This paper focuses on the topic identification for the Arabic language based on topic models. We study the Latent Dirichlet Allocation (LDA) as an unsupervised method for the Arabic topic identification. Thus, a deep study of LDA is carried out at two levels: Stemming process and the choice of LDA hyper-parameters. For the first level, we study the effect of different Arabic stemmers on LDA. For the second level, we focus on LDA hyper-parameters α and β and their impact on the topic identification. This study shows that LDA is an efficient method for Arabic topic identification especially with the right choice of hyper-parameters. Another important result is the high impact of the stemming algorithm on topic identification.