基于收视行为的互联网电视节目流行度预测模型  

Program Popularity Prediction Model of Internet TV Based on Viewing Behavior

作　　者：朱琛刚[1] 程光[2] 

机构地区：[1]东南大学计算机科学与工程学院,南京211189 [2]教育部计算机网络和信息集成重点实验室(东南大学),南京211189

出　　处：《电子与信息学报》2017年第10期2504-2512,共9页Journal of Electronics & Information Technology

基　　金：国家863计划项目(2015AA015603);江苏省未来网络创新研究院未来网络前瞻性研究项目(BY2013095-5-03);江苏省"六大人才高峰"高层次人才项目(2011-DZ024)~~

摘　　要：准确预测节目流行度是互联网电视节目系统设计与优化所要解决的关键问题之一。针对现有预测方法存在模型训练时间长、样本数量多、且对突发热点节目流行度预测效果差等问题,该文测量了某互联网电视平台280万用户的60亿条收视行为数据,采用行为动力学分类方法将节目流行度演化过程分为内源临界、内源亚临界、外源临界和外源亚临界4种类型,运用双种群粒子优化的最小二乘支持向量机对每种类型分别构建了一种互联网电视节目流行度预测模型BD3P,并将BD3P模型应用于实际数据测验。实验结果表明,与现有其他方法相比,BD3P模型预测精度可提升17%以上,并能有效缩短预测周期。Predicting program popularity is a key issue for design and optimization of Internet TV system. Existing prediction methods usually need large quantity of samples and long training time, while the prediction accuracy is poor for the burst hot programs. This paper introduces an Internet TV Program Popularity Prediction model based on viewing Behavioral Dynamics features （BD3P）. 6 billion view behavior records from 2.8 million subscribers of a certain Internet TV platform are measured, and the evolution process of program popularity is divided into 4 types based on behavioral dynamics features, which is endogenous, internal subcritical, exogenous and exogenous subcritical. The prediction models of Internet TV program popularity are constructed for each type using Least Squares Support Vector Machines （LSSVM） with double population Particle Swarm Optimization （PSO）, and these models are applied to the actual data test. The experimental results show that, compared to the existing prediction model, the prediction accuracy can be increased by more than 17%, and the forecast period can be effectively shortened.

关 键 词：互联网电视 流行度预测 行为动力学 最小二乘支持向量机 双种群粒子群优化 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]