Sentiment classification via supplementary information modeling

Zenan XU, Yetao FU, Xingming CHEN, Yanghui RAO*, Haoran XIE, Fu Lee WANG, Yang PENG

*Corresponding author for this work

Research output: Book Chapters | Papers in Conference ProceedingsConference paper (refereed)

1 Scopus Citations

Abstract

Traditional methods of annotating the sentiment of a document are based on sentiment lexicons, which have been proven quite efficient. However, such methods ignore the effect of supplementary features (e.g., negation and intensity words), while only consider the counts of positive and negative words, the sum of strengths, or the maximum sentiment score over the whole document primarily. In this paper, we propose to use convolutional neural network (CNN) and long short-term memory network (LSTM) to model the role of negation and intensity words, so as to address the limitations of lexicon-based methods. Results show that our model can not only successfully capture the effect of negation and intensity words, but also achieve significant improvements over state-of-the-art deep neural network baselines without supplementary features.
Original languageEnglish
Title of host publicationWeb and Big Data : Second International Joint Conference, APWeb-WAIM 2018, Macau, China, July 23-25, 2018, proceedings, part I
EditorsYi CAI, Yoshiharu ISHIKAWA, Jianliang XU
PublisherSpringer International Publishing AG
Pages54-62
Number of pages9
ISBN (Electronic)9783319968902
ISBN (Print)9783319968896
DOIs
Publication statusPublished - 2018
Externally publishedYes
EventAsia-Pacific Web (APWeb) and Web-Age Information Management (WAIM) Joint International Conference on Web and Big Data - Macau, Macao
Duration: 23 Jul 201825 Jul 2018
http://conferences.cis.umac.mo/apwebwaim2018/

Publication series

NameLecture Notes in Computer Science
PublisherSpringer
Volume10987
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

ConferenceAsia-Pacific Web (APWeb) and Web-Age Information Management (WAIM) Joint International Conference on Web and Big Data
Abbreviated titleAPWeb-WAIM 2018
CountryMacao
CityMacau
Period23/07/1825/07/18
Internet address

Fingerprint

Neural networks
Deep neural networks
Long short-term memory

Bibliographical note

The research has been supported by the National Natural Science Foundation of China (61502545, U1611264, U1711262), Guangdong Science and Technology Program grant (2017A050506025), a grant from the Research Grants Council of the Hong Kong Special Administrative Region, China (UGC/FDS11/E03/16), and the Internal Research Grant (RG 92/2017-2018R) of The Education University of Hong Kong.

Cite this

XU, Z., FU, Y., CHEN, X., RAO, Y., XIE, H., WANG, F. L., & PENG, Y. (2018). Sentiment classification via supplementary information modeling. In Y. CAI, Y. ISHIKAWA, & J. XU (Eds.), Web and Big Data : Second International Joint Conference, APWeb-WAIM 2018, Macau, China, July 23-25, 2018, proceedings, part I (pp. 54-62). (Lecture Notes in Computer Science; Vol. 10987). Springer International Publishing AG. https://doi.org/10.1007/978-3-319-96890-2_5
XU, Zenan ; FU, Yetao ; CHEN, Xingming ; RAO, Yanghui ; XIE, Haoran ; WANG, Fu Lee ; PENG, Yang. / Sentiment classification via supplementary information modeling. Web and Big Data : Second International Joint Conference, APWeb-WAIM 2018, Macau, China, July 23-25, 2018, proceedings, part I. editor / Yi CAI ; Yoshiharu ISHIKAWA ; Jianliang XU. Springer International Publishing AG, 2018. pp. 54-62 (Lecture Notes in Computer Science).
@inproceedings{258289609cf24bb1934542716afe3c46,
title = "Sentiment classification via supplementary information modeling",
abstract = "Traditional methods of annotating the sentiment of a document are based on sentiment lexicons, which have been proven quite efficient. However, such methods ignore the effect of supplementary features (e.g., negation and intensity words), while only consider the counts of positive and negative words, the sum of strengths, or the maximum sentiment score over the whole document primarily. In this paper, we propose to use convolutional neural network (CNN) and long short-term memory network (LSTM) to model the role of negation and intensity words, so as to address the limitations of lexicon-based methods. Results show that our model can not only successfully capture the effect of negation and intensity words, but also achieve significant improvements over state-of-the-art deep neural network baselines without supplementary features.",
author = "Zenan XU and Yetao FU and Xingming CHEN and Yanghui RAO and Haoran XIE and WANG, {Fu Lee} and Yang PENG",
note = "The research has been supported by the National Natural Science Foundation of China (61502545, U1611264, U1711262), Guangdong Science and Technology Program grant (2017A050506025), a grant from the Research Grants Council of the Hong Kong Special Administrative Region, China (UGC/FDS11/E03/16), and the Internal Research Grant (RG 92/2017-2018R) of The Education University of Hong Kong.",
year = "2018",
doi = "10.1007/978-3-319-96890-2_5",
language = "English",
isbn = "9783319968896",
series = "Lecture Notes in Computer Science",
publisher = "Springer International Publishing AG",
pages = "54--62",
editor = "Yi CAI and Yoshiharu ISHIKAWA and Jianliang XU",
booktitle = "Web and Big Data : Second International Joint Conference, APWeb-WAIM 2018, Macau, China, July 23-25, 2018, proceedings, part I",
address = "Switzerland",

