摘要
当前存在大量基于位置推荐的移动社交应用。提出一种基于CF算法的混合相似度和信任传播位置推荐系统的方案。方案中主要分三个考量要素,首先将用户偏好分为用户静态偏好和用户动态偏好,对于用户静态偏好主要是基于位置的种类信息和历史评价,而用户动态偏好主要是基于地理信息和二位云模型,最后用户的社会关系是基于信任传播的信息。该方法优势是不仅考虑用户偏好的多样性,而且通过信任传播可以有效缓解数据稀疏性问题。并应用Hadoop以提高计算平台的数据处理能力。该方案对比现有方法,基于CF算法的位置推荐预测用户对新位置的喜好更加准确且高效。
Nowadays there are a large number of mobile social applications based on location recommendation. This paper proposes a method to provide a hybrid similarity and trust propagation location recommendation system based on CF algorithm. The scheme consists of three elements,and the user preference is divided into user static preferences and user dynamic preferences. Static user preference is mainly type information and historical evaluation based on the location,and the dynamic user preference is based on geographic information and two-dimensional cloud model. The users＇ social relationship is based on the information that is propagated by trust. The advantage of this method is that not only the diversity of user preferences is considered,but also the data sparseness can be effectively alleviated by trust propagation.And apply Hadoop to improve the data processing ability of computing platform. Compared with the existing methods,the proposed location recommendation based on CF algorithm is more accurate and efficient to predict the users＇ preferences for new locations.
出处
《计算机应用与软件》
2017年第9期97-102,137,共7页
Computer Applications and Software
