北京邮电大学学报

  • EI核心期刊

北京邮电大学学报 ›› 2012, Vol. 35 ›› Issue (4): 24-27.doi: 10.13190/jbupt.201204.24.zhangch

• 论文 • 上一篇    下一篇

基于社会化媒体节点属性的信息预测

张闯,姜杨,吴铭,肖文君,李泰   

  1. 北京邮电大学 模式识别与智能系统实验室
  • 收稿日期:2011-09-21 修回日期:2012-02-21 出版日期:2012-08-28 发布日期:2012-07-08
  • 通讯作者: 姜杨 E-mail:13701167275@139.com
  • 作者简介:张闯(1975-),男,副教授,E-mail:zhangchuang@bupt.edu.cn
  • 基金资助:

    新一代宽带无线移动通信网科技重大专项(2011ZX03002-005-01);教育部青年创新基金项目(2012RC0129);111工程学科创新引智计划项目(B08004)

Information Predictions Based on Node Attributes of Social Media

ZHANG Chuang, JIANG Yang, WU Ming, XIAO Wen-jun, LI Tai   

  1. Pattern Recognition and Intelligent Systems Laboratory, Beijing University of Posts and Telecommunications
  • Received:2011-09-21 Revised:2012-02-21 Online:2012-08-28 Published:2012-07-08
  • Contact: Yang Jiang E-mail:13701167275@139.com

摘要:

针对多数研究仅将社会化媒体作为数据来源的现状,深入分析社会化媒体特点,重点将节点属性分为静态和动态进行研究,提出基于预测目标的节点影响力的概念. 在此基础上提出了一种基于节点属性进行信息预测的属性、节点数、倾向(ANV)模型. 实验采用后向传播(BP)神经网络预测方法,通过新浪微博数据预测电影票房. 仿真表明,带有节点属性的方法比没有节点属性的方法拟合和预测更为准确.

关键词: 社会化媒体, 节点属性, 预测模型, BP神经网络模型, 票房预测

Abstract:

The features of social media, such as node influence are important in information prediction. A new model with attribute, number and views of nodes (ANV) is proposed for event prediction. Experiment is designed to predict the performance of new released movies using back propagation (BP) neural network algorithm with the data of Sina Microblog. Compared to other models, when considering the node attribution, the proposed has higher prediction accuracy.

Key words: social media, node attribute, prediction model, BP neural network, box office earnings forecast

中图分类号: