Transformers tabanlı ile duygu sınıflandırma, son zamanlarda doğal dil işleme ve makine öğrenmesi alanında yaygın olarak çalışılan bir konudur. Metinler içerisinde karşılaşılan duygusal ifadelerin anlamlandırılması ve sınıflandırılması, sosyal medya analizi, piyasa araştırması, kullanıcı deneyimleri vb. gibi kullanılabileceği birçok alan mevcuttur. Bu sebeple, bu çalışmada Transformers tabanlı mimariler kullanılarak duygu sınıflandırmasının gerçekleştirilmesi hedeflenmiştir. Bu çalışmada, 150000 veriden oluşan TRSAv1 veriseti üzerinde, 8 farklı BERTurk ve 2 farklı ELECTRA varyasyonu üzerinde duygu sınıflandırma işlemi için kullanılmıştır. Bu modeller, Türkçe metinler üzerinde duygu sınıflandırılması çalışmalarında kullanılmak için önceden eğitilmiş modellerdir. Veri seti üzerinde 3 farklı metot kullanılarak modeller eğitilmiş ve sonuçlar karşılaştırmalı olarak değerlendirilmiştir. Yapılan deneyler sonucunda, modellerin duygu sınıflandırma performansları doğruluk ve F1-skor metrikleri kullanılarak ölçülmüştür. Deney sonuçları, Transformers modellerinin duygu sınıflandırması konusundaki etkinliğini ve kullanılan modellerin performans değerlendirmelerini ortaya koymuştur.

Question-answering systems facilitate information access processes by providing fast and accurate answers to questions that users express in natural language. Today, advances in Natural Language Processing (NLP) techniques increase the effectiveness of such systems and improve the user experience. However, for these systems to work effectively, an accurate understanding of the structural properties of language is required. Traditional rule-based and knowledge retrieval-based systems are not able to analyze the contextual meaning of questions and texts deeply enough and therefore cannot produce satisfactory answers to complex questions. For this reason, Transformer-based models that can better capture the contextual and semantic integrity of the language have been developed. In this study, within the scope of the developed models, the performances of BERTurk, ELECTRA Turkish and DistilBERTurk models for Turkish question-answer tasks were compared by fine-tuning under the same hyperparameters and the results obtained were evaluated. According to the findings, it was observed that higher Exact Match (EM) and F1 scores were obtained in models with case sensitivity; the best performance was obtained with 63.99 EM and 80.84 F1 scores in the BERTurk (Cased, 128k) model.

This study comparatively evaluates the performance of Convolutional Neural Networks (CNN), Vision Transformer (ViT), and Swin Transformer architectures for diagnosing plant leaf diseases. The experimental framework was evaluated using model effectiveness, F1 scores, and confusion matrices on the PlantVillage dataset, which contains 55,448 images from 39 different classes. The results show that the Swin Transformer achieved the highest diagnostic performance with a 100% F1 score, outperforming both CNN and ViT models. These findings confirm the high potential of transformer-based architectures for automated disease detection in agriculture.

Short Message Service is one of the most widely used communication tools today. This widespread use has also led to an increase in the number of unwanted messages. These messages, sent for the purposes of fraud, advertising, and promotion, are messages that users do not wish to receive. This study addresses the classification of messages received via SMS into spam and non-spam categories. To this end, a project was developed using a specialized SMS dataset. Preprocessing steps were applied to the dataset to remove meaningless features from the text, and then two different classical machine learning algorithms and two different deep learning algorithms were used. These algorithms are Naive Bayes, Random Forest, BERTurk, and Turkish ELECTRA, respectively. Classification models were developed using these algorithms. During the training process, the dataset was split into an 80% training set and a 20% test set; additionally, a 5-fold cross-validation method was applied to verify the stability of the results. The models were trained on the preprocessed data and analyzed and compared using performance metrics such as Precision, Recall, F1-Score, and Accuracy. Analyses conducted on the dataset indicate that the unprocessed BERTurk model, evaluated using 5-fold cross-validation, achieved the best result with an accuracy of 0.990. The results demonstrate that different algorithms offer distinct advantages depending on the data structure.