Enhancing Arabic Web Article Classification through Label Noise Reduction and Class-Balanced Training
International Journal on Advanced Electrical and Computer Engineering · 19.01.2026 · 15 (1S) · 110-118
DOI: 10.65521/ijaece.v15i1s.1347 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal on Advanced Electrical and Computer Engineering, online erschienen am 19.01.2026. Verfasst von Mohammed Ahmed Aledresi und 1 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY-ND 4.0).
Abstract (Originalsprache)
We present an applied, data-centric study on improving large-scale Arabic web-article classification by detecting and correcting label noise in a 451K-article corpus. Building on a practical AraBERTv02 fine-tuned baseline, we develop an automated noise-detection and relabelling pipeline that combines model confidence, MC-dropout ensemble agreement, and probabilistic flagging inspired by Confident Learning. Using stratified reduced experiments (Train=50k, Val=5k, Test=5k) we automatically relabelled 880 training examples and retrained the classifier. On the held-out 5k test set the cleaned model improved Macro-F1 from 0.9367 (baseline) to 0.9503, a statistically significant gain confirmed by McNemar’s test (p ≈ 1.9e-5) and bootstrap 95% CI for Δ Macro-F1 = [0.0082, 0.0190]. We analyse per-class gains, error modes, and cost trade-offs of automated relabelling versus manual review, and release anonymized preprocessing scripts and experiment artifacts to support reproducibility. Our results show that conservative automatic relabelling — judiciously combined with uncertainty estimation — can yield meaningful, reproducible improvements for large-scale Arabic article classification.
Zitieren
APA
Aledresi, M. A., & Alsubari, A. (2026). Enhancing Arabic Web Article Classification through Label Noise Reduction and Class-Balanced Training. International Journal on Advanced Electrical and Computer Engineering, 15(1S), 110-118. https://doi.org/10.65521/ijaece.v15i1s.1347
Vancouver
Aledresi MA, Alsubari A. Enhancing Arabic Web Article Classification through Label Noise Reduction and Class-Balanced Training. International Journal on Advanced Electrical and Computer Engineering. 2026;15(1S):110-118. doi:10.65521/ijaece.v15i1s.1347
BibTeX
@article{Aledresi2026,
author = {Aledresi, Mohammed Ahmed and Alsubari, Akram},
title = {Enhancing Arabic Web Article Classification through Label Noise Reduction and Class-Balanced Training},
journal = {International Journal on Advanced Electrical and Computer Engineering},
year = {2026},
volume = {15},
number = {1S},
pages = {110--118},
doi = {10.65521/ijaece.v15i1s.1347}
}
EUVIMED