The aim of this study is to try to investigate the problem of deviation in non-standard data generated by inquiry into Arabic language, using research engines or some Natural Language Processing programs (NLP) applied to Arabic. The deviation could be corrected by using expert systems and some neural networking. It will try to analyze these issues from both linguistic point of view and technological point of view in order to find the best way to build a model of work that could find solutions for some deviant cases using Artificial Intelligence (Expert Systems and neural networks). Above all, this work could serve as a starting point for farther research on the deviant cases.