![[2602.12778] Aspect-Based Sentiment Analysis for Future Tourism Experiences: A BERT-MoE Framework for Persian User Reviews](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.12778] Aspect-Based Sentiment Analysis for Future Tourism Experiences: A BERT-MoE Framework for Persian User Reviews
Summary
This article presents a novel BERT-MoE framework for aspect-based sentiment analysis (ABSA) tailored for Persian user reviews in the tourism sector, achieving significant efficiency and accuracy improvements.
Why It Matters
The study addresses the challenges of low-resource languages in sentiment analysis, particularly in tourism, which is vital for enhancing user experience and supporting sustainable AI practices. By releasing a new annotated dataset, it also paves the way for further research in multilingual NLP.
Key Takeaways
- Introduces a hybrid BERT-based model for ABSA in Persian tourism reviews.
- Achieves a weighted F1-score of 90.6%, outperforming existing models.
- Demonstrates a 39% reduction in GPU power consumption, promoting sustainable AI.
- Focuses on critical aspects like cleanliness and amenities in tourism.
- Releases a new annotated dataset to aid future multilingual NLP research.
Computer Science > Computation and Language arXiv:2602.12778 (cs) [Submitted on 13 Feb 2026] Title:Aspect-Based Sentiment Analysis for Future Tourism Experiences: A BERT-MoE Framework for Persian User Reviews Authors:Hamidreza Kazemi Taskooh, Taha Zare Harofte View a PDF of the paper titled Aspect-Based Sentiment Analysis for Future Tourism Experiences: A BERT-MoE Framework for Persian User Reviews, by Hamidreza Kazemi Taskooh and Taha Zare Harofte View PDF HTML (experimental) Abstract:This study advances aspect-based sentiment analysis (ABSA) for Persian-language user reviews in the tourism domain, addressing challenges of low-resource languages. We propose a hybrid BERT-based model with Top-K routing and auxiliary losses to mitigate routing collapse and improve efficiency. The pipeline includes: (1) overall sentiment classification using BERT on 9,558 labeled reviews, (2) multi-label aspect extraction for six tourism-related aspects (host, price, location, amenities, cleanliness, connectivity), and (3) integrated ABSA with dynamic routing. The dataset consists of 58,473 preprocessed reviews from the Iranian accommodation platform Jabama, manually annotated for aspects and sentiments. The proposed model achieves a weighted F1-score of 90.6% for ABSA, outperforming baseline BERT (89.25%) and a standard hybrid approach (85.7%). Key efficiency gains include a 39% reduction in GPU power consumption compared to dense BERT, supporting sustainable AI deployment in alignment with...
