@article{article_1655764, title={User interest classification on social media using machine learning and deep learning models: A multi-domain approach}, journal={Niğde Ömer Halisdemir Üniversitesi Mühendislik Bilimleri Dergisi}, volume={14}, pages={1428–1435}, year={2025}, DOI={10.28948/ngumuh.1655764}, author={Zeybek, Sultan and Kaçaman, Melih}, keywords={Makine Öğrenmesi, Sosyal Medya, Kullanıcı Profili Analizi, Sınıflandırma}, abstract={The analysis of user-generated textual content provides valuable insights into user preferences in various domains, including politics, entertainment, health, sports, food, and technology. This study aims to automatically classify X user profiles based on interests using machine learning and deep learning algorithms. The objective is to categorize users into six interest areas with techniques including Naive Bayes, Logistic Regression, and Support Vector Machines, as well as LSTM, GRU, Bidirectional RNN, Conv1D, and Dense networks. Machine learning and deep learning models were compared using a pooled dataset, revealing that deep learning approaches generally improved generalization ability. Results indicate that while deep learning models achieve higher performance with large datasets, machine learning algorithms also perform competitively in certain categories. The findings highlight the potential of these models to support applications such as targeted content delivery, personalized recommendation systems, and user profiling on social media platforms.}, number={4}, publisher={Niğde Ömer Halisdemir Üniversitesi}