In web-based item audit frameworks, clients are permitted to submit surveys about their bought things or administrations. Nonetheless, counterfeit surveys posted by fake clients frequently delude buyers and carry misfortunes to endeavors. Customary misrepresentation recognition calculation chiefly uses rule-based techniques, which is deficient for the rich client collaborations and chart organized information. As of late, chart based strategies have been proposed to deal with this present circumstance, however not many earlier works have seen the disguise fraudster's way of behaving and irregularity heterogeneous nature. Existing strategies have either not resolved these two issues or just somewhat, which brings about horrible showing. On the other hand, we propose another model named Fraud Aware Heterogeneous Graph Transformer (FAHGT), to address covers and irregularity issues in a brought together way. FAHGT embraces a sort mindful component planning system to deal with heterogeneous diagram information, then, at that point, executing different connection scoring techniques to lighten irregularity and find disguise.