In an open system, trust is one of the most complex concept in social relationships, involving many decision factors, such as assumptions, expectations and behaviors, etc. So, it is very difficult to quantify and forecast accurately. Combined with human social cognitive behaviors, a new dynamic trust forecasting model is proposed. Firstly, a new and adaptive trusted decision-making method based on historical evidences window is proposed, which can not only reduce the risk and improve system efficiency, but also solve trust measurement and forecasting problem when the direct evidences are insufficient. Then, a feedback trust information aggregating algorithm is used based on DTT (direct trust tree). Finally, Induced Ordered Weighted Averaging (IOWA) operator is introduced to construct the new combined direct dynamic trust forecasting model, to make up the shortage of traditional method, and the model hence can have a better rationality and a higher practicability. Simulations’ computing results show that compared with the existing trust forecasting metrics, the model in this paper is more robust on trust dynamic adaptability, and has more remarkable enhancements in the forecasting accuracy.