EUVIMEDEUVIMED entdecken
Crossref

Enhancing biomedical text interpretation through neuro-symbolic knowledge integration

Zahraa Tarek, Esraa Hassan

An International Journal of Optimization and Control: Theories & Applications (IJOCTA) · 05.05.2026 · 16 (3) · 816

DOI: 10.36922/ijocta026070024 ↗

Kurz gefasst

Fachartikel in der Zeitschrift An International Journal of Optimization and Control: Theories & Applications (IJOCTA), online erschienen am 05.05.2026. Verfasst von Zahraa Tarek und 1 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The integration of statistical learning with symbolic reasoning provides a promising pathway toward artificial intelligence systems that are not only highly accurate but also transparent and trustworthy. In this study, we propose a neuro-symbolic framework that integrates neural models for robust pattern recognition with symbolic reasoning mechanisms designed to enforce domain-informed logical constraints, thereby improving the interpretation of biomedical text. The proposed framework is evaluated through a comprehensive experimental protocol that assesses classification performance, statistical robustness, and prediction reliability. Experimental results on a multi-class biomedical text classification task demonstrate that the proposed approach consistently outperformed conventional machine learning baselines, including term frequency--inverse document frequency-based logistic regression. Specifically, the neuro-symbolic model achieved a macro-averaged F1-score of 0.863, compared to 0.724 for the baseline model, representing a 19.2\% relative improvement. Furthermore, the framework exhibited strong predictive stability across multiple biomedical categories, with classification accuracies ranging from 0.938 to 0.950 across topics. The highest performance was observed for sclerenchyma-related texts (accuracy = 0.950), while all other categories maintained accuracies above 0.94, indicating consistent and reliable classification performance. In addition, the proposed framework substantially improved model calibration, achieving a 63\% reduction in expected calibration error and yielding more reliable probability estimates for decision-support applications.

Zitieren

APA

Tarek, Z., & Hassan, E. (2026). Enhancing biomedical text interpretation through neuro-symbolic knowledge integration. An International Journal of Optimization and Control: Theories & Applications (IJOCTA), 16(3), 816. https://doi.org/10.36922/ijocta026070024

Vancouver

Tarek Z, Hassan E. Enhancing biomedical text interpretation through neuro-symbolic knowledge integration. An International Journal of Optimization and Control: Theories & Applications (IJOCTA). 2026;16(3):816. doi:10.36922/ijocta026070024

BibTeX
@article{Tarek2026,
  author = {Tarek, Zahraa and Hassan, Esraa},
  title = {Enhancing biomedical text interpretation through neuro-symbolic knowledge integration},
  journal = {An International Journal of Optimization and Control: Theories & Applications (IJOCTA)},
  year = {2026},
  volume = {16},
  number = {3},
  pages = {816},
  doi = {10.36922/ijocta026070024}
}
Kostenlos starten →