Trust is an important issue in e-commerce and e-service platforms. Factors to calculate trust level have been studied to provide valuable information to buyers to avoid fraud exchanges. While some malicious sellers or service providers and fraudulent transactions still widely exist. A typical attack is to attract buyers with ideal price or service but deliver a fake item or nothing at all after the payment. This attack is often carried out by a newly registered seller. This paper proposes a new trust evaluation model to indicate to some extent the trust level of a seller or service provider at different registered time.