EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Beyond Accuracy: Explainability of Transformer Models for Sentiment Analysis

Shahriyar Shahid, Abdul Aziz, Muhammad Bilal, Muhammad Shoaib, Muhammad Aaqib

International Journal of Innovations in Science and Technology · 26.12.2025 · 300

DOI: 10.33411/ijist/1704 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Journal of Innovations in Science and Technology, online erschienen am 26.12.2025. Verfasst von Shahriyar Shahid und 4 weiteren Autor:innen. Die Arbeit nennt 25 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Transformer-based models achieve strong performance on Urdu sentiment analysis; however, their predictions are often difficult to interpret, which can undermine trust in practical applications. In this paper, we present a Hybrid LIMESHAP Attribution (HLSAE) Module. We fine-tune Urdu BERT and introduce a token-level explanation pipeline that combines LIME and SHAP. The method is evaluated on a dataset of 50,000 Urdu movie reviews. Compared with LIME or SHAP individually, the hybrid approach produces more stable attributions across multiple runs and perturbations. In a human evaluation involving native Urdu speakers, the explanations achieve a fidelity score of 0.72, while the classifier maintains an F1-score of 79.8%. The resulting visualizations highlight sentiment-bearing linguistic cues and reveal instances where the model relies on spurious correlations. These results demonstrate that explainability can be incorporated without sacrificing classification accuracy in low-resource sentiment analysis, and the proposed workflow can be extended to similar languages.

Zitieren

APA

Shahid, S., Aziz, A., Bilal, M., Shoaib, M., & Aaqib, M. (2025). Beyond Accuracy: Explainability of Transformer Models for Sentiment Analysis. International Journal of Innovations in Science and Technology, 300. https://doi.org/10.33411/ijist/1704

Vancouver

Shahid S, Aziz A, Bilal M, Shoaib M, Aaqib M. Beyond Accuracy: Explainability of Transformer Models for Sentiment Analysis. International Journal of Innovations in Science and Technology. 2025:300. doi:10.33411/ijist/1704

BibTeX
@article{Shahid2025,
  author = {Shahid, Shahriyar and Aziz, Abdul and Bilal, Muhammad and Shoaib, Muhammad and Aaqib, Muhammad},
  title = {Beyond Accuracy: Explainability of Transformer Models for Sentiment Analysis},
  journal = {International Journal of Innovations in Science and Technology},
  year = {2025},
  pages = {300},
  doi = {10.33411/ijist/1704}
}
Kostenlos starten →