基于协同过滤的多节点信息资源分配推荐算法  

Multi-Node Information Resource Allocation Recommendation Algorithm Based on Collaborative Filtering

作　　者：佘学兵[1] 占清华[1] 邬昌兴 SHE Xue-bing;ZHAN Qing-hua;WU Chang-xing(Department of Information Engineering,Jiangxi University of Technology,Nanchang Jiangxi330098,China;School of Software,East China Jiaotong University,Nanchang Jiangxi330013,China)

机构地区：[1]江西科技学院信息工程学院,江西南昌330098 [2]华东交通大学软件学院,江西南昌330013

出　　处：《计算机仿真》2021年第6期419-423,共5页Computer Simulation

基　　金：2020年江西省教育科学“十三五”规划课题项目(20YB213)。

摘　　要：针对目前算法在多节点信息资源分配推荐时,未对多节点信息资源进行相似性计算,导致多节点信息资源分配时间长,信息资源分配正确率和推荐列表覆盖率较低的问题,提出基于协同过滤的多节点信息资源分配推荐算法。采用协同过滤算法,整合处理节点信息资源,构建信息数据评分模型,运用评分模型,查找节点信息数据的最近邻居集进行预测评分,利用相似性计算,完成多节点信息分类。根据二部图网络结构,资源分配分类节点信息,生成推荐列表并对用户进行推荐。实验结果表明,所提算法的信息资源分配正确率较高,能够有效缩短多节点信息资源分配时间,提高推荐列表覆盖率。Presently, the algorithm ignores the similarity calculation of multi-node information resources in multi-node information resource allocation recommendation, resulting in long multi-node information resource allocation time, low information resource allocation accuracy and recommendation list coverage. Therefore, this paper puts forward a multi-node information resource allocation recommendation algorithm based on collaborative filtering. Firstly, according to the collaborative filtering algorithm, the node information resources were integrated. Secondly, the information data scoring model was established. Then, the scoring model was adopted to find the nearest neighbor set of node information data to predict the scoring, and the similarity calculation was used to complete the multi node information classification. Finally, recommendation list and recommend users were generated by the network structure of bipartite graph and resource allocation classifies node information. The results show that the algorithm has high information resource allocation accuracy and recommendation list coverage, and can reduce multi-node information resource allocation time.

关 键 词：协同过滤 最近邻居集 评价预测 信息资源分配 二部图 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]