In order to improve the flexibility of trust computing in the model,the trusted domain was divided into three decision domains.Problem of adaptability was solved by introducing time adjustment into the model.Stability function tackled weight distribution problem during calculating total value.Consideration of interaction history and the distribution of trust while making evaluation strategy saved computing resources.Moreover,investigation table about direct recommendation entity was put forward to identify and restrict false recommendation problems effectively.Simulation results showed that this model has a better accuracy,convergence and adaptability in dealing with complex malicious attacks and fraud in the model.