Transactions on Data Analysis in Social Science

Transactions on Data Analysis in Social Science

Artificial Intelligence for Sentiment Analysis of Persian Online Reviews: A Focused Review of Models, Datasets, and Evaluation

Document Type : Original Article

Author
H. Ghayoumi Zadeh
Associate Professor, Department of Electrical Engineering, Vali-e-Asr University of Rafsanjan, Rafsanjan, Iran.
10.22034/tdas.2025.256843
Abstract
Persian online reviews provide evidence about consumer experiences, but the usefulness of automated sentiment analysis depends on what the labels measure and where the model is tested. This focused narrative review examines benchmark construction and evaluation for Persian product and service reviews. It covers lexicon and statistical methods, convolutional and recurrent neural networks, Persian and multilingual encoders, and instruction-tuned large language models. The reviewed studies use different observation units, label definitions, and test protocols. A binary food-delivery benchmark, a three-way product recommendation dataset, and a five-level sentence polarity corpus therefore support different claims even when all report an F1 score. Comparisons within individual studies show that the size of a model advantage varies by benchmark. Evaluations across several datasets also reveal source-specific performance differences, although these differences do not identify the effect of transferring a classifier trained on one platform to another. The review proposes a reporting framework covering dataset provenance, label alignment, leakage controls, per-class metrics, and external validation. An evaluation design for transfer between platforms and a taxonomy of review errors illustrate how to apply the framework in consumer research.
Keywords
Artificial Intelligence
Persian Sentiment Analysis
Online Reviews
Benchmark Evaluation
Domain Shift
[1]      Hosseini, P., Ramaki, A. A., Maleki, H., Anvari, M., & Mirroshandel, S. A. (2018). SentiPers: A sentiment analysis corpus for Persian. arXiv:1801.07737. https://arxiv.org/abs/1801.07737
[2]      PourMostafa Roshan Sharami, J., Abbasi Sarabestani, P., & Mirroshandel, S. A. (2020). DeepSentiPers: Novel deep learning models trained over proposed augmented Persian sentiment corpus. arXiv:2004.05328. https://arxiv.org/abs/2004.05328
[3]      Hooshvare Lab. (n.d.). Persian sentiment analysis [Dataset documentation]. https://hooshvare.github.io/docs/datasets/sa
[4]      Farahani, M., Gharachorloo, M., Farahani, M., & Manthouri, M. (2021). ParsBERT: Transformer-based model for Persian language understanding. Neural Processing Letters, 53, 3831–3847. https://doi.org/10.1007/s11063-021-10528-4
[5]      Ashrafi Asli, S. A., Sabeti, B., Majdabadi, Z., Golazizian, P., Fahmi, R., & Momenzadeh, O. (2020). Optimizing annotation effort using active learning strategies: A sentiment analysis case study in Persian. Proceedings of LREC 2020, 2855–2861. https://aclanthology.org/2020.lrec-1.348/
[6]      Masumi, M., Majd, S. S., Shamsfard, M., & Beigy, H. (2025). FaBERT: Pre-training BERT on Persian blogs. Proceedings of W-NUT 2025, 85–96. https://aclanthology.org/2025.wnut-1.10/
[7]      Basiri, M. E., & Kabiri, A. (2020). HOMPer: A new hybrid system for opinion mining in the Persian language. Journal of Information Science, 46(1), 101–117. https://doi.org/10.1177/0165551519827886
[8]      Shumaly, S., Yazdinejad, M., & Guo, Y. (2021). Persian sentiment analysis of an online store independent of pre-processing using convolutional neural network with fastText embeddings. PeerJ Computer Science, 7, e422. https://doi.org/10.7717/peerj-cs.422
[9]      SadraeiJavaheri, M., Moghaddaszadeh, A., Molazadeh, M., Naeiji, F., Aghababaloo, F., Rafiee, H., Amirmahani, Z., Abedini, T., Sheikhi, F. Z., & Salehoof, A. (2024). TookaBERT: A step forward for Persian NLU. arXiv:2407.16382. https://arxiv.org/abs/2407.16382
[10]   Arnob, N. M. K., & Mahi, A. B. S. (2026). One language, three of its voices: Evaluating multilingual LLMs across Persian, Dari, and Tajiki on translation and understanding tasks. Proceedings of SilkRoadNLP 2026, 98–104. https://doi.org/10.18653/v1/2026.silkroadnlp-1.10
[11]   Shangipour Ataei, T., Darvishi, K., Javdan, S., Minaei-Bidgoli, B., & Eetemadi, S. (2022). Pars-ABSA: A manually annotated aspect-based sentiment analysis benchmark on Farsi product reviews. Proceedings of LREC 2022, 7056–7060. https://aclanthology.org/2022.lrec-1.763/
[12]   Amiri, F., Scerri, S., & Khodashahi, M. (2015). Lexicon-based sentiment analysis for Persian text. Proceedings of RANLP 2015, 9–16. https://aclanthology.org/R15-1002/
[13]   Blitzer, J., Dredze, M., & Pereira, F. (2007). Biographies, Bollywood, boom-boxes and blenders: Domain adaptation for sentiment classification. Proceedings of ACL 2007, 440–447. https://aclanthology.org/P07-1056/
[14]   Lee, K., Ippolito, D., Nystrom, A., Zhang, C., Eck, D., Callison-Burch, C., & Carlini, N. (2022). Deduplicating training data makes language models better. Proceedings of ACL 2022, 8424–8445. https://aclanthology.org/2022.acl-long.577/
[15]   Dror, R., Baumer, G., Shlomov, S., & Reichart, R. (2018). The hitchhiker’s guide to testing statistical significance in natural language processing. Proceedings of ACL 2018, 1383–1392. https://aclanthology.org/P18-1128/
[16]   Guo, C., Pleiss, G., Sun, Y., & Weinberger, K. Q. (2017). On calibration of modern neural networks. Proceedings of ICML 2017, 1321–1330. https://proceedings.mlr.press/v70/guo17a.html
[17]   Gebru, T., Morgenstern, J., Vecchione, B., Vaughan, J. W., Wallach, H., Daumé III, H., & Crawford, K. (2021). Datasheets for datasets. Communications of the ACM, 64(12), 86–92. https://doi.org/10.1145/3458723
[18]   Mitchell, M., Wu, S., Zaldivar, A., Barnes, P., Vasserman, L., Hutchinson, B., Spitzer, E., Raji, I. D., & Gebru, T. (2019). Model cards for model reporting. Proceedings of FAT* 2019, 220–229. https://doi.org/10.1145/3287560.3287596
Transactions on Data Analysis in Social Science
Volume 7, Issue 4
Autumn 2025
Pages 309-318

  • Receive Date 20 August 2025
  • Revise Date 22 October 2025
  • Accept Date 08 December 2025