{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,11]],"date-time":"2026-06-11T14:34:30Z","timestamp":1781188470910,"version":"3.54.1"},"reference-count":27,"publisher":"MDPI AG","issue":"2","license":[{"start":{"date-parts":[[2026,2,5]],"date-time":"2026-02-05T00:00:00Z","timestamp":1770249600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["JTAER"],"abstract":"<jats:p>The rapid expansion of e-commerce has significantly influenced consumer purchasing behavior, making user reviews a critical source of product-related information. However, the large volume of low-quality and superficial reviews limits the ability to obtain reliable insights. This study aims to classify Turkish e-commerce reviews as either useful or useless, thereby highlighting high-quality content to support more informed consumer decisions. A dataset of 15,170 Turkish product reviews collected from major e-commerce platforms was analyzed using traditional machine learning approaches, including Support Vector Machines and Logistic Regression, and transformer-based models such as BERT and RoBERTa. In addition, a novel Multi-Transformer Fusion Framework (MTFF) was proposed by integrating BERT and RoBERTa representations through concatenation, weighted-sum, and attention-based fusion strategies. Experimental results demonstrated that the concatenation-based fusion model achieved the highest performance with an F1-score of 91.75%, outperforming all individual models. Among standalone models, Turkish BERT achieved the best performance (F1: 89.37%), while the BERT + Logistic Regression hybrid approach yielded an F1-score of 88.47%. The findings indicate that multi-transformer architectures substantially enhance classification performance, particularly for agglutinative languages such as Turkish. To improve the interpretability of the proposed framework, SHAP (SHapley Additive exPlanations) was employed to analyze feature contributions and provide transparent explanations for model predictions, revealing that the model primarily relies on experience-oriented and semantically meaningful linguistic cues. The proposed approach can support e-commerce platforms by automatically prioritizing high-quality and informative reviews, thereby improving user experience and decision-making processes.<\/jats:p>","DOI":"10.3390\/jtaer21020059","type":"journal-article","created":{"date-parts":[[2026,2,5]],"date-time":"2026-02-05T16:30:47Z","timestamp":1770309047000},"page":"59","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":1,"title":["Explainable Turkish E-Commerce Review Classification Using a Multi-Transformer Fusion Framework and SHAP Analysis"],"prefix":"10.3390","volume":"21","author":[{"ORCID":"https:\/\/orcid.org\/0009-0002-5342-2015","authenticated-orcid":false,"given":"S\u0131la","family":"\u00c7etin","sequence":"first","affiliation":[{"name":"Department of Information Systems Engineering, Faculty of Computer and Informatics Sciences, Sakarya University, Sakarya 54050, Turkey"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-9688-9868","authenticated-orcid":false,"given":"Esin Ay\u015fe","family":"Zaimo\u011flu","sequence":"additional","affiliation":[{"name":"Department of Information Systems Engineering, Faculty of Computer and Informatics Sciences, Sakarya University, Sakarya 54050, Turkey"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1968","published-online":{"date-parts":[[2026,2,5]]},"reference":[{"key":"ref_1","unstructured":"N\u00e9dellec, C., and Rouveirol, C. (1998). Text Categorization with Support Vector Machines: Learning with Many Relevant Features. Machine Learning: ECML-98, Springer."},{"key":"ref_2","doi-asserted-by":"crossref","first-page":"114802","DOI":"10.1016\/j.eswa.2021.114802","article-title":"Detecting Abusive Instagram Comments in Turkish Using Convolutional Neural Network and Machine Learning Methods","volume":"174","year":"2021","journal-title":"Expert Syst. Appl."