服务推荐者声誉记录分析方法研究

Research of Analysis Method of Presenters' Reputation Record

为了满足推荐系统对服务推荐者可信性的要求,提出一种基于声誉记录分析的可信推荐者发现方法,通过用户偏好向量计算不同用户间的偏好相似度,计算领域相关度、推荐响应率、满意率并对推荐者进行过滤,通过引入惩罚因子计算推荐者当前声誉,对推荐者声誉记录进行定域改变和倾向改变过滤,计算出声誉记录的偏度系数和峰度系数,结合期望与方差确定推荐信任源。实验结果表明,该方法可以提高推荐者声誉计算准确性,寻找到更为可信的推荐者从而提升服务推荐的有效性。

In order to satisfy the requirement of service presenters’ credibility from recommendation system, this paper put forward a method based on the analysis of reputation record to find the most credible presenters. The preference similarity between different users is calculated with users’ preference vectors. The relevance of the fields, recommendation response rate and recommendation satisfaction rate are then calculated to filter the presenters. The presenters’ current reputation is calculated by introducing the penalty factor. We filter the presenters’ reputation record with localized changes and tendentious changes and calculate the skewness coefficient and kurtosis coefficient of the reputation records. We get the presenters’ excellent reputation values combined with the expectation and variance to choose the credible presenters. The experimental results show that this method can improve the accuracy of presenters’ reputation calculation and the effectiveness of service recommendation.