Social networks have become a customary news media source in recent times. However the openness and unrestricted way sharing the information on social networks fosters spreading rumors which may cause severe damages economically, socially, etc. Motivated by this, our paper focuses on the rumor detection problem in Algerian Arabizi. Studying linguistic rules of the Algerian Arabizi, we propose an lemmatiser and a parser for analysing and standardizing the text in order to produce better rumor detection models. An approach for classifying rumors and news in social networks based on the expression of emotions and positions of users is proposed. The experiments were done on many ngram representation and the best one has reached more than 94% for f-score. In addition to that this research deals with resources creation for Algerian Arabizi which is an under-resourced dialect. A corpus and several lexicons have been built and which can be the subject of other works dealing with this dialect.