Graph Matching Text Classification Based on KG

doi:10.11871/jfdc.issn.2096-742X.2022.02.004

Abstract

Abstract:

[Objective] In the field of natural language processing (NLP), text classification is a well-developed task that benefits many downstream tasks such as article retrieval, recommendation systems, and question answering. Inspired by the role of knowledge graph (KG) in the field of text reasoning, this article explores the way of utilizing the reasoning capability of KG to support text classification. [Methods] This paper proposes graph matching text classification based on KG. Specifically, this paper constructs the corresponding KG for each class according to the task. The model utilizes the semantics and structure information of these KGs to evaluate the relevance of the text to each class’s KG and then classifies the text by synthesizing the evaluations of all KGs. [Conclusions] In order to prove the effectiveness of our proposed model, this paper builds all KGs of classes in two datasets and conducts experiments on those datasets. The experiment results prove that the proposed model achieves high accuracy under the premise of allowing some data to be rejected and further promotes the application of the method.

Key words: text classification, knowledge graph, graph matching, knowledge graph construction, information extraction

LAN Ge,WANG Jinyu,SUN Yufei,ZHANG Yuzhi. Graph Matching Text Classification Based on KG[J]. Frontiers of Data and Computing, 2022, 4(2): 39-49.

Figures/Tables 6

Fig.1

Fig.2

Table 1

Table 2

Table 3

Fig.3

References 32

[1]	Chen J, Li K, Rong H, et al. A disease diagnosis and treatment recommendation system based on big data min-ing and cloud computing[J]. Information Sciences, 2018, 435(1):124-149. doi: 10.1016/j.ins.2018.01.001
[2]	Reddy S, Lapata M, Steedman M. Large-scale semantic parsing without question-answer pairs[J]. Transactions of the Association for Computational Linguistics, 2014, 2(1):377-392. doi: 10.1162/tacl_a_00190
[3]	Chen X, Sun C, Wang J, et al. Aspect sentiment class-ification with document-level sentiment preference mod-eling[C]// Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, 2020: 3667-3677.
[4]	巫影, 陈定方, 唐小兵, 等. 神经网络综述[J]. 科技进步与对策, 2002, 19(6): 133-134.
[5]	Wang S, Huang M, Deng Z. Densely connected CNN with multi-scale feature attention for text classification[C]// IJCAI, 2018: 4468-4474.
[6]	Banerjee I, Ling Y, Chen M C, et al. Comparative eff-ectiveness of convolutional neural network (CNN) and recurrent neural network (RNN) architectures for radiol-ogy text report classification[J]. Artificial intelligence in medicine, 2019, 97(1): 79-88. doi: 10.1016/j.artmed.2018.11.004
[7]	Ji S, Pan S, Cambria E, et al. A survey on knowledge gr-aphs: Representation, acquisition, and applications[J]. IEEE Transactions on Neural Networks and Learning Systems, 2021, 33(2):494-514, doi: 10.1109/TNNLS.2021.3070843
[8]	李蕾, 王楠, 钟义信, 等. 基于语义网络的概念检索研究与实现[J]. 情报学报, 2000, 19(5): 525-531.
[9]	Friedman N, Geiger D, Goldszmidt M, et al. Bayesian Network Classifiers[J]. Machine Learning, 1997, 29 (2): 131-163. doi: 10.1023/A:1007465528199
[10]	Patrick E A, Fischer III F P. A generalized k-nearest neig-hbor rule[J]. Information and control, 1970, 16(2): 128-152. doi: 10.1016/S0019-9958(70)90081-1
[11]	Schuldt C, Laptev I, Caputo B. Recognizing human acti-ons: a local SVM approach[C]// Proceedings of the 17th International Conference on Pattern Recognition, ICPR 2004, IEEE, 2004, 3: 32-36.
