Enhancing Arabic Aspect-Based Sentiment Analysis Using End-to-End Model
The majority of research on the Aspect-Based Sentiment Analysis (ABSA) tends to split this task into two subtasks: one for extracting aspects, Aspect Term Extraction (ATE), and another for identifying sentiments toward particular aspects, Aspect Sentiment Classification (ASC). Although these subtask...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2023-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10356063/ |
| _version_ | 1797376385179713536 |
|---|---|
| author | Ghada M. Shafiq Taher Hamza Mohammed F. Alrahmawy Reem El-Deeb |
| author_facet | Ghada M. Shafiq Taher Hamza Mohammed F. Alrahmawy Reem El-Deeb |
| author_sort | Ghada M. Shafiq |
| collection | DOAJ |
| description | The majority of research on the Aspect-Based Sentiment Analysis (ABSA) tends to split this task into two subtasks: one for extracting aspects, Aspect Term Extraction (ATE), and another for identifying sentiments toward particular aspects, Aspect Sentiment Classification (ASC). Although these subtasks are closely related, they are performed independently; while performing the Aspect Sentiment Classification task, it is assumed that the aspect terms are pre-identified, which ignores the practical interaction required to properly perform the ABSA. This study addresses these limitations using a unified End-to-End (E2E) approach, which combines the two subtasks into a single sequence labeling task using a unified tagging schema. The proposed model was evaluated by fine-tuning the Arabic version of the Bidirectional Encoder Representations from Transformers (AraBERT) model with a Conditional Random Fields (CRF) classifier for enhanced target-polarity identification. The experimental results demonstrated the efficiency of the proposed fine-tuned AraBERT-CRF model, which achieved an overall F1 score of 95.11% on the SemEval-2016 Arabic Hotel Reviews dataset. The model’s predictions are then subjected to additional processing, and the results indicate the superiority of the proposed model, achieving an F1 score of 97.78% for the ATE task and an accuracy of 98.34% for the ASC task, outperforming previous studies. |
| first_indexed | 2024-03-08T19:37:47Z |
| format | Article |
| id | doaj.art-e86e810c9be8426390d485de597e6488 |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-03-08T19:37:47Z |
| publishDate | 2023-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-e86e810c9be8426390d485de597e64882023-12-26T00:07:56ZengIEEEIEEE Access2169-35362023-01-011114206214207610.1109/ACCESS.2023.334275510356063Enhancing Arabic Aspect-Based Sentiment Analysis Using End-to-End ModelGhada M. Shafiq0https://orcid.org/0000-0001-8540-3623Taher Hamza1https://orcid.org/0000-0001-5735-8562Mohammed F. Alrahmawy2https://orcid.org/0000-0001-8978-8268Reem El-Deeb3https://orcid.org/0000-0002-4984-6416Department of Computer Science, Faculty of Computers and Information, Mansoura University, Mansoura, EgyptDepartment of Computer Science, Faculty of Computers and Information, Mansoura University, Mansoura, EgyptDepartment of Computer Science, Faculty of Computers and Information, Mansoura University, Mansoura, EgyptDepartment of Computer Science, Faculty of Computers and Information, Mansoura University, Mansoura, EgyptThe majority of research on the Aspect-Based Sentiment Analysis (ABSA) tends to split this task into two subtasks: one for extracting aspects, Aspect Term Extraction (ATE), and another for identifying sentiments toward particular aspects, Aspect Sentiment Classification (ASC). Although these subtasks are closely related, they are performed independently; while performing the Aspect Sentiment Classification task, it is assumed that the aspect terms are pre-identified, which ignores the practical interaction required to properly perform the ABSA. This study addresses these limitations using a unified End-to-End (E2E) approach, which combines the two subtasks into a single sequence labeling task using a unified tagging schema. The proposed model was evaluated by fine-tuning the Arabic version of the Bidirectional Encoder Representations from Transformers (AraBERT) model with a Conditional Random Fields (CRF) classifier for enhanced target-polarity identification. The experimental results demonstrated the efficiency of the proposed fine-tuned AraBERT-CRF model, which achieved an overall F1 score of 95.11% on the SemEval-2016 Arabic Hotel Reviews dataset. The model’s predictions are then subjected to additional processing, and the results indicate the superiority of the proposed model, achieving an F1 score of 97.78% for the ATE task and an accuracy of 98.34% for the ASC task, outperforming previous studies.https://ieeexplore.ieee.org/document/10356063/Sentiment analysisaspect-basedAraBERTCRFtransfer learning |
| spellingShingle | Ghada M. Shafiq Taher Hamza Mohammed F. Alrahmawy Reem El-Deeb Enhancing Arabic Aspect-Based Sentiment Analysis Using End-to-End Model IEEE Access Sentiment analysis aspect-based AraBERT CRF transfer learning |
| title | Enhancing Arabic Aspect-Based Sentiment Analysis Using End-to-End Model |
| title_full | Enhancing Arabic Aspect-Based Sentiment Analysis Using End-to-End Model |
| title_fullStr | Enhancing Arabic Aspect-Based Sentiment Analysis Using End-to-End Model |
| title_full_unstemmed | Enhancing Arabic Aspect-Based Sentiment Analysis Using End-to-End Model |
| title_short | Enhancing Arabic Aspect-Based Sentiment Analysis Using End-to-End Model |
| title_sort | enhancing arabic aspect based sentiment analysis using end to end model |
| topic | Sentiment analysis aspect-based AraBERT CRF transfer learning |
| url | https://ieeexplore.ieee.org/document/10356063/ |
| work_keys_str_mv | AT ghadamshafiq enhancingarabicaspectbasedsentimentanalysisusingendtoendmodel AT taherhamza enhancingarabicaspectbasedsentimentanalysisusingendtoendmodel AT mohammedfalrahmawy enhancingarabicaspectbasedsentimentanalysisusingendtoendmodel AT reemeldeeb enhancingarabicaspectbasedsentimentanalysisusingendtoendmodel |