},{"key":"ref_3","doi-asserted-by":"crossref","first-page":"386","DOI":"10.1016\/j.jocs.2017.11.006","article-title":"Deep Recurrent Neural Network vs. Support Vector Machine for Aspect-Based Sentiment Analysis of Arabic Hotels\u2019 Reviews","volume":"27","author":"Qawasmeh","year":"2018","journal-title":"J. Comput. Sci."},{"key":"ref_4","doi-asserted-by":"crossref","first-page":"1986","DOI":"10.1016\/j.phpro.2012.05.312","article-title":"Text Classifier Based on an Improved SVM Decision Tree","volume":"33","author":"Xu","year":"2012","journal-title":"Phys. Procedia"},{"key":"ref_5","doi-asserted-by":"crossref","unstructured":"\u00d6zmen, C.G., and G\u00fcnd\u00fcz, S. (2025). Comparison of Machine Learning Models for Sentiment Analysis of Big Turkish Web-Based Data. Appl. Sci., 15.","DOI":"10.3390\/app15052297"},{"key":"ref_6","doi-asserted-by":"crossref","unstructured":"Deniz, E., Erbay, H., and Co\u015far, M. (2022). Multi-Label Classification of E-Commerce Customer Reviews via Machine Learning. Axioms, 11.","DOI":"10.3390\/axioms11090436"},{"key":"ref_7","doi-asserted-by":"crossref","first-page":"349","DOI":"10.1016\/j.procs.2024.03.014","article-title":"Comparison of Effectiveness of Logistic Regression, Naive Bayes, and Random Forest Algorithms in Predicting Student Arguments","volume":"234","author":"Wahyuningsih","year":"2024","journal-title":"Procedia Comput. Sci."},{"key":"ref_8","doi-asserted-by":"crossref","first-page":"110761","DOI":"10.1016\/j.knosys.2023.110761","article-title":"Logistic Regression Matching Pursuit Algorithm for Text Classification","volume":"277","author":"Li","year":"2023","journal-title":"Knowl.-Based Syst."},{"key":"ref_9","unstructured":"Devlin, J., Chang, M.W., Lee, K., and Toutanova, K. (2019, January 2\u20137). BERT: Pre-Training of Deep Bidirectional Transformers for Language Understanding. Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT), Minneapolis, MN, USA."},{"key":"ref_10","doi-asserted-by":"crossref","first-page":"231","DOI":"10.1016\/j.aej.2025.04.061","article-title":"BERT-SVM: A Hybrid BERT and SVM Method for Semantic Similarity Matching Evaluation of Paired Short Texts in English Teaching","volume":"126","author":"Xiao","year":"2025","journal-title":"Alex. Eng. J."},{"key":"ref_11","doi-asserted-by":"crossref","first-page":"108979","DOI":"10.1016\/j.compbiomed.2024.108979","article-title":"A Transformer-Based Unified Multimodal Framework for Alzheimer\u2019s Disease Assessment","volume":"180","author":"Yu","year":"2024","journal-title":"Comput. Biol. Med."},{"key":"ref_12","unstructured":"Liu, Y., Ott, M., Goyal, N., Du, J., Joshi, M., Chen, D., Levy, O., Lewis, M., Zettlemoyer, L., and Stoyanov, V. (2019). RoBERTa: A Robustly Optimized BERT Pretraining Approach. arXiv."},{"key":"ref_13","doi-asserted-by":"crossref","unstructured":"Z\u00fcmbero\u011flu, K.B., Dik, S.Z., Karadeniz, B.S., and Sahmoud, S. (2025). Towards Better Sentiment Analysis in the Turkish Language: Dataset Improvements and Model Innovations. Appl. Sci., 15.","DOI":"10.3390\/app15042062"},{"key":"ref_14","first-page":"13","article-title":"Performance Analysis of the Most Downloaded Turkish and English Language Models on the HuggingFace Platform","volume":"61","author":"Gencer","year":"2025","journal-title":"J. Sci. Rep.-A"},{"key":"ref_15","first-page":"110","article-title":"Makine \u00d6\u011frenmesi ile E-Ticaret \u00dcr\u00fcn Yorumlar\u0131n\u0131n Otomatik Analizi","volume":"52","year":"2023","journal-title":"Avrupa Bilim Ve Teknol. Derg."