微博网络中用户关注行为预测  

Predicting the User Follow Behavior in Microblogging Network

作　　者：曹云忠[1,2] 邵培基[1] 李良强[1] 

机构地区：[1]电子科技大学经济与管理学院,四川成都611731 [2]四川农业大学建筑与城乡规划学院,四川成都611830

出　　处：《系统工程》2015年第7期146-152,共7页Systems Engineering

基　　金：国家自然科学基金资助项目(71101018)

摘　　要：在微博网络中挖掘博主的潜在关注用户,建立相应的关注预测指标和模型,对于增加博主活跃粉丝数、增强信息传播具有重要意义。利用微博粉丝人际关系网络,在众多与博主交互的用户中,预测潜在的活跃粉丝。在融合微博主影响力、用户活跃度等因素的基础上,提出了以兴趣相似度为核心的用户关注预测指标。利用模糊近似支持向量机作为预测工具,对模糊隶属度算法进行了优化,推导了基于矩阵的模糊近似支持向量机预测模型。利用KDD Cup 2012腾讯微博公开数据进行了实证研究。实验结果表明,本文提出的指标能很好地用于微博用户关注行为的预测,通过与其他支持向量机模型及模糊化算法相比,本文提出的优化方法具有较高的预测精度,且预测结果稳定,在微博用户关注行为预测中具有较强的实用性。Mining potential users in microblogging network,establishing the appropriate follow predictors and models for active bloggers are very important for increasing the number of fans,and enhancing dissemination of information.the Microblogging network user is regarded as the research object.A model on Microblogging fans is employed to predict potential active fans in many interactions with blogger user.The main influence factors such as user activity,and other factors on the basis of proposed user interest similarity are viewed as the core follow predictor.Fuzzy proximal support vector machine is used as a forecasting tool which fuzzy membership algorithm is optimized for ease of experimental derived matrix representation of fuzzy proximal support vector machine model.Concern the predictor and fuzzy the approximate support vector machine predictive validity of the use of publicly available data of the KDD Cup 2012 Tencent microblogging empirical study to verify the user.Experimental results show that,compared with other support vector machine model and fuzzy algorithm,predictor and prediction methods proposed in this paper has higher prediction accuracy and stability of the predicted results.

关 键 词：微博 用户关注 预测 模糊近似支持向量机 

分 类 号：C931[经济管理—管理学]