I 摘要 “5G已来，未来已至”，在互联网大数据时代，面对腾讯视频、爱奇艺、优 酷视频、咪咕视频等长视频业务，以及今日头条、西瓜视频、火山视频、抖音等 短视频业务的冲击。传统广电业务，中国电信的IPTV业务，中国移动的“魔百和” 业务，以及其他运营商的互联网电视业务，如何应对新兴技术带来的挑战，是一 个亟待回答的问题。现今用户对视频应用类的收视需求可选择性越来越多，在视 频行业，如何精准定位用户收视喜好，更好的为用户推送电视节目，把用户留在 家庭客厅，并且使用户愿意为付费节目埋单；电视收视用户如何能够更快捷的从 海量电视节目中寻找自己感兴趣、喜欢的电视节目内容；内容生产与版权提供者 如何让高昂的内容生产与版权购买成本转化成利润收入等，都是S省“魔百和” 业务运营面临的很现实的亟待解决的问题。 为了解决此问题，本课题在大数据背景下，研究S省“魔百和”业务运营中， 通过对收视用户的精准画像，结合媒体资源库，对用户进行个性化精准推送用户 偏爱的电视节目，实现智能有效用的运营模式。 课题具体研究过程，通过抽取S省“魔百和”业务在网用户219898个样本量 进行智能运营与传统运营方式实验对比研究分析论证。智能运营主要是通过“魔 百和”业务的集成播控系统，采集“魔百和”电视用户的收视行为数据，根据用 户不同的收视行为数据，构建用户画像，形成用户特有的属性标签库。后台CMS 内容管理系统通过编辑不同的内容，将内容节目组织成为有意义的内容服务，并 映射到用户对应的收视行为标签下，形成数字版权内容库。在标签体系和数字版 权内容库的基础之上，以用户为中心，根据用户的收视习惯，将用户的收视行为 标签类型，进行建模分群，再对特定群体用户，通过后台智能推荐系统，向用户 针对性的推送用户偏好的电视节目内容。 最后，经过实验研究对比论证，基于用户画像的“精耕细作”式智能推送内 容，相对比“粗放圈地”式的传统推荐内容，能够更好的提高用户的收视活跃率、 提高用户的付费转化率，提升“魔百和”电视业务的增值收入。鉴于课题研究结 论，建议S省“魔百和”业务继续扩大智能运营试点范围，并完善用户画像、内 容标签库的标签体系以及运营数据的实时性。 关键词：“魔百和”，用户画像，标签体系，智能运营 ABSTRACT II ABSTRACT “5G is in place and the future is about to coming”, in the era of Internet Big Data, with the booming of long video service provided by Tencent Video, Iqiyi, Youku, Migu, and short video service provided by Bytedance, ixigua, huoshan, Tik Tok, the traditional broadcasting and radio service providers is under serious impact. How IPTV service provided by Chinartelecom, Mobaihe service provided by Chinamobile and other operators’ IPTV services cope with challenges brought by the newly emerging technologies, is a question that needs to be solved urgently. In recent years, there are more and more options that are made available to the video audience to meet their demand. Mobaihe’s business operation team needs to figure out the answers to the underlying questions such as how to target the customers’ preference to push more interesting content respectively, to make the customers stay in living room and willing to pay for the content; How to enable the customers to find out the TV programme that is interesting to themselve; How to convert the skyrocketing content creation cost and copyright costs into profit revenue. In hope of answering the critical questions, this paper will analyze the intelligent and effective measures applied in the operation of S Province Mobaihe Service. Through precise user portrait, and combine with affluent media library, the congenial content can be targeted to the customers accurately. This paper analyses a samples of 219,898 Mobaihe users in S Province, compares with different operation efficiency under the traditional operation mode and Intelligent Operation Mode. Intelligent Operation Mode refers to a novel method that captures the user behaviour through Mobaihe Integrated System, and create the user portrait and tags by analyzing the user behaviour. The backstage CMS organizes content by editing various content and maps the content to the users that are tagged with related interests, thereby the digital database is created. On the foundation of tag system and digital copyright database, the users view behaviour will be defined as tag categories and user groups will be divided. The backstage intelligent recommendation system will push nichetargeting content to the users respectively. Lastly, through the experimental research and argument, the author finds out that the targeted push mode is significantly more effective than the traditional sketchy way ABSTRACT III of content-pushing, and increases user activity and paid-user conversion rate remarkably. The result of this research shows that it is necessary for S Province Mobaihe Operation Team to broaden the scope of intelligent operation, improve user portrait, content tags and other tagging system, and eventually improve the effectiveness of operational data. Keywords: “Mobaihe”, User Portrait, Tagging System, Intelligent Operation 目 录 IV 目 录 第一章 绪论 .................................................................................................................... 1 1.1 课题研究背景及意义 ....................................................................................... 1 1.1.1 课题研究背景 ........................................................................................ 1 1.1.2 课题研究意义 ........................................................................................ 2 1.2 课题研究内容与创新 ....................................................................................... 4 1.3 课题主要研究思路与研究方法 ....................................................................... 5 1.3.1 研究思路 ................................................................................................ 5 1.3.2 研究方法 ................................................................................................ 6 1.4 本章小结 ........................................................................................................... 7 第二章 智能运营技术理论与技术研究 ........................................................................ 8 2.1 用户画像技术研究 ........................................................................................... 8 2.1.1 用户画像的定义 .................................................................................... 8 2.1.2 用户标签体系 ........................................................................................ 8 2.1.3 用户画像的构建 .................................................................................... 9 2.1.4 用户画像技术的应用 .......................................................................... 11 2.2 智能运营推送技术研究................................................................................... 12 2.2.1 智能运营推送技术 .............................................................................. 12 2.2.2智能推送技术应用现状 ....................................................................... 13 2.3 智能运营技术研究现状................................................................................... 14 2.3.1 用户画像技术研究现状 ...................................................................... 14 2.3.2 内容推送技术研究现状 ...................................................................... 15 2.4 本章小结 ......................................................................................................... 16 第三章 S省魔百和业务发展分析 ............................................................................. 17 3.1 魔百和业务发展现状 ..................................................................................... 17 3.1.1 魔百和行业发展现状 .......................................................................... 17 3.1.2 魔百和内容引入现状 .......................................................................... 17 3.1.3 魔百和运营发展需求 .......................................................................... 18 3.2 魔百和业务发展问题 ..................................................................................... 18 3.2.1 魔百和业务末端收视用户需求痛点 .................................................. 18 3.2.2 魔百和产业链上内容提供者的诉求 .......................................