},{"key":"ref_16","doi-asserted-by":"crossref","unstructured":"Demir, E., and Bilgin, M. (2023, January 14\u201317). Sentiment Analysis from Turkish News Texts with BERT-Based Language Models and Machine Learning Algorithms. Proceedings of the 2023 8th International Conference on Computer Science and Engineering (UBMK), Burdur, Turkey.","DOI":"10.1109\/UBMK59864.2023.10286719"},{"key":"ref_17","unstructured":"Mayda, \u0130., and U\u011furlu, Y. (2024, January 23\u201326). Predicting the Usefulness of Turkish Consumer Reviews on E-Commerce Websites. Proceedings of the 2024 32nd Signal Processing and Communications Applications Conference (SIU), Mersin, Turkey."},{"key":"ref_18","first-page":"497","article-title":"BERTurk-Based Sentiment Analysis on E-Commerce Multi-Domain Product Reviews","volume":"25","author":"Teke","year":"2025","journal-title":"Afyon Kocatepe Univ. J. Sci. Eng."},{"key":"ref_19","doi-asserted-by":"crossref","first-page":"51","DOI":"10.12785\/ijcds\/150105","article-title":"A Fusion Architecture of BERT and RoBERTa for Enhanced Performance of Sentiment Analysis of Social Media Platforms","volume":"15","author":"Kumar","year":"2024","journal-title":"Int. J. Comput. Digit. Syst."},{"key":"ref_20","doi-asserted-by":"crossref","first-page":"202","DOI":"10.1016\/j.procs.2024.10.244","article-title":"BERT and RoBERTa Models for Enhanced Detection of Depression in Social Media Text","volume":"245","author":"Kurniadi","year":"2024","journal-title":"Procedia Comput. Sci."},{"key":"ref_21","first-page":"15","article-title":"E-Ticaret Sitelerindeki T\u00fcrk\u00e7e \u00dcr\u00fcn Yorumlar\u0131 \u00dczerine Makine \u00d6\u011frenmesi Algoritmalar\u0131 ile Duygu Analizi","volume":"6","year":"2023","journal-title":"Veri Bilim."},{"key":"ref_22","unstructured":"Vural, M., and Aydo\u011fan, M. (2024, January 14\u201315). Duygu Analizi G\u00f6revi \u00dczerinde Geni\u015f Dil Modellerinin Performans\u0131n\u0131n K\u0131yaslanmas\u0131. Proceedings of the 14th International Academic Research Congress (ICAR), Online."},{"key":"ref_23","unstructured":"Arzu, A., and Aydo\u011fan, M. (2025, December 08). Transformers Tabanl\u0131 Mimariler ile T\u00fcrk\u00e7e Duygu S\u0131n\u0131fland\u0131rmas\u0131 Kar\u015f\u0131la\u015ft\u0131rmal\u0131 Analizi. Available online: https:\/\/dergipark.org.tr\/tr\/download\/article-file\/3364427."},{"key":"ref_24","unstructured":"Oflazer, K. (2025, December 08). T\u00fcrk\u00e7e ve Do\u011fal Dil \u0130\u015fleme. T.B.V. Bilgisayar Bilimleri ve M\u00fchendisli\u011fi Dergisi. Available online: https:\/\/dergipark.org.tr\/tr\/pub\/tbbmd\/issue\/22245\/238795."},{"key":"ref_25","unstructured":"\u00c7abuk, M. (2025, December 08). E-Ticaret \u00dcr\u00fcn Yorumlar\u0131 Veri Seti. Kaggle. Available online: https:\/\/www.kaggle.com\/datasets\/mujdatcabuk\/eticaret-urun-yorumlari."},{"key":"ref_26","unstructured":"Loshchilov, I., and Hutter, F. (2017). Decoupled Weight Decay Regularization. arXiv."},{"key":"ref_27","doi-asserted-by":"crossref","first-page":"223","DOI":"10.1080\/01431160500275762","article-title":"Comparing Accuracy Assessments to Infer Superiority of Image Classification Methods","volume":"27","author":"Jia","year":"2006","journal-title":"Int. J. Remote Sens."}],"container-title":["Journal of Theoretical and Applied Electronic Commerce Research"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/0718-1876\/21\/2\/59\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,2,5]],"date-time":"2026-02-05T16:42:06Z","timestamp":1770309726000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/0718-1876\/21\/2\/59"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,2,5]]},"references-count":27,"journal-issue":{"issue":"2","published-online":{"date-parts":[[2026,2]]}},"alternative-id":["jtaer21020059"],"URL":"https:\/\/doi.org\/10.3390\/jtaer21020059","relation":{},"ISSN":["0718-1876"],"issn-type":[{"value":"0718-1876","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,2,5]]}}}