In the service oriented computing, the evaluation that the user for the service which its uses is different from the difference cf service spots. Even if the identical service, the different user' s evaluation criteria is dissimilar, the selection of recommender is related to not only context but also service spots. In order to enable recommendation more reliable and effective , this paper put forward recommendation algorithm based on service attention spot similarity. In order to solve the problem of bind searching, it generated the user clusters with clustering algorithm. It searched recommenders based on similarity between users in the cluster class. It not only improved the reliability of the recommendation, but also improved the efficiency of the searching. The experiments show that this algorithm has obvious advantages compared with traditional algorithm on the recommendation accuracy and searching efficiency.%针对服务计算环境下用户对其所使用服务的评分,依据其服务关注点的不同而不同,即使是同一个服务,不同用户的评价标准也不一样,推荐者的选取不仅与其所处环境上下文有关,还与推荐者对服务的关注点有关.为了使用户推荐更加可靠、有效,提出基于服务关注点相似度的推荐算法.该算法解决了用户盲目搜索推荐者的问题,使用聚类算法生成用户聚类簇,根据用户间的相似度在聚类簇内进行推荐者的搜索,既提高了推荐的可靠性,又提高了搜索的效率.实验显示,此算法比传统算法在推荐准确性与推荐搜索效率上存在明显优势.
展开▼
