Triplet Embedding Convolutional Recurrent Neural Network for Long Text Semantic Analysis

Jingxuan Liu; Ming Zhu; Huajiang Ouyang; Guozi Sun; Huakang Li

doi:10.1007/978-3-031-20891-1_43

Triplet Embedding Convolutional Recurrent Neural Network for Long Text Semantic Analysis

Jingxuan Liu
, Ming Zhu
, Huajiang Ouyang
, Guozi Sun
, Huakang Li^*

^*Corresponding author for this work

School of AI and Advanced Computing

Xi'an Jiaotong-Liverpool University
Nanjing University of Posts and Telecommunications
University of Liverpool

Research output: Chapter in Book or Report/Conference proceeding › Conference Proceeding › peer-review

1 Citation (Scopus)

Abstract

Deep Recurrent Neural Network has an excellent performance in sentence semantic analysis. However, due to the curse of the computational dimensionality, the application in the long text is minimal. Therefore, we propose a Triplet Embedding Convolutional Recurrent Neural Network for long text analysis. Firstly, a triplet from each sentence of the long text. Then the most crucial head entity into the CRNN network, composed of CNN and Bi-GRU networks. Both relation and tail entities are input to a CNN network through three splicing layers. Finally, the output results into the global pooling layer to get the final results. Entity fusion and entity replacement are also used to retain the text’s structural and semantic information before triplet extraction in sentences. We have conducted experiments on a large-scale criminal case dataset. The results show our model significantly improves the judgment prediction task.

Original language	English
Title of host publication	Web Information Systems Engineering – WISE 2022 - 23rd International Conference, Proceedings
Editors	Richard Chbeir, Helen Huang, Fabrizio Silvestri, Yannis Manolopoulos, Yanchun Zhang, Yanchun Zhang
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	607-615
Number of pages	9
ISBN (Print)	9783031208904
DOIs	https://doi.org/10.1007/978-3-031-20891-1_43
Publication status	Published - 2022
Event	23rd International Conference on Web Information Systems Engineering, WISE 2021 - Biarritz, France Duration: 1 Nov 2022 → 3 Nov 2022

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	13724 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	23rd International Conference on Web Information Systems Engineering, WISE 2021
Country/Territory	France
City	Biarritz
Period	1/11/22 → 3/11/22

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 16 Peace, Justice and Strong Institutions

Keywords

CRNN
Entity fusion and entity replacement
Long text analysis
Triplet embedding

Access to Document

10.1007/978-3-031-20891-1_43

Cite this

Liu, J., Zhu, M., Ouyang, H., Sun, G., & Li, H. (2022). Triplet Embedding Convolutional Recurrent Neural Network for Long Text Semantic Analysis. In R. Chbeir, H. Huang, F. Silvestri, Y. Manolopoulos, Y. Zhang, & Y. Zhang (Eds.), Web Information Systems Engineering – WISE 2022 - 23rd International Conference, Proceedings (pp. 607-615). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 13724 LNCS). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-031-20891-1_43

Liu, Jingxuan ; Zhu, Ming ; Ouyang, Huajiang et al. / Triplet Embedding Convolutional Recurrent Neural Network for Long Text Semantic Analysis. Web Information Systems Engineering – WISE 2022 - 23rd International Conference, Proceedings. editor / Richard Chbeir ; Helen Huang ; Fabrizio Silvestri ; Yannis Manolopoulos ; Yanchun Zhang ; Yanchun Zhang. Springer Science and Business Media Deutschland GmbH, 2022. pp. 607-615 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{e6af45e2304c4e059045125622b00a62,

title = "Triplet Embedding Convolutional Recurrent Neural Network for Long Text Semantic Analysis",

abstract = "Deep Recurrent Neural Network has an excellent performance in sentence semantic analysis. However, due to the curse of the computational dimensionality, the application in the long text is minimal. Therefore, we propose a Triplet Embedding Convolutional Recurrent Neural Network for long text analysis. Firstly, a triplet from each sentence of the long text. Then the most crucial head entity into the CRNN network, composed of CNN and Bi-GRU networks. Both relation and tail entities are input to a CNN network through three splicing layers. Finally, the output results into the global pooling layer to get the final results. Entity fusion and entity replacement are also used to retain the text{\textquoteright}s structural and semantic information before triplet extraction in sentences. We have conducted experiments on a large-scale criminal case dataset. The results show our model significantly improves the judgment prediction task.",

keywords = "CRNN, Entity fusion and entity replacement, Long text analysis, Triplet embedding",

author = "Jingxuan Liu and Ming Zhu and Huajiang Ouyang and Guozi Sun and Huakang Li",

note = "Publisher Copyright: {\textcopyright} 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.; 23rd International Conference on Web Information Systems Engineering, WISE 2021 ; Conference date: 01-11-2022 Through 03-11-2022",

year = "2022",

doi = "10.1007/978-3-031-20891-1\_43",

language = "English",

isbn = "9783031208904",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Science and Business Media Deutschland GmbH",

pages = "607--615",

editor = "Richard Chbeir and Helen Huang and Fabrizio Silvestri and Yannis Manolopoulos and Yanchun Zhang and Yanchun Zhang",

booktitle = "Web Information Systems Engineering – WISE 2022 - 23rd International Conference, Proceedings",

