TY  - JOUR
AU  - Khan, Bilal
AU  - Khan, Khairullah
AU  - Khan, Fida Muhammad
AU  - Noureen, Haseena
AU  - Ali, Ahmad
AU  - Shah, Mohsin
PY  - 2024
DA  - 2025/05/25
TI  - Comparing Fine-Tuned RoBERTa with Traditional Machine Learning Models for Stance Detection in Political Tweets
JO  - ICCK Transactions on Advanced Computing and Systems
T2  - ICCK Transactions on Advanced Computing and Systems
JF  - ICCK Transactions on Advanced Computing and Systems
VL  - 1
IS  - 2
SP  - 78
EP  - 96
DO  - 10.62762/TACS.2025.928069
UR  - https://www.icck.org/article/abs/TACS.2025.928069
KW  - stance detection
KW  - Roman Urdu
KW  - machine learning
KW  - SVM
KW  - random forest
KW  - logistic regression
KW  - naïve Bayes
KW  - decision tree
KW  - RoBERTa
AB  - Stance detection identifies a text’s position or attitude toward a given subject. A major challenge in Roman Urdu is the lack of a publicly available dataset for political stance detection. To address this gap, we constructed a high-quality dataset of 8,374 political tweets and comments using the Twitter API, annotated with stance labels: agree, disagree, and unrelated. The dataset captures diverse political viewpoints and user interactions. For feature representation, we employed TF-IDF due to its effectiveness in handling high-dimensional, context-sensitive Roman Urdu text. Several machine learning classifiers were evaluated, with Random Forest achieving the highest accuracy of 95%. Additionally, we fine-tuned the transformer-based RoBERTa model, which outperformed traditional methods with 97% accuracy. Our results demonstrate the potential of combining machine learning and deep learning for stance detection in low-resource languages. This study not only introduces a novel dataset but also provides a robust evaluation of methods, highlighting the importance of modern AI techniques in processing informal and multilingual text data.
SN  - 3068-7969
PB  - Institute of Central Computation and Knowledge
LA  - English
ER  - 

@article{Khan2024Comparing,
  author = {Bilal Khan and Khairullah Khan and Fida Muhammad Khan and Haseena Noureen and Ahmad Ali and Mohsin Shah},
  title = {Comparing Fine-Tuned RoBERTa with Traditional Machine Learning Models for Stance Detection in Political Tweets},
  journal = {ICCK Transactions on Advanced Computing and Systems},
  year = {2024},
  volume = {1},
  number = {2},
  pages = {78-96},
  doi = {10.62762/TACS.2025.928069},
  url = {https://www.icck.org/article/abs/TACS.2025.928069},
  abstract = {Stance detection identifies a text’s position or attitude toward a given subject. A major challenge in Roman Urdu is the lack of a publicly available dataset for political stance detection. To address this gap, we constructed a high-quality dataset of 8,374 political tweets and comments using the Twitter API, annotated with stance labels: agree, disagree, and unrelated. The dataset captures diverse political viewpoints and user interactions. For feature representation, we employed TF-IDF due to its effectiveness in handling high-dimensional, context-sensitive Roman Urdu text. Several machine learning classifiers were evaluated, with Random Forest achieving the highest accuracy of 95\%. Additionally, we fine-tuned the transformer-based RoBERTa model, which outperformed traditional methods with 97\% accuracy. Our results demonstrate the potential of combining machine learning and deep learning for stance detection in low-resource languages. This study not only introduces a novel dataset but also provides a robust evaluation of methods, highlighting the importance of modern AI techniques in processing informal and multilingual text data.},
  keywords = {stance detection, Roman Urdu, machine learning, SVM, random forest, logistic regression, naïve Bayes, decision tree, RoBERTa},
  issn = {3068-7969},
  publisher = {Institute of Central Computation and Knowledge}
}