中文摘要：

针对现有的局部信任度计算方法忽视了交互经验的时效性及充分性等问题，提出一种证据理论框架下的局部信任度计算方法，该方法采用了时效因子计算函数来区分不同时间内的交互经验在局部信任度计算中的重要性；同时，采用了半梯形函数来区分基于不同交互经验计算得到的局部信任度的有效性。实验分析表明，该方法对实体行为改变有较强的敏感性，能有效地降低对各种恶意实体的局部信任度。

英文摘要：

This paper proposed a new computation method for local trust value based on D-S theory of evidence, while the ex- istent methods for local trust value were based on count of success and failure times of interaction and ignored the effect on the local trust degree of adequacy and timeliness of interaction experience. In the method, introduced a normalized sigmoid func- tion to distinguish the importance of interactive experiences occurred in different time and used a semi-trapezoidal function to distinguish the adequacy of interactive experiences. Analysis and experiment show that it is sensitive to entity behavior altering and effective in reducing the local trust degree placed on malicious entity.