الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 114 |  |  |
| | | from | 114 | | |
Abstract
In the past few years, social media has exploded all over the world and has become an indispensable aspect of our existence. Social media can be a double-edged sword as it has many benefits in reducing barriers to communication, expanding social circles, aiding learning, and assisting entrepreneurs and companies. On the other hand, it can be used to spread rumors, spread false concepts, hate, and violence. from this standpoint, the need arose to analyze short Arabic texts on social media so that we could benefit from these analyses. Especially after the growth of the number of Arab users on social media thanks to the spread of the Internet and the improvement of infrastructure. Also, Arabic has become the fourth most significant language on the Internet. Depending on the similarity analysis of Arabic short texts using deep learning techniques, we can benefit from these analyzes in various fields, but the Arabic language has a complex nature, due to its ambiguity and rich morphological-logical system. This nature, with its limited research resources and dialectal diversity, poses challenges to progress in Arabic language analysis research. In addition to monitoring a major challenge in the increasing use of the Franco-Arabic language among the youth community on social media, the scarcity of its resources and the lack of research conducted to study this challenge.
Recently, deep learning has shown remarkable improvements in the field of text analysis in the English language. Little research has been done on the use of deep learning in analyzing Arabic text as well as challenging Franco-Arabic. In this thesis, we introduce intelligent natural language processing models that analyze the meaning similarity of Arabic short texts on social media by measuring context similarity between posts and their relevance to each other. We will highlight the challenges in dealing with the Franco-Arabic language and present new methods using deep learning techniques to deal with this challenge.