一种内生式信誉系统不实评价过滤方法

Endogenous unfair rating filtering method for reputation system

提出一种应用于Beta reputation信誉系统的内生式不实评价过滤方法。该方法以交易次数的对数做为权重进行均值计算,以此限制评价者虚报过大交易量;以加权方差定义容限,识别并过滤与平均信誉度差别较大的评价。仿真结果表明,该方法能有效过滤不实评价,提高信誉计算的精度。

In this paper,the authors propose an endogenous unfair rating filtering method for Beta reputation system.The method uses logarithm of interaction times as weight when calculating the average rating,to limit the possible exaggeration of interaction times.Besides,it defines tolerance by weighted variance,to recognize and filter out unfair ratings which are too different from the mean of all ratings.Tile simulation demonstrates that the method can filter out unfair ratings effectively,improving the precision of the reputation system greatly.