Title: Multilingual Sentiment Analysis Using a Single LSTM Architecture: A Comparative Study on Turkish and English Datasets

Abstract: Sentiment analysis has become an essential task in Natural Language Processing (NLP), particularly with the growing availability of multilingual textual data. While most studies in the literature focus on monolingual models trained separately for each language, the present study proposes a unified deep learning–based framework that performs sentiment analysis on both Turkish and English texts using a single LSTM architecture. Two datasets were employed: the publicly available IMDb movie review dataset for English and a manually labeled dataset consisting of approximately 2,000 Turkish sentences. Texts in both languages were preprocessed, tokenized, and transformed into fixed-length vector representations through embedding and LSTM layers, and binary sentiment classification was performed using a sigmoid activation function. Experimental results demonstrate that the model achieves high accuracy on the English dataset, benefiting from its large and well-balanced structure, while comparatively lower generalization performance is observed for the Turkish dataset due to its smaller size and limited domain coverage. The findings highlight the importance of dataset scale and linguistic characteristics in multilingual sentiment analysis and show that LSTM-based architecture provides an effective baseline for bilingual sentiment classification. Future work will focus on expanding Turkish data resources and integrating transformer-based multilingual models to improve performance across morphologically rich languages.

Keywords: Sentiment Analysis, Natural Language Processing (NLP), Long Short-Term Memory (LSTM), Deep Learning, Cross-Language Modeling

