PulseAugur
EN
LIVE 07:49:25

New system detects offensive Nastaliq Urdu content with 87% accuracy

By PulseAugur Editorial · [1 sources] ·

Researchers have developed an explainable system for detecting offensive content in Nastaliq Urdu, a script primarily used on platforms like Facebook. The study involved curating a new corpus of over 7,600 comments, fine-tuning four Urdu-capable transformer models, and comparing their performance against baseline classifiers. The best-performing model achieved 87.05% accuracy, outperforming baselines by a significant margin. Further analysis using word-level SHAP attributions revealed that while performance partly relies on identifying slurs, it also incorporates function words and political vocabulary, suggesting a complex understanding of offensiveness. AI

IMPACT Improves AI's ability to moderate content in under-resourced languages, potentially enabling broader global platform safety.

RANK_REASON Academic paper detailing a new corpus, model benchmarking, and explainability techniques for a specific language and script. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

New system detects offensive Nastaliq Urdu content with 87% accuracy

How we ranked this

Signal score
20 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
Academic paper detailing a new corpus, model benchmarking, and explainability techniques for a specific language and script. [lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration
Single-source cluster
Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics
paper, safety
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
Breaking (< 6h)
Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

Full methodology in our editorial standards.

COVERAGE [1]

  1. arXiv cs.AI TIER_1 English(EN) · Muhammad Shahid Iqbal Malik ·

    An Explainable System for Offensive Content Detection in Nastaliq Urdu: Corpus Curation, Transformer Benchmarking, and Word-Level Attribution

    arXiv:2610.15795v1 Announce Type: cross Abstract: Automated moderation systems are deployed at scale, yet they remain opaque to the users and moderators affected by their decisions, and their coverage is almost entirely confined to English. For Urdu written in the Nastaliq script…

RELATED ENTITIES

RELATED TOPICS