[12]	Harris Z S. Distributional structure[J]. Word, 1954, 10(2-3): 146-162. doi: 10.1080/00437956.1954.11659520
[13]	Jing L P, Huang H K, Shi H B. Improved feature selection approach TFIDF in text mining[C]// Proceedings. Inter-national Conference on Machine Learning and Cyberne-tics. IEEE, 2002, 2: 944-946.
[14]	Mikolov T, Chen K, Corrado G, et al. Efficient estimation of word representations in vector space[C]. ICLR. [S.l.: s.n.], 2013: 1-12.
[15]	Sturman D J, Zeltzer D. A survey of glove-based input[J]. IEEE Computer graphics and Applications, 1994, 14(1): 30-39. doi: 10.1109/38.250916
[16]	Liu G, Guo J. Bidirectional LSTM with attention mech-anism and convolutional layer for text classification[J]. Neurocomputing, 2019, 337(1): 325-338. doi: 10.1016/j.neucom.2019.01.078
[17]	Yu S, Su J, Luo D. Improving bert-based text class-ificat-ion with auxiliary sentence and domain knowledge[J]. IEEE Access, 2019, 7(1): 176600-176612. doi: 10.1109/ACCESS.2019.2953990
[18]	Yao L, Mao C, Luo Y. Graph convolutional networks for text classification[C]// Proceedings of the AAAI confer-ence on artificial intelligence, 2019, 33(01): 7370-7377.
[19]	M. Abdollahi, X. Gao, Y. Mei, S. Ghosh, J. Li. An ont-ology-based two-stage approach to medical text classif-ication with feature selection byparticle swarm optimisat-ion[C]. in: 2019 IEEE Congress on Evolutionary Com-putation (CEC), IEEE, 2019: 119-126.
[20]	M. Mishra, J. Huan, S. Bleik, M. Song. Biomedical text categorization with concept graph representations using a controlled vocabulary[C]. in: Proceedings of the 11th International Workshop on Data Mining in Bioin-formatics, 2012: 26-32.
[21]	Yao L, Mao C, Luo Y. Clinical text classification with rule-based features and knowledge-guided convolutional neural networks[J]. BMC medical informatics and deci-sion making, 2019, 19(3): 31-39.
[22]	J. Chen, Y. Hu, J. Liu, Y. Xiao, H. Jiang. Deep short text classification with knowledge powered attention[C]. in: Proceedings of the AAAI Conference on Artificial Intelli-gence, 2019, 33: 6252-6259.
[23]	Pujara J, Miao H, Getoor L, et al. Knowledge graph iden-tification[C]// International semantic web conference. Spr-inger, Berlin, Heidelberg, 2013: 542-557.
[24]	Ren H, Dai H, Dai B, et al. Smore: Knowledge graph completion and multi-hop reasoning in massive know-ledge graphs[Z]. 2021.
[25]	Auer S, Bizer C, Kobilarov G, et al. Dbpedia: A nucleus for a web of open data[M]// The semantic web. Springer, Berlin, Heidelberg, 2007: 722-735.
[26]	Bekoulis G, Deleu J, Demeester T, et al. Adversarial train-ing for multicontext joint entity and relation extraction [Z]. 2019.
[27]	Hailong Jin, Chengjiang Li, Jing Zhang, Lei Hou, Juanzi Li, Peng Zhang. XLORE2: Large-Scale Cross-Lingual Knowledge Graph Construction and Application[J]. Data Intelligence, 2019, 1(01):77-98. doi: 10.1162/dint_a_00003
[28]	阮彤, 孙程琳, 王昊奋, 方之家, 殷亦超. 中医药知识图谱构建与应用[J]. 医学信息学杂志, 2016, 37(04):8-13.
[29]	吕华揆, 洪亮, 马费成. 金融股权知识图谱构建与应用[J]. 数据分析与知识发现, 2020, 4(05):27-37.
[30]	邱轶. 《中国图书馆分类法》部分类目设置的研究[J]. 图书馆建设, 2011, 1(6): 53-56.
[31]	Chen Y. Convolutional neural network for sentence class-ification[D]. University of Waterloo, 2015.
[32]	SUN Y, WANG S, LI Y, et al. Ernie: Enhanced repr-esentation throughknowledge integration[Z]. 2019.

名称	训练集样本数	测试集样本数	类别数
U20	600	56	20
T122	10989	1519	122

模型	准确率	训练样本数	拒绝率
textCNN	0.9262	600	0.2226
ERNIE	0.9670	600	0.2321
ours	0.9803	100	0.2256

模型	准确率	训练样本数	拒绝率
textCNN	0.8114	10989	0.2462
ERNIE	0.7450	10989	0.2409
ours	0.9095	1200	0.2416