• Abadi, M., Barham, P., Chen, J., Chen, Z., Davis, A., Dean, J., Devin, M., Ghemawat, S., Irving, G., Isard, M., et al. Tensorflow: A system for large-scale machine learning. arXiv preprint arXiv:1605.08695 (2016), https://arxiv.org/abs/1605.08695
• Ant, K., Soğukpınar, U., Amasyalı, M.F. Comparison of templates with word2vec in finding semantic relations between words. Journal of Intelligent Systems with Applications 1(1), 13–17 (2018), https://joiswa.com/issue.php?issue=1
• Atayolu, Y., Kutlu, Y. Effect of text preprocessing methods on the performance of social media posts classification. Journal of Intelligent Systems with Applications 7(1), 1–6 (2023), https://joiswa.com/indir.php?id=304
• Berus, Y., Bente¸sen Yakut, Y. Derin Öğrenme (1d-cnn, rnn, lstm, bilstm) ile enerji tüketim tahmini: Diyarbakır avm Örne˘gi. DÜMF Mühendislik Dergisi (2024). https://doi.org/10.24012/dumf.1415055
• Bird, S. Nltk: The natural language toolkit. In: Proceedings of the COLING/ACL 2006 Interactive Presentation Sessions. pp. 69–72. As sociation for Computational Linguistics (2006)
• Can, V.B., Şatır, E. Mitigating gender bias in english–spanish machine translation: A parameter-efficient fine-tuning approach. Journal of Intelligent Systems with Applications 8(1), 20–24 (2025)
• Demir, Ö., Chawai, A.I.B., Doğan, B. Türkçe metinlerde sözlük tabanli yaklaşimla duygu analizi ve görselleştirme. Int. Periodical of Recent Technologies in Applied Engineering 1(2), 14–22 (2020). https://doi.org/10.35333/porta.2019.98
• Demirbilek, M., Demirbilek, S. Google yorumları Üzerinden makine Öğrenme yöntemleri ve amazon comprehend ile duygu analizi. Üniversite Araştırmaları Dergisi 6(4), 452–461 (2023). https://doi.org/10.32329/uad.1383794
• Dönmez, H.G., Becerikli, Y. Türkçe twitter verileri Üzerinde duygu analizi: Lstm, cnn-lstm, bert algoritmalarının karşılaştırılması. International Journal of Advanced Natural Sciences and Engineering Re searches 7(6), 132–138 (2023). https://doi.org/10.59287/ijanser.1146
• Görentaş, M.B., Uçkan, T., Arlı, N.B. Uyuşmazlık mahkemesi karar larının makine Öğrenmesi yöntemleri ile sınıflandırılması. Yüzüncü Yıl Üniversitesi Fen Bilimleri Enstitüsü Dergisi 28(3), 947–961 (2023). https://doi.org/10.53433/yyufbed.1292275
• Hochreiter, S., Schmidhuber, J. Long short-term mem ory. Computation 9(8), https://doi.org/10.1162/neco.1997.9.8.1735
• İnan, E., Mostafapour, V., Tekbacak, F. Domain-specific evaluation dataset generator for multilingual text analysis. Journal of Intelligent Systems with Applications 2(2), 140–147 (2019), https://joiswa.com/abstract.php?id=84
• Karaca, A., Aydın, Ö. Transformatör mimarisi tabanlı derin Öğrenme yöntemi ile türkçe haber metinlerine başlık Üretme. Gazi Üniversitesi Mühendislik-Mimarlık Fakültesi Dergisi 39(1), 485–496 (2023). https://doi.org/10.17341/gazimmfd.963240
• Maas, A.L., Daly, R.E., Pham, P.T., Huang, D., Ng, A.Y., Potts, C. Learning word vectors for sentiment analysis. In: Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies. pp. 142–150. Association for Computational Linguistics (2011)
• Onan, A. Evrişimli sinir ağı mimarilerine dayalı türkçe duygu analizi. European Journal of Science and Technology pp. 374–380 (2020). https://doi.org/10.31590/ejosat.780609
• Onan, A. Türkçe Metin Madenciliği için Dikkat Mekanizması Tabanlı Derin Öğrenme Mimarilerinin Değerlendirilmesi. European Journal of Science and Technology (2022). https://doi.org/10.31590/ejosat.1082379
• Onan, A. Türkçe Metin Madenciliği için Çalışan Bellek Bağlantıları Tabanlı Uzun Kısa Süreli Bellek Mimarisi. European Journal of Science and Technology (2022). https://doi.org/10.31590/ejosat.1080239
• Tokcaer, S. Türkçe metinlerde duygu analizi. Journal of Yaşar University 16(63), 1516–1536 (2021). https://doi.org/10.19168/jyasar.928843
• Ülgen, O., Öğrenci, A.S. Structured learning based turkish sentiment analysis. Journal of Intelligent Systems with Applications 2(2), 90–93 (2019), https://joiswa.com/abstract.php?id=71
• Uyrun, Ö.F., Sabuncu, İ. Sosyal Medya ve Diğer Yatırım Aracı Verilerine Dayalı Hisse Senedi Değeri Tahmini. Acta Infologica 5(2), 267–285 (2021). https://doi.org/10.26650/acin.934130
• Yürütücü, Y. Ön eğitimli dil modelleriyle duygu analizi. İstanbul Sabahattin Zaim Üniversitesi Fen Bilimleri Enstitüsü Dergisi 5(1), 46 53 (2023). https://doi.org/10.47769/izufbed.1312032
• Zada, A.J.J., Albayrak, A. Duygu analizi ve topluluk Öğrenmesi yaklaşımları ile kullanıcı yorumlarının analizi. Düzce Üniver sitesi Bilim Ve Teknoloji Dergisi 11(4), 1725–1732 (2023). https://doi.org/10.29130/dubited.1102181
• Çancıoğlu, E., ¸Sahin, S., İşler, Y. Heart sounds analysis and classification based on long-short term memory. Journal of Intelligent Systems with Applications 3(1), 25–28 (2020), https://joiswa.com/issue.php?issue=5
• Çetin, F.S., Eryiğit, G. Türkçe hedef tabanlı duygu analizi ˙ Için alt görevlerin ˙ Incelenmesi– hedef terim, hedef kategori ve duygu sınıfı belirleme. Bili¸sim Teknolojileri Dergisi 11(1), 43–56 (2018). https://doi.org/10.17671/gazibtd.325865
• Öğe, B.C., Kayaalp, F. Farklı sınıflandırma algoritmaları ve metin temsil yöntemlerinin duygu analizinde performans kar¸sıla¸stırılması. Düzce Üniversitesi Bilim ve Teknoloji Dergisi 9(6), 406–416 (2021). https://doi.org/10.29130/dubited.1015320
• Şahınaslan, Ö, Dalyan, H., Şahınaslan, E. Naive Bayes Sınıflandırıcısı Kullanılarak Youtube Verileri Üzerinden Çok Dilli Duygu Analizi. Bilişim Teknolojileri Dergisi 15(2), 221–229 (2022). https://doi.org/10.17671/gazibtd.999960