查询结果:   顾军华,高星,王守彬,武君艳,张素琪.基于大数据的IPTV视频评估模型[J].计算机应用与软件,2018,35(8):231 - 237.
中文标题
基于大数据的IPTV视频评估模型
发表栏目
多媒体技术应用
摘要点击数
1163
英文标题
IPTV VIDEO EVALUATION MODEL BASED ON BIG DATA
作 者
顾军华 高星 王守彬 武君艳 张素琪 Gu Junhua Gao Xing Wang Shoubin Wu Junyan Zhang Suqi
作者单位
河北工业大学计算机科学与软件学院 天津 300401 河北省大数据计算重点实验室 天津 300401 天津商业大学信息工程学院 天津 300134   
英文单位
School of Computer Science and Software, Hebei University of Technology, Tianjin 300401, China Hebei Province Key Laboratory of Big Data Computing, Tianjin 300401, China School of Information Engineering, Tianjin University of Commerce, Tianjin 300134, China   
关键词
大数据 IPTV视频评估模型 隐式评分 Spark BP神经网络
Keywords
Big data IPTV video evaluation model Implicit rating Spark BP neural network
基金项目
河北省科技计划项目(17210305D);天津市科技计划项目(16ZXHLSF0023);天津市自然科学基金项目(15JCQNJC00600)
作者资料
顾军华,教授,主研领域:智能信息处理,数据挖掘。高星,硕士生。王守彬,硕士生。武君艳,硕士生。张素琪,讲师。 。
文章摘要
随着网络信息技术的发展以及“三网融合”的推进,交互式网络电视IPTV成为越来越多用户的选择,成为新媒体中的一支主力军,但快速发展的同时也面临着巨大的挑战。如何有效评估供应商提供的大量视频,选择符合用户需求的视频成为IPTV发展的关键问题。提出利用新媒体和传统媒体的视频大数据和IPTV历史收视大数据,在Spark平台上使用BP神经网络建立视频评估模型。基于新媒体和传统媒体从视频收视度、视频影响度和视频内容三个方面完善视频评估体系;基于IPTV历史收视大数据,建立反映IPTV受众群体喜好的视频隐式评分策略,使用BP神经网络构建视频评估模型;针对大数据的海量性,在Spark并行化平台上建立视频评估模型,实现数据的并行训练,完成模型的建立。实验结果证明,新的视频评估模型能从IPTV受众群体的角度有效评估视频,在Spark平台上进行评估模型的训练,能够有效提高大数据量的评估模型训练速度。
Abstract
With the development of network information technology and the progress of the “triple play”, IPTV has become a choice for more and more users and has been a major force in new media. But it also faces great challenges at the same time. How to effectively evaluate the large number of video resources provided by suppliers and select videos that meet the needs of users has become a key issue in the development of IPTV. The BP neural network was used in this paper to establish the video evaluation model on the Spark platform by using the large video data of new media, traditional media and IPTV history. First, based on the new media and traditional media, the video evaluation system was perfected from three aspects: video viewing degree, video influence degree and video content. Then, based on the IPTV history, a video implicit scoring strategy was set up to reflect the preferences of the IPTV audience, and the video evaluation model was built using the BP neural network; finally, aiming at the mass of big data, a video evaluation model was established on the Spark parallel platform to realize data parallel training and complete the establishment of the model. Experimental results show that the new video evaluation model can effectively evaluate video from the perspective of IPTV users. The training of the model on the Spark platform can effectively improve the training speed.
下载PDF全文