DeformAR: A Visual Analytics Framework for Evaluation of Arabic Named Entity Recognition

ACL ARR 2025 May Submission4212 Authors

Abstract: Arabic Named Entity Recognition (ANER) presents challenges due to its linguistic characteristics. We present DeformAR, a visual analytics framework for evaluating and interpreting Arabic NER models via a structured, component-based approach. DeformAR combines quantitative metrics and qualitative visualisation across data and model subcomponents to identify performance weaknesses and explain system behaviour. In a case study on ANERCorp, DeformAR identifies annotation mistakes, explains calibration issues, and reveals interaction effects between subcomponents. To our knowledge, this is the first framework to support both evaluation and interpretability for Arabic NER.

Paper Type: Long

Research Area: Interpretability and Analysis of Models for NLP

Research Area Keywords: named entity recognition and relation extraction, calibration/uncertainty, explanation faithfulness, less-resourced languages, evaluation methodologies, human-in-the-loop, evaluation and metrics

Contribution Types: Model analysis & interpretability, Approaches to low-resource settings, Publicly available software and/or pre-trained models, Data analysis

Languages Studied: Arabic

Submission Number: 4212