基于数据挖掘的社交网络信息推荐与预测方法研究  

Research on Social Network Information Recommendation and Prediction Method Based on Data Mining

作　　者：陈婉[1] 朱世杰[2] CHEN Wan;ZHU Shi-jie(Information Network Center, Zhengzhou Tourism College, Zhengzhou 450009, China;Students＇Affairs Division, Zhengzhou Tourism College, Zhengzhou 450009, China)

机构地区：[1]郑州旅游职业学院信息网络中心,河南郑州450009 [2]郑州旅游职业学院学生处,河南郑州450009

出　　处：《内蒙古师范大学学报（自然科学汉文版）》2018年第2期127-131,共5页Journal of Inner Mongolia Normal University(Natural Science Edition)

基　　金：河南省基础与前沿技术研究计划项目(142300410463)

摘　　要：针对现有推荐与预测方法中存在的不足,提出了一种基于数据挖掘的社交网络信息推荐与预测方法.该方法在传统预测模型的基础上构建了双向社交网络推荐与预测框架,并在框架内整合了用户类别、行为和内容相似性特征构建广义的拓扑特征集合,通过协同过滤算法对用户的权重特征进行聚类分析,提升模型的预测效果.实验结果表明,该方法有效提升了静态数据环境下的挖掘精度.With the development of two-way social networks such as＂We Chat＂and＂Micro-blog＂,the recommendation and prediction of social information become more and more important.In view of the shortcomings of the existing recommendation and prediction methods,this paper proposes a method of social network information recommendation and prediction based on data mining.Based on the traditional prediction model,this method constructs a framework for recommendation and prediction of bi-directional social networks,and integrates user categories,behaviors and content similarity features into the framework to construct generalized topological feature sets.The method uses collaborative filtering algorithm to cluster the user’s weight characteristics,so as to improve the prediction effect of the model.Experimental results show that the proposed method can effectively improve the mining accuracy in static data environment.

关 键 词：信息推荐 数据挖掘 预测模型 特征提取 特征融合 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]