Social emotion classification based on noise-aware training

Xin LI, Yanghui RAO*, Haoran XIE, Xuebo LIU, Tak-Lam WONG, Fu Lee WANG

*Corresponding author for this work

Research output: Journal PublicationsJournal Article (refereed)

2 Citations (Scopus)

Abstract

Social emotion classification draws many natural language processing researchers’ attention in recent years, since analyzing user-generated emotional documents on the Web is quite useful in recommending products, gathering public opinions, and predicting election results. However, the documents that evoke prominent social emotions are usually mixed with noisy instances, and it is also challenging to capture the textual meaning of short messages. In this work, we focus on reducing the impact of noisy instances and learning a better representation of sentences. For the former, we introduce an “emotional concentration” indicator, which is derived from emotional ratings to weight documents. For the latter, we propose a new architecture named PCNN, which utilizes two cascading convolutional layers to model the word-phrase relation and the phrase–sentence relation. This model regards continuous tokens as phrases based on an assumption that neighboring words are very likely to have internal relations, and semantic feature vectors are generated based on the phrase representation. We also present a Bayesian-based model named WMCM to learn document-level semantic features. Both PCNN and WMCM classify social emotions by capturing semantic regularities in language. Experiments on two real-world datasets indicate that the quality of learned semantic vectors and the performance of social emotion classification can be improved by our models.

Original languageEnglish
Article number101605
JournalData and Knowledge Engineering
Volume123
DOIs
Publication statusPublished - Sep 2019
Externally publishedYes

Fingerprint

Emotion
Regularity
Language
Rating
Elections
Public opinion
Experiment
World Wide Web
Natural language processing

Bibliographical note

This paper is an extended version of our previous conference paper (Li et al., 2016).

Keywords

  • Convolutional neural network
  • Emotional concentration
  • Social emotion classification
  • Topic modeling

Cite this

LI, Xin ; RAO, Yanghui ; XIE, Haoran ; LIU, Xuebo ; WONG, Tak-Lam ; WANG, Fu Lee. / Social emotion classification based on noise-aware training. In: Data and Knowledge Engineering. 2019 ; Vol. 123.
@article{79cd5dcfcb7c463b9452ac8bb6479a48,
title = "Social emotion classification based on noise-aware training",
abstract = "Social emotion classification draws many natural language processing researchers’ attention in recent years, since analyzing user-generated emotional documents on the Web is quite useful in recommending products, gathering public opinions, and predicting election results. However, the documents that evoke prominent social emotions are usually mixed with noisy instances, and it is also challenging to capture the textual meaning of short messages. In this work, we focus on reducing the impact of noisy instances and learning a better representation of sentences. For the former, we introduce an “emotional concentration” indicator, which is derived from emotional ratings to weight documents. For the latter, we propose a new architecture named PCNN, which utilizes two cascading convolutional layers to model the word-phrase relation and the phrase–sentence relation. This model regards continuous tokens as phrases based on an assumption that neighboring words are very likely to have internal relations, and semantic feature vectors are generated based on the phrase representation. We also present a Bayesian-based model named WMCM to learn document-level semantic features. Both PCNN and WMCM classify social emotions by capturing semantic regularities in language. Experiments on two real-world datasets indicate that the quality of learned semantic vectors and the performance of social emotion classification can be improved by our models.",
keywords = "Convolutional neural network, Emotional concentration, Social emotion classification, Topic modeling",
author = "Xin LI and Yanghui RAO and Haoran XIE and Xuebo LIU and Tak-Lam WONG and WANG, {Fu Lee}",
note = "This paper is an extended version of our previous conference paper (Li et al., 2016).",
year = "2019",
month = "9",
doi = "10.1016/j.datak.2017.07.008",
language = "English",
volume = "123",
journal = "Data and Knowledge Engineering",
issn = "0169-023X",
publisher = "Elsevier",

}

Social emotion classification based on noise-aware training. / LI, Xin; RAO, Yanghui; XIE, Haoran; LIU, Xuebo; WONG, Tak-Lam; WANG, Fu Lee.

In: Data and Knowledge Engineering, Vol. 123, 101605, 09.2019.

Research output: Journal PublicationsJournal Article (refereed)

TY - JOUR

T1 - Social emotion classification based on noise-aware training

AU - LI, Xin

AU - RAO, Yanghui

AU - XIE, Haoran

AU - LIU, Xuebo

AU - WONG, Tak-Lam

AU - WANG, Fu Lee

N1 - This paper is an extended version of our previous conference paper (Li et al., 2016).

PY - 2019/9

Y1 - 2019/9

N2 - Social emotion classification draws many natural language processing researchers’ attention in recent years, since analyzing user-generated emotional documents on the Web is quite useful in recommending products, gathering public opinions, and predicting election results. However, the documents that evoke prominent social emotions are usually mixed with noisy instances, and it is also challenging to capture the textual meaning of short messages. In this work, we focus on reducing the impact of noisy instances and learning a better representation of sentences. For the former, we introduce an “emotional concentration” indicator, which is derived from emotional ratings to weight documents. For the latter, we propose a new architecture named PCNN, which utilizes two cascading convolutional layers to model the word-phrase relation and the phrase–sentence relation. This model regards continuous tokens as phrases based on an assumption that neighboring words are very likely to have internal relations, and semantic feature vectors are generated based on the phrase representation. We also present a Bayesian-based model named WMCM to learn document-level semantic features. Both PCNN and WMCM classify social emotions by capturing semantic regularities in language. Experiments on two real-world datasets indicate that the quality of learned semantic vectors and the performance of social emotion classification can be improved by our models.

AB - Social emotion classification draws many natural language processing researchers’ attention in recent years, since analyzing user-generated emotional documents on the Web is quite useful in recommending products, gathering public opinions, and predicting election results. However, the documents that evoke prominent social emotions are usually mixed with noisy instances, and it is also challenging to capture the textual meaning of short messages. In this work, we focus on reducing the impact of noisy instances and learning a better representation of sentences. For the former, we introduce an “emotional concentration” indicator, which is derived from emotional ratings to weight documents. For the latter, we propose a new architecture named PCNN, which utilizes two cascading convolutional layers to model the word-phrase relation and the phrase–sentence relation. This model regards continuous tokens as phrases based on an assumption that neighboring words are very likely to have internal relations, and semantic feature vectors are generated based on the phrase representation. We also present a Bayesian-based model named WMCM to learn document-level semantic features. Both PCNN and WMCM classify social emotions by capturing semantic regularities in language. Experiments on two real-world datasets indicate that the quality of learned semantic vectors and the performance of social emotion classification can be improved by our models.

KW - Convolutional neural network

KW - Emotional concentration

KW - Social emotion classification

KW - Topic modeling

UR - http://www.scopus.com/inward/record.url?scp=85026356456&partnerID=8YFLogxK

U2 - 10.1016/j.datak.2017.07.008

DO - 10.1016/j.datak.2017.07.008

M3 - Journal Article (refereed)

AN - SCOPUS:85026356456

VL - 123

JO - Data and Knowledge Engineering

JF - Data and Knowledge Engineering

SN - 0169-023X

M1 - 101605

ER -