Building Arabic morphological analysis with natural language processing algorithm based on patterns

Morphological analysis is a necessary step in the great majority of Natural Language Processing (NLP) applications, including text categorization, machine translation, indexing, information retrieval, and many more. Arabic is a deep-meaning language with a complex word structure that is rich in derivatives. That's why it requires a lot of work to cope with. Arabic words are constructed from roots and patterns. Arabic is a language with a lot of inflections and a complicated morphological structure, it is a derivational language with a very rich derivational morphology, with nearly all words deriving from roots through patterns. We have created and extended numerous patterns to accommodate all kinds of Arabic words, even though the language has highly specific patterns. After patterns are used to identify roots, words' morphological structures can be ascertained, or words or stems can be formed from these patterns. This morphological analyzer can recognize roots, nouns, verbs, and all types of plurals and singulars, and it can also generate all inflections for any word. In this thesis proposed two methods; first, it does not rely on any linguistic rules, which eliminates any possibility of confusion between the original letters and the letters added to the word; second, it retains all word affixes, suffixes, and infixes, resulting in a shorter processing time. Finally, the algorithm was evaluated using performance metrics and 0.94 for Precision, 0.97 for Recall, 0.95 for F1-score, and 0.92 for Accuracy were obtained. This shows that the algorithm has a high level of performance.