}

XU, Z, FU, Y, CHEN, X, RAO, Y, XIE, H, WANG, FL & PENG, Y 2018, Sentiment classification via supplementary information modeling. in Y CAI, Y ISHIKAWA & J XU (eds), Web and Big Data : Second International Joint Conference, APWeb-WAIM 2018, Macau, China, July 23-25, 2018, proceedings, part I. Lecture Notes in Computer Science, vol. 10987, Springer International Publishing AG, pp. 54-62, Asia-Pacific Web (APWeb) and Web-Age Information Management (WAIM) Joint International Conference on Web and Big Data, Macau, Macao, 23/07/18. https://doi.org/10.1007/978-3-319-96890-2_5

Sentiment classification via supplementary information modeling. / XU, Zenan ; FU, Yetao; CHEN, Xingming; RAO, Yanghui; XIE, Haoran; WANG, Fu Lee; PENG, Yang.

Web and Big Data : Second International Joint Conference, APWeb-WAIM 2018, Macau, China, July 23-25, 2018, proceedings, part I. ed. / Yi CAI; Yoshiharu ISHIKAWA; Jianliang XU. Springer International Publishing AG, 2018. p. 54-62 (Lecture Notes in Computer Science; Vol. 10987).

Research output: Book Chapters | Papers in Conference ProceedingsConference paper (refereed)

TY - GEN

T1 - Sentiment classification via supplementary information modeling

AU - XU, Zenan

AU - FU, Yetao

AU - CHEN, Xingming

AU - RAO, Yanghui

AU - XIE, Haoran

AU - WANG, Fu Lee

AU - PENG, Yang

N1 - The research has been supported by the National Natural Science Foundation of China (61502545, U1611264, U1711262), Guangdong Science and Technology Program grant (2017A050506025), a grant from the Research Grants Council of the Hong Kong Special Administrative Region, China (UGC/FDS11/E03/16), and the Internal Research Grant (RG 92/2017-2018R) of The Education University of Hong Kong.

PY - 2018

Y1 - 2018

N2 - Traditional methods of annotating the sentiment of a document are based on sentiment lexicons, which have been proven quite efficient. However, such methods ignore the effect of supplementary features (e.g., negation and intensity words), while only consider the counts of positive and negative words, the sum of strengths, or the maximum sentiment score over the whole document primarily. In this paper, we propose to use convolutional neural network (CNN) and long short-term memory network (LSTM) to model the role of negation and intensity words, so as to address the limitations of lexicon-based methods. Results show that our model can not only successfully capture the effect of negation and intensity words, but also achieve significant improvements over state-of-the-art deep neural network baselines without supplementary features.

AB - Traditional methods of annotating the sentiment of a document are based on sentiment lexicons, which have been proven quite efficient. However, such methods ignore the effect of supplementary features (e.g., negation and intensity words), while only consider the counts of positive and negative words, the sum of strengths, or the maximum sentiment score over the whole document primarily. In this paper, we propose to use convolutional neural network (CNN) and long short-term memory network (LSTM) to model the role of negation and intensity words, so as to address the limitations of lexicon-based methods. Results show that our model can not only successfully capture the effect of negation and intensity words, but also achieve significant improvements over state-of-the-art deep neural network baselines without supplementary features.

UR - http://www.scopus.com/inward/record.url?scp=85050548597&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-96890-2_5

DO - 10.1007/978-3-319-96890-2_5

M3 - Conference paper (refereed)

SN - 9783319968896

T3 - Lecture Notes in Computer Science

SP - 54

EP - 62

BT - Web and Big Data : Second International Joint Conference, APWeb-WAIM 2018, Macau, China, July 23-25, 2018, proceedings, part I

A2 - CAI, Yi

A2 - ISHIKAWA, Yoshiharu

A2 - XU, Jianliang

PB - Springer International Publishing AG

ER -

XU Z, FU Y, CHEN X, RAO Y, XIE H, WANG FL et al. Sentiment classification via supplementary information modeling. In CAI Y, ISHIKAWA Y, XU J, editors, Web and Big Data : Second International Joint Conference, APWeb-WAIM 2018, Macau, China, July 23-25, 2018, proceedings, part I. Springer International Publishing AG. 2018. p. 54-62. (Lecture Notes in Computer Science). https://doi.org/10.1007/978-3-319-96890-2_5