An Experimental Study on Dynamic Lifelong Learning With GPT for Mitigating Catastrophic Forgetting in Aspect-Based Sentiment Analysis

GPT is widely recognized as one of the most versatile and powerful large language models, excelling across diverse domains. However, its significant computational demands often render it economically unfeasible for individuals and small businesses, underscoring the need for efficient, domain-specifi...

Full description

Saved in:
Bibliographic Details
Main Authors: Huang Huang, Mumtaz Begum Mustafa, Adeleh Asemi
Format: Article
Language:English
Published: IEEE 2025-01-01
Series:IEEE Access
Subjects:
Online Access:https://ieeexplore.ieee.org/document/10909082/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849762371404300288
author Huang Huang
Mumtaz Begum Mustafa
Adeleh Asemi
author_facet Huang Huang
Mumtaz Begum Mustafa
Adeleh Asemi
author_sort Huang Huang
collection DOAJ
description GPT is widely recognized as one of the most versatile and powerful large language models, excelling across diverse domains. However, its significant computational demands often render it economically unfeasible for individuals and small businesses, underscoring the need for efficient, domain-specific alternatives. In Aspect-Based Sentiment Analysis (ABSA), existing models are typically optimized for single domains, facing challenges in performing effectively across multiple domains. A key issue, known as “catastrophic forgetting,” arises when models trained on one domain lose previously learned knowledge upon exposure to new domain data. This leads to two significant problems: limited cross-domain generalization and difficulty in retaining prior knowledge while learning domain-specific information. To address these challenges, we introduce the Dynamic Lifelong Learning Aspect-Based Sentiment Analysis GPT (DllaGPT), a model designed to handle multiple domains while mitigating catastrophic forgetting. Leveraging datasets from four ABSA domains—Laptops, Restaurants, Tweets, and Finance—this study fine-tunes a pretrained GPT model from HuggingFace sequentially across domains. DllaGPT employs a mechanism to retain real data from earlier domains during new domain training, effectively preserving prior knowledge. Experimental results highlight that DllaGPT achieves an average accuracy of 0.85 and a Backward Transfer (BWT) score of -0.09 across the four domains, showcasing its high accuracy and robust lifelong learning capabilities.
format Article
id doaj-art-ff0e08c131c14651bb89fbeef8ca330c
institution DOAJ
issn 2169-3536
language English
publishDate 2025-01-01
publisher IEEE
record_format Article
series IEEE Access
spelling doaj-art-ff0e08c131c14651bb89fbeef8ca330c2025-08-20T03:05:45ZengIEEEIEEE Access2169-35362025-01-0113903169033210.1109/ACCESS.2025.354735910909082An Experimental Study on Dynamic Lifelong Learning With GPT for Mitigating Catastrophic Forgetting in Aspect-Based Sentiment AnalysisHuang Huang0https://orcid.org/0009-0006-8282-9446Mumtaz Begum Mustafa1Adeleh Asemi2https://orcid.org/0000-0002-9193-2430Department of Software Engineering, Faculty of Computer Science and Information Technology, University of Malaya, Kuala Lumpur, MalaysiaDepartment of Software Engineering, Faculty of Computer Science and Information Technology, University of Malaya, Kuala Lumpur, MalaysiaDepartment of Software Engineering, Faculty of Computer Science and Information Technology, University of Malaya, Kuala Lumpur, MalaysiaGPT is widely recognized as one of the most versatile and powerful large language models, excelling across diverse domains. However, its significant computational demands often render it economically unfeasible for individuals and small businesses, underscoring the need for efficient, domain-specific alternatives. In Aspect-Based Sentiment Analysis (ABSA), existing models are typically optimized for single domains, facing challenges in performing effectively across multiple domains. A key issue, known as “catastrophic forgetting,” arises when models trained on one domain lose previously learned knowledge upon exposure to new domain data. This leads to two significant problems: limited cross-domain generalization and difficulty in retaining prior knowledge while learning domain-specific information. To address these challenges, we introduce the Dynamic Lifelong Learning Aspect-Based Sentiment Analysis GPT (DllaGPT), a model designed to handle multiple domains while mitigating catastrophic forgetting. Leveraging datasets from four ABSA domains—Laptops, Restaurants, Tweets, and Finance—this study fine-tunes a pretrained GPT model from HuggingFace sequentially across domains. DllaGPT employs a mechanism to retain real data from earlier domains during new domain training, effectively preserving prior knowledge. Experimental results highlight that DllaGPT achieves an average accuracy of 0.85 and a Backward Transfer (BWT) score of -0.09 across the four domains, showcasing its high accuracy and robust lifelong learning capabilities.https://ieeexplore.ieee.org/document/10909082/Aspect-based sentiment analysisGPTlifelong learninglanguage modeling
spellingShingle Huang Huang
Mumtaz Begum Mustafa
Adeleh Asemi
An Experimental Study on Dynamic Lifelong Learning With GPT for Mitigating Catastrophic Forgetting in Aspect-Based Sentiment Analysis
IEEE Access
Aspect-based sentiment analysis
GPT
lifelong learning
language modeling
title An Experimental Study on Dynamic Lifelong Learning With GPT for Mitigating Catastrophic Forgetting in Aspect-Based Sentiment Analysis
title_full An Experimental Study on Dynamic Lifelong Learning With GPT for Mitigating Catastrophic Forgetting in Aspect-Based Sentiment Analysis
title_fullStr An Experimental Study on Dynamic Lifelong Learning With GPT for Mitigating Catastrophic Forgetting in Aspect-Based Sentiment Analysis
title_full_unstemmed An Experimental Study on Dynamic Lifelong Learning With GPT for Mitigating Catastrophic Forgetting in Aspect-Based Sentiment Analysis
title_short An Experimental Study on Dynamic Lifelong Learning With GPT for Mitigating Catastrophic Forgetting in Aspect-Based Sentiment Analysis
title_sort experimental study on dynamic lifelong learning with gpt for mitigating catastrophic forgetting in aspect based sentiment analysis
topic Aspect-based sentiment analysis
GPT
lifelong learning
language modeling
url https://ieeexplore.ieee.org/document/10909082/
work_keys_str_mv AT huanghuang anexperimentalstudyondynamiclifelonglearningwithgptformitigatingcatastrophicforgettinginaspectbasedsentimentanalysis
AT mumtazbegummustafa anexperimentalstudyondynamiclifelonglearningwithgptformitigatingcatastrophicforgettinginaspectbasedsentimentanalysis
AT adelehasemi anexperimentalstudyondynamiclifelonglearningwithgptformitigatingcatastrophicforgettinginaspectbasedsentimentanalysis
AT huanghuang experimentalstudyondynamiclifelonglearningwithgptformitigatingcatastrophicforgettinginaspectbasedsentimentanalysis
AT mumtazbegummustafa experimentalstudyondynamiclifelonglearningwithgptformitigatingcatastrophicforgettinginaspectbasedsentimentanalysis
AT adelehasemi experimentalstudyondynamiclifelonglearningwithgptformitigatingcatastrophicforgettinginaspectbasedsentimentanalysis