欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
以大众分类（folksonomy）为基础的社会化网络中，通常用标签对资源特征和用户喜好进行描述，本文将标签语义引人传统的推荐系统，从用户的标注行为入手，提出一种基于超图投影的推荐方法，该方法利用投影图中节点的连边权重进行节点相似性度量，使待推荐对象在投影图上随机游走，根据待推荐对象在节点上停留概率获得推荐，实验结果表明标签的引入提高了推荐质量，算法在精确性和多样性上均有很好的改进。
英文摘要：
In folksonomy-based social networks, tags are important information which can be used to describe features of recourses as well as users＇ topic preferences. The latent semantics of Tags are introduced into traditional recommendation system, then starting from user＇ s annotation an approach for recommendation was proposed based on hypergraph projection. In our approach, the similarity of nodes is measured by edge weights in projected hypergraph, the objects to be recommended random walks on the projected hypergraph, so that the result of recommendation can be obtained according to its＇ probability of staying on the node. The experimental results clearly show that our approach is effective, both diversity and accuracy can be improved by considering the latent semantics of Tags.
同期刊论文项目
同项目期刊论文
期刊信息
位置：