Ultimate Arabic News Dataset: A New Efficient Dataset for Arabic Text Classification

The rapid growth of the internet in recent years and the huge increase in the number of internet users has been accompanied by a huge increase in the number of data being shared at every moment, especially text data. With such a large amount of data, text classification (aka categorization) has become one of the most active research topics in recent years. Text classification is one of the most common natural language processing (NLP) tasks where unclassified texts are categorized into predefined categories. Many studies have been conducted in the field of text classification, but research on Arabic text classification has been very limited due to many reasons, including the complexity of the Arabic language and the absence of publicly available, high-resolution Arabic datasets. In this study, a new huge Arabic news dataset called "Ultimate Arabic News Dataset" is presented. The data was collected with high accuracy and efficiency from different Arab news sources, and the documents were divided into ten different categories. The dataset contains four different versions ranging from the original data, without any processing techniques, to the versions on which all pre-processing techniques were applied for the researcher to choose the appropriate dataset for their studies. The different versions of the dataset were evaluated using four classifiers and three term weighting methods and the dataset, in all its versions, provided very good results. With the good results shown by the new dataset, which will be presented and discussed in the experimental study section, it is expected that this dataset will effectively contribute to helping other researchers and developing research dealing with the tasks of natural language processing and classification of Arabic text data.