}

Liu, J, Zhu, M, Ouyang, H, Sun, G & Li, H 2022, Triplet Embedding Convolutional Recurrent Neural Network for Long Text Semantic Analysis. in R Chbeir, H Huang, F Silvestri, Y Manolopoulos, Y Zhang & Y Zhang (eds), Web Information Systems Engineering – WISE 2022 - 23rd International Conference, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 13724 LNCS, Springer Science and Business Media Deutschland GmbH, pp. 607-615, 23rd International Conference on Web Information Systems Engineering, WISE 2021, Biarritz, France, 1/11/22. https://doi.org/10.1007/978-3-031-20891-1_43

Triplet Embedding Convolutional Recurrent Neural Network for Long Text Semantic Analysis. / Liu, Jingxuan; Zhu, Ming; Ouyang, Huajiang et al.
Web Information Systems Engineering – WISE 2022 - 23rd International Conference, Proceedings. ed. / Richard Chbeir; Helen Huang; Fabrizio Silvestri; Yannis Manolopoulos; Yanchun Zhang; Yanchun Zhang. Springer Science and Business Media Deutschland GmbH, 2022. p. 607-615 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 13724 LNCS).

Research output: Chapter in Book or Report/Conference proceeding › Conference Proceeding › peer-review

TY - GEN

T1 - Triplet Embedding Convolutional Recurrent Neural Network for Long Text Semantic Analysis

AU - Liu, Jingxuan

AU - Zhu, Ming

AU - Ouyang, Huajiang

AU - Sun, Guozi

AU - Li, Huakang

PY - 2022

Y1 - 2022

N2 - Deep Recurrent Neural Network has an excellent performance in sentence semantic analysis. However, due to the curse of the computational dimensionality, the application in the long text is minimal. Therefore, we propose a Triplet Embedding Convolutional Recurrent Neural Network for long text analysis. Firstly, a triplet from each sentence of the long text. Then the most crucial head entity into the CRNN network, composed of CNN and Bi-GRU networks. Both relation and tail entities are input to a CNN network through three splicing layers. Finally, the output results into the global pooling layer to get the final results. Entity fusion and entity replacement are also used to retain the text’s structural and semantic information before triplet extraction in sentences. We have conducted experiments on a large-scale criminal case dataset. The results show our model significantly improves the judgment prediction task.

AB - Deep Recurrent Neural Network has an excellent performance in sentence semantic analysis. However, due to the curse of the computational dimensionality, the application in the long text is minimal. Therefore, we propose a Triplet Embedding Convolutional Recurrent Neural Network for long text analysis. Firstly, a triplet from each sentence of the long text. Then the most crucial head entity into the CRNN network, composed of CNN and Bi-GRU networks. Both relation and tail entities are input to a CNN network through three splicing layers. Finally, the output results into the global pooling layer to get the final results. Entity fusion and entity replacement are also used to retain the text’s structural and semantic information before triplet extraction in sentences. We have conducted experiments on a large-scale criminal case dataset. The results show our model significantly improves the judgment prediction task.

KW - CRNN

KW - Entity fusion and entity replacement

KW - Long text analysis

KW - Triplet embedding

UR - https://www.scopus.com/pages/publications/85142681247

U2 - 10.1007/978-3-031-20891-1_43

DO - 10.1007/978-3-031-20891-1_43

M3 - Conference Proceeding

AN - SCOPUS:85142681247

SN - 9783031208904

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 607

EP - 615

BT - Web Information Systems Engineering – WISE 2022 - 23rd International Conference, Proceedings

A2 - Chbeir, Richard

A2 - Huang, Helen

A2 - Silvestri, Fabrizio

A2 - Manolopoulos, Yannis

A2 - Zhang, Yanchun

PB - Springer Science and Business Media Deutschland GmbH

T2 - 23rd International Conference on Web Information Systems Engineering, WISE 2021

Y2 - 1 November 2022 through 3 November 2022

ER -

Liu J, Zhu M, Ouyang H, Sun G, Li H. Triplet Embedding Convolutional Recurrent Neural Network for Long Text Semantic Analysis. In Chbeir R, Huang H, Silvestri F, Manolopoulos Y, Zhang Y, Zhang Y, editors, Web Information Systems Engineering – WISE 2022 - 23rd International Conference, Proceedings. Springer Science and Business Media Deutschland GmbH. 2022. p. 607-615. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-031-20891-1_43