Abstract: In the personalized business process customization environment,recommendation technology is introduced to improve the user's customization efficiency,but the existing recommendation technology has many problems,such as the new users,the lack of diversity of recommendation results and low recommendation quality.To give full play to the advantages of recommended technologies and solve the above problems,a service recommendation scheme suitable for business process customization environment was proposed.In this scheme,the clustering algorithm was used to solve new user problems,then the collaborative filtering was optimized to improve the diversity of recommendation results.The content filtering was built to screen out “pseudo-neighbors” for improving the recommendation quality,and the candidate services were recommended to the user to improve customization efficiency.The experimental research proved that the scheme proposed could solve the problem of new users,improve the diversity of recommendation results and recommendation quality,and enhance the customization efficiency of the users.

Key words: cloud service platform, business process customization, user clustering, collaborative filtering, content-based filtering, hybrid recommendation technology

摘要： 在个性化业务流程定制环境中,为提高用户的定制效率引入推荐技术,但现有推荐技术存在诸多问题,如新用户、推荐结果缺乏多样性与推荐质量偏低等。为充分发挥推荐技术的优势并解决上述问题,提出一种适用于业务流程定制环境的服务推荐方案,该方案首先利用聚类算法来解决新用户问题,再通过优化协同过滤来提高推荐结果的多样性,然后构建内容过滤筛除“伪邻居”来提高推荐质量,最后将获得的候选服务推荐给用户来提高定制效率。通过实验研究证实,所提方案既能解决新用户问题,又能改善推荐结果的多样性与推荐质量,还能提高用户的定制效率。

关键词: 云服务平台, 业务流程定制, 用户聚类, 协同过滤, 基于内容的过滤, 混合推荐技术