针对网络推断(NBI)算法的二部图实现算法忽略二部图权重而导致实际评分值高的项目没有得到优先推荐这一问题,提出加权网络推断(WNBI)算法的加权二部图实现算法.该算法以项目的评分作为二部图中用户与项目的边权,按照用户-项目间边权占该节点权重和的比例分配资源,从而实现评分值高的项目得到优先推荐.通过在数据集MovieLens上的实验表明,相比NBI算法,WNBI算法命中高评分值项目数目增多,同时在推荐列表长度小于20的情况下,命中项目的数量和命中高评分项目数量均有明显增加.%In Network-Based Inference (NBI) algorithm, the weight of edge between user and item is ignored; therefore, the items with high rating have not got the priority to be recommended. In order to solve the problem, a Weigted Network-Based Inference (WNBI) algorithm was proposed. The edge between user and item was weighted with item's rating by proposed algorithm, the resources were allocated according to the ratio of the edge's weight to total edges' weight of the node, so that high rating items could be recommended with priority. The experimental results on data set MovieLens demonstrate that the number of hit high rating items by WNBI increases obviously in contrast with NBI, especially when the length of recommendation list is shorter than 20, the numbers of hit items and hit high rating items both increase.
展开▼
