Because of the quick development of organization information, the validness and dependability of organization data have become progressively significant and have introduced difficulties. A large portion of the strategies for counterfeit survey recognition start with printed highlights and social elements. Notwithstanding, they are tedious and effortlessly identified by false clients. Albeit the vast majority of the current brain network-based strategies address the issues introduced by the complicated semantics of audits, they don't represent the understood examples among clients, surveys, and items; also, they don't consider the convenience of data with respect to ne-grained perspectives in recognizing counterfeit audits. In this paper, we propose a consideration based staggered intuitive neural network model with multilevel imperatives that mines the staggered implied articulation method of audits and coordinates four aspects, to be specific, clients, survey messages, items and fine-grained perspectives, into audit portrayals. And predict the fake reviews using Navie bayes, SVM and Logistic regression algorithms. Finally, we show the simulation results of these models in detecting the fake reviews.