摘要： BBK 信任模型提出了对实体信任度进行计算的方法，其中推荐信任合成算法不能有效地抵制恶意推荐带来的影响。文章基于推荐路径互相独立、推荐路径中善意推荐路径数量远大于恶意推荐路径数量和恶意推荐信任值合善意推荐信任值相差很大的假设，用相似程度参数Sdegre 对推荐信任值分类，选择其中所占比例最大的一类信任值进行合成，有效地排除占少数的恶意推荐，从而有效抵制恶意推荐带来的影响。

关键词: BBK；主观信任模型；恶意推荐；推荐信任合成

Abstract: BBK presents a method for the valuation of trustworthiness, but the combination of recommendation trust of it can not effectively resist the effect of malicious recommendation. In this paper in term of the assumption that the recommendation paths are independent and the quality of benign recommendation paths is much bigger than malicious ones and the value of benign recommendation is much bigger than malicious ones, it classifies the recommendation values in term of similar degree parameter Sdegree and chooses the bigger group to combine, so it can exclude malicious recommendation which is smaller and can effectively resist the effect of malicious recommendation

Key words: BBK; Subjective trust model; Malicious recommendation; Combination of recommendation trust