Social media applications have become the mainstream of Web application. User-oriented and content generated by users are pivotal characteristics of social media sites. Data sharing and recommendation approaches play an important role in dealing with the problem of information overload in social media environment. In this paper, we analyze the flaws of current group-based information sharing mechanism and the common problem of traditional recommender approaches, and then we propose a novel approach of group automatic generating for social media sharing and recommendation. Intuitively, the essential idea of our approach is that we switch user's preference from the media objects to the interest elements which media objects imply. Then we gather the users who have common preference, namely users have the same interestingness in a set of interest elements, together as Common Preference Group (CPG). We also propose a new social media data sharing and recommendation system architecture based on CPG and design a CPG automatic mining algorithm. By compare our CPG mining algorithm with other algorithm which has similar functionality, it is shown that our algorithm could be applicable to real social media application with massive users.%社会媒体应用已成为Web应用的主流,以用户为中心并且海量媒体数据由用户自生成是社会媒体Web应用的重要特征.应对目前社会媒体环境中信息过载的问题,信息的共享和推荐机制发挥着重要的作用.文中分析了目前主流社会媒体网站基于用户自建组的信息共享机制所存在的问题以及传统推荐技术在效率上的问题,提出了一种新的基于用户偏好自动分类的社会媒体数据共享和推荐方法.直观上讲,该方法的本质是把用户对具体媒体对象的偏好转化成用户对媒体对象所蕴含兴趣元素的偏好,然后把具有相同偏好的用户,即对若干兴趣元素上的兴趣度都相同,自动聚合成为一个“共同偏好组(CPG)”.文中提出了基于CPG的社会媒体信息共享和推荐的架构,设计实现了CPG的自动生成算法,通过随机生成模拟数据集实验详细分析了算法性能的影响因素,并与现有类似功能算法进行了效率对比,实验结果表明算法可适用于具有海量用户的社会媒体应用.
展开▼
