RoBERTa-Based Multi-Feature Integrated BiLSTM and CNN Model for Ceramic Review Analysis

To address the limitation that the Robustly Optimized BERT Pretraining Approach (RoBERTa) may not effectively capture local dependencies and salient features within the text, we propose a feature fusion framework based on RoBERTa’s multi-output architecture. By feeding different outputs o...

Full description

Saved in:
Bibliographic Details
Main Authors: LiHua Yang, Jun Wang, WangRen Qiu
Format: Article
Language:English
Published: IEEE 2025-01-01
Series:IEEE Access
Subjects:
Online Access:https://ieeexplore.ieee.org/document/11031473/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849422106261979136
author LiHua Yang
Jun Wang
WangRen Qiu
author_facet LiHua Yang
Jun Wang
WangRen Qiu
author_sort LiHua Yang
collection DOAJ
description To address the limitation that the Robustly Optimized BERT Pretraining Approach (RoBERTa) may not effectively capture local dependencies and salient features within the text, we propose a feature fusion framework based on RoBERTa’s multi-output architecture. By feeding different outputs of RoBERTa into Convolutional Neural Networks (CNN) and Bidirectional Long Short-Term Memory (BiLSTM) networks, the model effectively captures both local static patterns and global contextual dependencies, thereby enhancing its capability to handle complex textual inputs. Additionally, an attention mechanism highlights semantically important features, further improving feature representation and classification performance. To evaluate the effectiveness of the proposed model, we conducted comparative experiments on the Amazon, IMDB, SST-2, and Ceramic datasets. The F1 scores achieved were 94.25%, 92.52%, 91.51%, and 81.32%, respectively, substantially outperforming existing models, particularly in handling domain-specific and nuanced texts. These results demonstrate that the proposed multi-feature fusion model significantly enhances sentiment classification performance.
format Article
id doaj-art-9a624c77fa46442185ccd58b7ea71afc
institution Kabale University
issn 2169-3536
language English
publishDate 2025-01-01
publisher IEEE
record_format Article
series IEEE Access
spelling doaj-art-9a624c77fa46442185ccd58b7ea71afc2025-08-20T03:31:14ZengIEEEIEEE Access2169-35362025-01-011310366810368110.1109/ACCESS.2025.357902211031473RoBERTa-Based Multi-Feature Integrated BiLSTM and CNN Model for Ceramic Review AnalysisLiHua Yang0https://orcid.org/0000-0002-8296-2769Jun Wang1https://orcid.org/0009-0004-9886-9368WangRen Qiu2https://orcid.org/0000-0001-7659-8553School of Information Engineering, Jingdezhen Ceramic University, Jingdezhen, Jiangxi, ChinaSchool of Information Engineering, Jingdezhen Ceramic University, Jingdezhen, Jiangxi, ChinaSchool of Information Engineering, Jingdezhen Ceramic University, Jingdezhen, Jiangxi, ChinaTo address the limitation that the Robustly Optimized BERT Pretraining Approach (RoBERTa) may not effectively capture local dependencies and salient features within the text, we propose a feature fusion framework based on RoBERTa’s multi-output architecture. By feeding different outputs of RoBERTa into Convolutional Neural Networks (CNN) and Bidirectional Long Short-Term Memory (BiLSTM) networks, the model effectively captures both local static patterns and global contextual dependencies, thereby enhancing its capability to handle complex textual inputs. Additionally, an attention mechanism highlights semantically important features, further improving feature representation and classification performance. To evaluate the effectiveness of the proposed model, we conducted comparative experiments on the Amazon, IMDB, SST-2, and Ceramic datasets. The F1 scores achieved were 94.25%, 92.52%, 91.51%, and 81.32%, respectively, substantially outperforming existing models, particularly in handling domain-specific and nuanced texts. These results demonstrate that the proposed multi-feature fusion model significantly enhances sentiment classification performance.https://ieeexplore.ieee.org/document/11031473/RoBERTafeature extractiondeep learningattention mechanismsentiment analysis
spellingShingle LiHua Yang
Jun Wang
WangRen Qiu
RoBERTa-Based Multi-Feature Integrated BiLSTM and CNN Model for Ceramic Review Analysis
IEEE Access
RoBERTa
feature extraction
deep learning
attention mechanism
sentiment analysis
title RoBERTa-Based Multi-Feature Integrated BiLSTM and CNN Model for Ceramic Review Analysis
title_full RoBERTa-Based Multi-Feature Integrated BiLSTM and CNN Model for Ceramic Review Analysis
title_fullStr RoBERTa-Based Multi-Feature Integrated BiLSTM and CNN Model for Ceramic Review Analysis
title_full_unstemmed RoBERTa-Based Multi-Feature Integrated BiLSTM and CNN Model for Ceramic Review Analysis
title_short RoBERTa-Based Multi-Feature Integrated BiLSTM and CNN Model for Ceramic Review Analysis
title_sort roberta based multi feature integrated bilstm and cnn model for ceramic review analysis
topic RoBERTa
feature extraction
deep learning
attention mechanism
sentiment analysis
url https://ieeexplore.ieee.org/document/11031473/
work_keys_str_mv AT lihuayang robertabasedmultifeatureintegratedbilstmandcnnmodelforceramicreviewanalysis
AT junwang robertabasedmultifeatureintegratedbilstmandcnnmodelforceramicreviewanalysis
AT wangrenqiu robertabasedmultifeatureintegratedbilstmandcnnmodelforceramicreviewanalysis