A Green Pipeline for Out-of-Domain Public Sentiment Analysis

Xie, M; Jiang, J; Shen, T; Wang, Y; Gerrard, L; Clarke, A

A Green Pipeline for Out-of-Domain Public Sentiment Analysis

Xie, M Jiang, J

Shen, T Wang, Y Gerrard, L Clarke, A

Permalink

Publisher:: Springer Nature
Publication Type:: Conference Proceeding
Citation:: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 2022, 13087 LNAI, pp. 190-202
Issue Date:: 2022-01-01

Closed Access

	Filename	Description	Size
	A Green Pipeline for Out-of-Domain Public Sentiment Analysis.pdf	Published version	414.35 kB	Adobe PDF	View/Open

Copyright Clearance Process

Recently Added
In Progress
Closed Access

This item is closed access and not available.

Full metadata record

Field	Value	Language
dc.contributor.author	Xie, M
dc.contributor.author	Jiang, J https://orcid.org/0000-0001-5301-7779
dc.contributor.author	Shen, T
dc.contributor.author	Wang, Y
dc.contributor.author	Gerrard, L
dc.contributor.author	Clarke, A
dc.date.accessioned	2023-03-21T02:00:29Z
dc.date.available	2023-03-21T02:00:29Z
dc.date.issued	2022-01-01
dc.identifier.citation	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 2022, 13087 LNAI, pp. 190-202
dc.identifier.isbn	9783030954048
dc.identifier.issn	0302-9743
dc.identifier.issn	1611-3349
dc.identifier.uri	http://hdl.handle.net/10453/167891
dc.description.abstract	In the changing social and economic environment, organisations are keen to act promptly and appropriately to changes. Sentiment analysis can be applied to social media data to capture timely information of new events and the corresponding public opinions. However, currently both the social topics and trending words are changing just as rapidly as the target topics and domains that organisations are interested in investigating. Therefore, there is a need for a well-trained sentiment analysis model able to handle out-of-domain input. Current solutions mainly focus on using domain adaptation techniques, but these solutions require domain-specific data and inevitably introduce extra overheads. To tackle this challenge, we propose a green Artificial Intelligence (AI) solution for a sentiment analysis pipeline (GreenSAP) to gain a better understanding of the changing public opinions on social media. Specifically, we propose to leverage the expressively powerful capability of the pre-trained Transformer encoder, and make use of several publicly-available sentiment analysis datasets from various domains and scenarios to develop a pipeline model. A sarcasm detection model is also included to eliminate false positive predictions. In experiments, this model significantly outperforms its competitors on three public benchmark datasets and on two of our labelled out-of-domain datasets for real-world applications.
dc.language	en
dc.publisher	Springer Nature
dc.relation.ispartof	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
dc.relation.ispartofseries	Lecture Notes in Computer Science
dc.relation.isbasedon	10.1007/978-3-030-95405-5_14
dc.rights	info:eu-repo/semantics/closedAccess
dc.subject.classification	Artificial Intelligence & Image Processing
dc.title	A Green Pipeline for Out-of-Domain Public Sentiment Analysis
dc.type	Conference Proceeding
utslib.citation.volume	13087 LNAI
pubs.organisational-group	/University of Technology Sydney
pubs.organisational-group	/University of Technology Sydney/Faculty of Engineering and Information Technology
pubs.organisational-group	/University of Technology Sydney/Strength - AAII - Australian Artificial Intelligence Institute
utslib.copyright.status	closed_access	*
dc.date.updated	2023-03-21T02:00:28Z
pubs.publication-status	Published
pubs.volume	13087 LNAI

Abstract:

In the changing social and economic environment, organisations are keen to act promptly and appropriately to changes. Sentiment analysis can be applied to social media data to capture timely information of new events and the corresponding public opinions. However, currently both the social topics and trending words are changing just as rapidly as the target topics and domains that organisations are interested in investigating. Therefore, there is a need for a well-trained sentiment analysis model able to handle out-of-domain input. Current solutions mainly focus on using domain adaptation techniques, but these solutions require domain-specific data and inevitably introduce extra overheads. To tackle this challenge, we propose a green Artificial Intelligence (AI) solution for a sentiment analysis pipeline (GreenSAP) to gain a better understanding of the changing public opinions on social media. Specifically, we propose to leverage the expressively powerful capability of the pre-trained Transformer encoder, and make use of several publicly-available sentiment analysis datasets from various domains and scenarios to develop a pipeline model. A sarcasm detection model is also included to eliminate false positive predictions. In experiments, this model significantly outperforms its competitors on three public benchmark datasets and on two of our labelled out-of-domain datasets for real-world applications.

Please use this identifier to cite or link to this item:

http://hdl.handle.net/10453/167891