Breaking the 0.01mm gap between the ideal and reality of big data in radio and television

Founded in 1994, Shenzhen Tongzhou Electronics Co., Ltd. is a high-tech enterprise focusing on providing leading smart home products and services for global users. The company continues to actively expand the layout around the strategic development idea of ​​"smart continent, connect the future", continuously enhance the company's core competitiveness, and further expand multi-functional smart set-top boxes, smart routing, smart gateways, smart speakers, network broadband products, digital front-end Equipment, as well as the research and development and promotion of the new generation of DVB+IP+OTT business integration platform, to consolidate the innovation and market advantages of the business integration platform and smart home business. On this basis, the company actively deploys emerging fields and markets such as smart homes, smart communities, smart buildings, smart parks and the Internet of Things, and expands new growth space through resource integration and model innovation. Comprehensively assist operators to upgrade to "smart" and "Internet of Things", and promote the development of the global digital video and Internet of Things industries. At the same time, the company is also actively evaluating and exploring new applications of artificial intelligence, 8K , 5G and other new technologies in the video field.

 

Breakthrough: The Big Data Project of Tongzhou Radio and Television Landed on Guiguang Network

There is no doubt that the radio and television industry has abundant data resources. According to the content announced in the 2018 National Radio and Television Industry Statistical Bulletin of the State Administration of Radio, Film and Television, the broadcast time of national public broadcasting programs in 2018 was 15.2674 million hours; the broadcasting time of national public TV programs was 19.2503 million hours. In 2018, there were 218 million actual users of cable broadcasting and TV nationwide, and 201 million actual users of national cable digital TV.

 

Since the radio and television are sitting on such a huge amount of data, how to make these data " speak" becomes the top priority! How to realize the reuse of these data, so that it can guide production, realize accurate advertising, closely link user behavior data, so as to realize the development of audiences into users in a true sense, and bring infinite space for innovation and development for the future business of radio and television. The 0.01mm gap between this ideal and reality needs to be broken through.

 

Fortunately, Tongzhou has made it a reality in the Guiguang network big data project. Through the aggregation, mining, analysis and application of these massive data in data collection technology, data processing and data structure analysis, new vitality is stimulated for the production, operation and maintenance management and user services of Guiguang Network.

 

As of November 2019 , the Big Data System of Guiguang Network has collected more than 3.1 million set-top boxes, generating more than 1.5 billion pieces of raw data every day. Through data cleaning, an average of about 40 million complete playback records are generated every day.

 

In 2019 , after realizing the data collection of all two-way set-top boxes, Guiguang Network began to build a data analysis platform. Through the analysis of the collected data, it realized the ranking analysis of live broadcast ratings, timely relocation of on-demand broadcasts, user loss early warning, and network traffic monitoring early warning. , special area page order rate, end-user behavior event analysis and other functions, in addition, based on the analysis results of the big data system, it is connected to the SMP network management, which realizes the statistics and early warning of user-on-demand error codes, and the extraction of user bills, helping operation and maintenance personnel to quickly locate faults. At the same time, the big data platform of Guiguang Network provides data to third-party business systems such as recommendation systems and advertising systems to realize new services such as intelligent recommendation, precise advertising, and personalized portals to improve user experience.

 

Construction: Build a platform architecture for interconnection mechanism

在进行贵广网络大数据平台搭建时，同洲在平台搭建策略上坚持做好整体规划，以支撑大数据应用服务为目标，建立互联互通机制，推动各项数据的共通互联互享；另一方面立足业务现实，分步推进大数据平台的建设与完善。

 

在平台架构上，主要分为数据采集层、应用层和展现层。透过主要架构的搭建，助力贵广网络透过大数据平台采集、存储、挖掘数据资产，通过大数据平台对媒体内容数据、用户服务数据进行全方位分析挖掘，从而提升媒体内容生产的能力，创新网络视听业务产品，升级媒体运营服务、加强媒体传播的监测监管。

 

其中在大数据处理流程上，从数据采集、数据预处理、数据存储、数据处理、数据分析挖掘、统一数据服务到数据应用等环节的处理实现闭环。即将数据应用产生的数据作为新数据源采集进入大数据处理平台，形成完整的应用机制。

 

At present, the first phase of Guiguang Network Big Data Platform has a total of 35 servers, which can store 104TB of data in total. According to the current daily new data estimation, the offline original data can be stored for a maximum of 52 months, and currently it can also support 33 months of storage. , the offline data report display strategy is the last 3 months; the online original storage data can be stored for a maximum of 23 months, and can currently be stored for 15 months, and the online data report display strategy is the last 1 week. The report data displays the data in the recent 5 minutes in real time.

