EUVIMEDEUVIMED entdecken
Crossref

Deep Learning vs. Machine Learning in Sentiment Classification: A Comparative Analysis of Mobile Game Tweets from the X Platform

Erol Kına, Recep Özdağ

Erzincan Üniversitesi Fen Bilimleri Enstitüsü Dergisi · 31.08.2025 · 18 (2) · 639-658

DOI: 10.18185/erzifbed.1667207 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Erzincan Üniversitesi Fen Bilimleri Enstitüsü Dergisi, online erschienen am 31.08.2025. Verfasst von Erol Kına und 1 weiteren Autor:innen. Die Arbeit wurde bisher 3-mal zitiert (Crossref) und nennt 30 Literaturangaben.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This paper presents an overview based on comparison of different machine and deep learning methods applied to perform the sentiment analysis of tweets related to mobile games. The dataset, gathered from Twitter (X) between 2020-2021, was preprocessed and vectorized using Count Vectorizer and TF-IDF methodology. Traditional machine learning (ML) models such as Linear Support Vector Classifier (SVC), Logistic Regression (LR), Ridge Classifier (RC), and Voting Classifier (VC) were benchmarked against a few deep learning (DL) architectures such as the TEMSAP-CNNLSTM model stand-alone and BERT-enhanced versions. The study used precision, F1-score, recall, accuracy, and the AUC to check the performance of the model. The results revealed that DL models outperformed traditional ML classifiers, with these models achieving the highest classification performance of 97,10% and achieving impressive success in minimizing false negatives and false positives. The Ridge Classifier exhibited the lowest performance, correctly classified twitter reviews at an accuracy of 76,76%, indicating its limitations in sentiment classification. In addition, ensemble learning techniques like the Voting Classifier performed much better than individual machine learning models, thus re-establishing the benefits of model aggregation. This study demonstrated that transformer-based models such as BERT have shown remarkable success in sentiment classification of text data related to mobile games. What is even more promising for furthering academic and industrial agendas is that it will provide informed insights into how to make the best selection for enhancing the analysis of user sentiment and identifying the best models to make the play of mobile games more entertaining.

Zitieren

APA

Kına, E., & Özdağ, R. (2025). Deep Learning vs. Machine Learning in Sentiment Classification: A Comparative Analysis of Mobile Game Tweets from the X Platform. Erzincan Üniversitesi Fen Bilimleri Enstitüsü Dergisi, 18(2), 639-658. https://doi.org/10.18185/erzifbed.1667207

Vancouver

Kına E, Özdağ R. Deep Learning vs. Machine Learning in Sentiment Classification: A Comparative Analysis of Mobile Game Tweets from the X Platform. Erzincan Üniversitesi Fen Bilimleri Enstitüsü Dergisi. 2025;18(2):639-658. doi:10.18185/erzifbed.1667207

BibTeX
@article{Kna2025,
  author = {Kına, Erol and Özdağ, Recep},
  title = {Deep Learning vs. Machine Learning in Sentiment Classification: A Comparative Analysis of Mobile Game Tweets from the X Platform},
  journal = {Erzincan Üniversitesi Fen Bilimleri Enstitüsü Dergisi},
  year = {2025},
  volume = {18},
  number = {2},
  pages = {639--658},
  doi = {10.18185/erzifbed.1667207}
}
Kostenlos starten →