基于标签对家庭IPTV业务个性化推送机制的用户体验优化研究  

User Experience Optimization Research on Personalized Push Approach of IPTV Service Based on Label of Contents

作　　者：朱映波[1] 刁建伟[1] 康波[2] 刘胜强[2] 

机构地区：[1]中国电信股份有限公司数字音乐运营中心,广州510810 [2]中国电信股份有限公司广东研究院,广州510630

出　　处：《电信科学》2014年第7期113-120,共8页Telecommunications Science

摘　　要：爱音乐卡拉OK业务作为一种IPTV业务,现阶段基于遥控器的用户操作成本高于PC和智能手机,需要引入内容的个性化主动推送机制以提升用户体验。但卡拉OK作为一种家庭业务,具有服务器端和客户端对客户主体认定不对称的特点,依据普通的推荐算法极易导致不同家庭成员之间的错位推荐。同时,传统推荐算法本身也存在马太效应、冷启动等缺陷,这些都会导致用户体验下降。针对这些用户体验问题,从业务的用户使用场景出发,对推荐算法进行选择与改良,并结合内容标签提升推荐的针对性,以克服算法缺陷,提升用户体验。研究的首要着眼点在于用户体验效果,而非算法的复杂性。iMusci Karaok is a service on IPTV developed and operated by China Telecom. It is needed to introduce active personalized push approach of content in order to improve user experience, due to higher cost of remote controller than PC and mobile phone. However, Karaok is a household service, and is ununiform to identify users on server side and terminal side, which will bring inaccurate result when normal recommendation algorithm is adopted. Meanwhile, traditional recommendation algorithm has defects of Matthew effect and cold reboot, which will decrease user experience. To tackle such problems of user experience, the recommendation algorithm was selected and optimized by combination of content label so as to increase recommendation objectively. This research is aiming at enhancing user experience effect rather than algorithm complexity.

关 键 词：关联规则 标签 个性化推送 层次分析法 权重 奇异值分解 独立性检验 马太效应 

分 类 号：TN94[电子电信—信号与信息处理]