76-83.
[6] 田世海,吕德丽. 改进潜在语义分析和支持向量机算法用于突发安全事件舆情预警*[J]. 数据分析与知识发现, 2017, 1(2): 11-18.
[7] 杨爽,陈芬. 基于SVM多特征融合的微博情感多级分类研究*[J]. 数据分析与知识发现, 2017, 1(2): 73-79.
[8] 刘红光,马双刚,刘桂锋. 基于降噪自动编码器的中文新闻文本分类方法研究*[J]. 现代图书情报技术, 2016, 32(6): 12-19.
[9] 张晔,张晗,尹玢璨,赵玉虹. 基于电子病历利用支持向量机构建疾病预测模型*——以重度急性胰腺炎早期预警为例[J]. 现代图书情报技术, 2016, 32(2): 83-89.
[10] 张策,都云程,梁然. 采用URL特征的Hub网页识别方法研究*[J]. 现代图书情报技术, 2016, 32(1): 24-31.
[11] 杜思奇, 李红莲, 吕学强. 汉语组块分析在产品特征提取中的应用研究[J]. 现代图书情报技术, 2015, 31(9): 26-30.
[12] 黄孝喜, 张华, 陆蓓, 王荣波, 吴铤. 一种基于词语抽象度的汉语隐喻识别方法[J]. 现代图书情报技术, 2015, 31(4): 34-40.
[13] 何跃, 宋灵犀, 齐丽云. 负面事件中的品牌网络口碑溢出效应研究——以“圆通夺命快递”事件为例[J]. 现代图书情报技术, 2015, 31(10): 58-64.
[14] 胡吉明, 陈果. 超球支持向量机文本分类方法改进[J]. 现代图书情报技术, 2014, 30(9): 74-80.
[15] 路永和, 梁明辉. 遗传算法在改进文本特征提取方法中的应用[J]. 现代图书情报技术, 2014, 30(4): 48-57.
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
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
电话/传真:(010)82626611-6626,82624938
E-mail:jishu@mail.las.ac.cn