Youssef Tag Eldin

Youssef Tag Eldin

Machine Learning Engineer with experience in deep learning, classical ML, and applied AI

A Data Centric HitL Framework for Conducting a Systematic Error Analysis of NLP Datasets using Explainable AI

Published in Scientific Reports, 2025

Published in Nature Scientific Reports — Featured case study on Arabic emotion detection using a novel data-centric AI approach.

X-Deep Framework

Framework Diagram The X-Deep Human-in-the-Loop framework for systematic dataset debugging

Abstract

The interest in data-centric AI has been recently growing. As opposed to model-centric AI, data-centric approaches aim at iteratively and systematically improving the data throughout the model life cycle rather than in a single pre-processing step.

X-Deep Framework: A Human-in-the-Loop framework designed to debug NLP datasets using Explainable AI techniques.

Explainable AI Analysis

LIME and SHAP Example LIME and SHAP outputs revealing how repeated words influence emotion classification

Key Contributions

  • Case Study: Emotion detection in Arabic text using LIME and SHAP
  • Multi-Classifier Analysis: Naive Bayes, Logistic Regression, GRU, MARBERT
  • Pattern Discovery: Identified spurious correlations, bias patterns, and anomalies
  • Actionable Insights: Mitigation strategies for improved data collection and augmentation

Publication Details

Published: August 19, 2025 | Volume: 15, Article 30406
DOI: 10.1038/s41598-025-13452-y

View Full Portfolio Project

Recommended citation: El-Sayed, A., Nasr, A., Mohamed, Y. et al. A Data Centric HitL Framework for Conducting a Systematic Error Analysis of NLP Datasets using Explainable AI. Sci Rep 15, 30406 (2025). https://doi.org/10.1038/s41598-025-13452-y
Download Paper

Direct Link