基于用户兴趣和服务满足度的服务搜索和推荐算法

Service search and recommendation algorithm based on user's interest and service satisfaction

随着互联网中服务的不断增加,如何为目标用户自动并精准推荐软件服务正在成为软件服务技术大规模应用和推广过程中必须解决的关键问题。但由于获取用户使用服务的信息比较困难,所以现有的大部分研究方法没有考虑用户的兴趣,从而不能产生个性化的搜索和推荐。而Mashup开发模式不断广泛的应用,所提供的信息恰好为推荐提供了丰富、便捷的信息来源。因此,提出了一种基于用户、Mashup、服务、标签的服务社会化网络的搜索和推荐算法,此算法能同时推荐原子级服务和组合服务,并且既考虑了用户对服务的兴趣剖面,又考虑了服务对标签的满足度剖面。实验从准确率、召回率和平均绝对误差三个指标进行分析表明,算法具有较好的推荐效果。

With the increasing services in the Internet, how to recommend software service automatically and accurately for tar- get users is becoming a key issue of software technology that must be addressed in the process of promoting large-scale applica- tion. But it is difficult to get information about the service used by users, so most of the existing methods do not consider the user＇ s interests, and thus can not produce personalized search and recommendation. With the constant application of Mashup development model, which provide a rich, convenient source of information for recommendation. Therefore, based on user, Mash- up, service and tag,this paper proposed a search and recommendation algorithm for social networking service. This algorithm could not only recommends atomic-level services but also combination services. Both the users＇ interest profile of service and the services＇ satisfaction profile of tag are taken into account. Experiment from three indicators ： the precision rate, recall rate and mean absolute error all show this algorithm has a better recommendation result：