NOOR dataset

A Classical Arabic dataset for stemming and morphological analysis containing approximately 260,000 word–stem pairs. Only a subset of 7,042 word–stem pairs (derived from the Holy Quran) is publicly available. The data is formatted for supervised learning and provided in structured formats such as CSV and XML,PDF supporting character-level  sequence-to-sequence NLP models. The dataset can be used for Arabic stemming, lemmatization, morphological analysis, root extraction, and evaluation of deep learning models in Classical Arabic NLP.