BOUBAAYA Hocine2025-11-052025https://dspace.univ-bba.dz/handle/123456789/956This research focuses on the process of stemming in Arabic texts, a fundamental step in Arabic Natural Language Processing (NLP). It aims to propose a novel hybrid stemming method that combines statistical techniques, semantic resources, and machine learning models to enhance the accuracy of root extraction. The work includes a critical review of existing Arabic stemming approaches, a comparative evaluation of statistical methods, and the development of a flexible statistical model based on morphological rules. The proposed method is tested on a corpus of Arabic texts, and the results demonstrate its superiority in terms of precision and linguistic coverage compared to traditional stemmers.frNatural Language ProcessingStatistical MethodsStemmingMorphologyRoot ExtractionArabic LanguageArabic CorporaLexical Resources.Thesis