Back to all available

Ensemble learning for Arabic sentiment analysis: A systematic comparison using transformer-based models

computational-linguistics
data-mining-and-machine-learning
natural-language-and-speech
sentiment-analysis
neural-networks

Abstract

Background. Sentiment analysis is impeded by morphological complexity and dialectal variation in Arabic. Transformer-based models are mature but underexplored for Arabic sentiment classification using ensemble methods. We report an evaluation of ensemble learning techniques that combine four transformer architectures, applied to the Hotel Arabic Reviews Dataset (HARD): CAMeLBERT, MARBERT, AraBERT, and XLM-RoBERTa.

Methods. Four ensemble strategies namely Hard Voting / Soft Voting / Weighted Voting / Stacking are systematically contrasted against individual models over 5 random seeds.

Results. Results show stacking is 96.40% ( ± 0.14%) better than the best individual model based on CAMeLBERT with statistical significance (p < 0.05). Manual inspection indicates that ensemble methods are particularly good at mixed-sentiment reviews but that 43.3% of errors are caused by dataset label noise rather than model limitations.

Conclusions. Experiments with general-purpose (LABR, 4,366 entries) and domain-specific (Custom Hotel, 2,000 entries) sentiment lexicons revealed no statistically significant improvements (p > 0.05), suggesting a performance ceiling at 96.42% F1 where transformer models internalize sentiment patterns. Our results show that ensemble diversity based on domain-specific CAMeLBERT/MARBERT and multilingual XLM-ROBERTa models leads to tangible gains in Arabic sentiment analysis with 0.16% accuracy gain with increased robustness.
Ask to review this manuscript

Notes for potential reviewers

  • Volunteering is not a guarantee that you will be asked to review. There are many reasons: reviewers must be qualified, there should be no conflicts of interest, a minimum of two reviewers have already accepted an invitation, etc.
  • This is NOT OPEN peer review. The review is single-blind, and all recommendations are sent privately to the Academic Editor handling the manuscript. All reviews are published and reviewers can choose to sign their reviews.
  • What happens after volunteering? It may be a few days before you receive an invitation to review with further instructions. You will need to accept the invitation to then become an official referee for the manuscript. If you do not receive an invitation it is for one of many possible reasons as noted above.
  • PeerJ Computer Science does not judge submissions based on subjective measures such as novelty, impact or degree of advance. Effectively, reviewers are asked to comment on whether or not the submission is scientifically and technically sound and therefore deserves to join the scientific literature. Our Peer Review criteria can be found on the "Editorial Criteria" page - reviewers are specifically asked to comment on 3 broad areas: "Basic Reporting", "Experimental Design" and "Validity of the Findings".
  • Reviewers are expected to comment in a timely, professional, and constructive manner.
  • Until the article is published, reviewers must regard all information relating to the submission as strictly confidential.
  • When submitting a review, reviewers are given the option to "sign" their review (i.e. to associate their name with their comments). Otherwise, all review comments remain anonymous.
  • All reviews of published articles are published. This includes manuscript files, peer review comments, author rebuttals and revised materials.
  • Each time a decision is made by the Academic Editor, each reviewer will receive a copy of the Decision Letter (which will include the comments of all reviewers).

If you have any questions about submitting your review, please email us at [email protected].