Noor dataset

Noor dataset include two thousand Arabic words, selecting them from various sources such as the Holy Quran and classical Arabic words. These words are assigned to two groups of human experts so that they determine the correct stem for each word.
The first collection of selected words comprises non-repetitive words from the Quran, categorized based on morphological structure. This collection contains over 16,000 words, focusing exclusively on their Quranic usage and the corresponding labels are only the word stems. The necessity of morphological analysis in Quranic texts, as an example of classical Arabic, underscores the importance of this evaluation.
The second collection consists of 10,000 words from non-repetitive text data in general classical Arabic texts. From a pool of more than 2,600,000 non-repetitive words, 10,000 were chosen. Since this dataset aims to be a gold standard, each stem is verified and labeled by a pair of experts.