Enhancing Arabic Sentiment Analysis in E-Commerce Reviews on Social Media Through a Stacked Ensemble Deep Learning Approach
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Sentiment analysis (SA) employs natural language processing techniques to extract opinions from textual data.Applying SA to the Arabic language presents numerous challenges, including ambiguity, the presence of multiple dialects, a need for additional resources, and morphological variation.The domain of Arabic SA has witnessed significant advancements with the application of deep learning (DL) approaches, such as convolutional neural networks (CNNs).The performance of single DL models has been further improved by hybrid models combining CNNs with bidirectional long short-term memory (Bi-LSTM) or bidirectional gated recurrent units (Bi-GRU).It is anticipated that the accuracy of these DL models can be enhanced through stacked deep learning ensembles.In this study, a stacked ensemble approach is proposed that accurately predicts Arabic sentiment by leveraging the predictive capabilities of CNN, Bi-GRU, Bi-LSTM, and hybrid DL models (CNN-Bi-GRU and CNN-Bi-LSTM).The proposed model's efficacy is evaluated using four extensive datasets: the HARD dataset, the BRAD dataset, the ARD dataset, and a real dataset composed of 71,583 Arabic reviews.Experimental results demonstrate the suitability of the proposed model for analyzing sentiments in Arabic texts.The method's first step involves feature extraction using the AraBERT model.Subsequently, five DL models are developed and trained, including CNN, Bi-GRU, Bi-LSTM, a hybrid CNN-Bi-GRU model, and a hybrid CNN-LSTM model.Finally, the outputs of the base classifiers are concatenated using the multilayer perceptron algorithm.Our approach achieves an improved accuracy of 0.9256 compared to basic and hybrid deep learning methods.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.000 | 0.002 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it