 

Application: Provide full-process service support for your network business

In the construction of the Guiguang network big data platform project, Tongzhou mainly focuses on production and operation scenarios in terms of data application services: such as radio and television big data sharing services, user classification, personalized recommendation, and business analysis. User service scenarios: such as viewing behavior analysis, user activity analysis, customer service, etc.

 

User trend analysis function: collect the daily activity of all interactive users on the entire network through offline data, analyze recent new users and online user data based on regional information, and present it in reports for operational analysis.

 

User churn warning function: In the case of information collection based on big data, use user playback behavior data, payment behavior data, use machine learning methods, infer user data that may be lost after a period of time, and provide data support for system operation and user retention ,Improve work efficiency.

 

Live channel viewing data analysis and reporting function: collect live data information of all interactive users on the entire network through offline data, calculate the live channel ratings and other data, and output reports for presentation.

 

On-demand, time-shift and review data analysis and user playlist report functions: collect on-demand data information of all interactive users on the entire network through offline data, calculate on-demand media resources, column on-demand ranking, user time-shift playback and user playback playback Analysis, and output detailed reports such as user playlists based on regional information.

 

Multicast switching list function: For IP platform users, it collects the multicast switching status of users during the process of watching live broadcasts, and outputs reports to analyze the network status information in the end user area.

 

Online data analysis and report function: According to the user behavior reported by the terminal probe, the user behavior events are classified and collected, analyzed and cleaned, and the user behavior report function is output.

 

Report on user order rate and on-demand conversion rate of users in the special area: According to the user's on-demand behavior and order information data in a specific area, analyze and calculate, and output the report function of the user's on-demand conversion rate and the order rate of users in the special area.

 

There are also traffic data reporting functions, data collection and data output functions, and so on.

 

Future: Gradually plan and build a new pattern to empower Guiguang Network

The application of the Tongzhou big data project in the Guiguang network will continue to promote the intelligent development of the Guiguang network, comprehensively enhance the commercial value of the Guiguang network, and provide support for the upgrading of the Guiguang network industry.

 

In the follow-up planning and construction, Tongzhou will continue to provide more functional applications that can be widely used in the collection and production of radio and television content, program recommendation and management, user tracking and service, and public opinion monitoring and management, including:

 

Data collection visual editing function: optimize and transform the current data collection system and report output function, make big data into visual editing, data results can be customized, selected all fields, customized report style, customized export As a result, the query conditions are customized.

 

User portrait function: Based on user behavior analysis based on data collection, each specific information of the user is abstracted into tags, and these tags are used to concretize the user image, establish a user portrait function, and provide differentiated services for specific user groups.

 

Viewing report analysis: support big data analysis of advertising operation, accurate advertising; support intelligent recommendation, improve user ARPU value; cluster analysis, trajectory analysis to retain customers, big data analysis can understand each customer's daily favorite watching through cluster analysis The program type and channel type, as well as the possible purchase degree of the latest blockbuster, allow the account manager to pick out the top 20% of users who are most likely to buy the product, and take the method of door-to-door visit or telephone promotion to quickly and effectively conclude transactions;

 

Application in operation and maintenance: effectively handle faults - user terminals upload their status data, management and control information and other data through a two-way network. On the basis of management operations such as writing, log reporting, upgrade management, and fault diagnosis, it realizes the visual management of user terminals, the display of terminal operating status, the control of terminals, and the remote diagnosis of faults. The terminal management system can provide information on the current system network and business operation status in various granularities with customized reports, provide system managers with an understanding of the current network operation status, provide a basis for high-level system decision-making, and improve radio and television operators. Service quality and management efficiency; O&M trend prediction—In the operation of the management system, enough data will be accumulated to form the operation status curve of the entire network, and the management system will calculate the network operation trend data according to certain algorithms. When the system detects that the current network operating state deviates from the trend curve, the system sends out an early warning message, which is convenient for operation and maintenance personnel to detect and adjust the network in time, and solve the problem in the state that has not yet occurred in a large area, etc.

 

The practice of Tongzhou Guiguang Network Big Data Project is to exert the value of radio and television big data, so as to truly form the data assets of radio and television enterprises and create the due value of data.