A Presenter Discovery Method Based on Analysis of Reputation Record

In order to satisfy the requirement of service presenters’ credibility for recommendation system, this paper puts forward a method based on the analysis of the presenters’ reputation record to find the credible presenters. First, we calculate the preference similarity between different users with their preference vectors and get the presenter initial set. Then we calculate the domain correlation, recommendation response rate and recommendation satisfaction rate to filter the presenter initial set and get presenter candidate set. According to the recommendation history we calculate the presenters’ current reputation with introducing the penalty factor. Finally, we filter the presenters’ reputation record with localized changes and tendentious changes and calculate the skewness coefficient and kurtosis coefficient of the reputation records. Then we get the presenters’ excellent reputation values combined with the expectation and variance to choose the credible presenters who have higher reputation values. The experimental results show that it can improve the accuracy of presenter reputation calculation and the effectiveness of service recommendation.