The news multimedia integration platform is not only a content aggregator, but also the creator of information distribution channels and user interaction channels. Based on the RealSightCI customer intelligence analysis platform and SaCaDeepCogni knowledge service platform, CCTV News highly integrates various new media products with each other, covering all types of user scenarios in all aspects, enabling real-time interaction with users based on news features, and becoming users through the Internet , Especially the main entry point for news on the mobile Internet.
CCTV News Client is the official client of CCTV News Center, which can update the content in real time and release the latest news as soon as possible. Client users can not only learn the latest information through CCTV News, but also follow up on the program in real time, participate in news discussions and interactions.
face the challenge
CCTV News Client has now formed a new media platform with client, Weibo and WeChat as the basic layout. Clients urgently need to achieve a deep integration and integrated platform and network operation of the new media collection, aggregation, and distribution platform, so that the CCTV news client has become a first-class mainstream media with powerful public opinion guidance and speaking power in the new media field.
Following the overall technical strategy of key media fusion projects, through loose coupling and unified technical standards, docking with the general platform, sharing user databases and video databases, and reusing unified underlying platform resources, reducing the overall development and construction costs. Build a content recommendation system based on a variety of algorithm recommendations, based on user behavior, social relationships, content, tags, popularity, geographic location, etc .; support intelligent recommendations for programs and advertisements; have fast interaction capabilities with websites, mobile apps, and OTT.
Comprehensive intelligent operation plan
By recording and analyzing user behavior and news content, the obtained user model is more relevant to the user's true ability and interest model, and it can better reflect the user's reading style. Users and news can be calculated with more accurate recommendation algorithms to meet users' personalized reading needs. With the message push module, the news delivery is more efficient, timely and accurate, and it strongly supports the intelligent operation business requirements of the new media integration platform.
User Analysis System
According to the collected data, all aspects of the user, user behavior, and content viewed by the user are analyzed. Perform multi-dimensional portraits for users, construct user interest graphs, and extract user labels. Support the usage habits analysis of typical user groups, and help optimize the news navigation architecture and news release strategy in order to optimize the user experience in a targeted manner.
Intelligent recommendation system
Provide multi-dimensional annotations for news content. It is divided according to 40 channels and 400 topics, and provides functions such as news channels, topic concepts, and keyword extraction. The accuracy of channel classification is more than 90%. It provides a content foundation for the entire intelligent operation. Based on the news tag data of the intelligent learning system and the user tag data of the user analysis system, a variety of built-in recommendation algorithms are used to construct an intelligent recommendation model based on the response business scenario. Personalize news recommendations for users based on their tags and interests, and find user groups that match them based on their content. The recommended clickthrough rate is 3.4 times higher than the popular ranking, and is 800% of random recommended news.
Smart Push Service
The recommendation results (programs, advertisements, personalized information, etc.) of the intelligent recommendation system are pushed to the target users according to business rules through the push engine.
The construction of the CCTV news integration platform has greatly accelerated the automatic processing of news content. Through automatic classification, related news, hot topics and tracking functions, it helps editors to understand the current news trends and hot content faster and more accurately. Shorten the cycle of news production and release, and grasp news trends.
In addition, by recording and analyzing user behavior and news content, the user model obtained through analysis is more relevant to the user's true ability and interest model, and can better reflect the user's reading style. Users and news can be calculated with more accurate recommendation algorithms to meet users' personalized reading needs. With the message push module, the delivery of news is more efficient, timely and accurate. Strongly supports the intelligent operation business requirements of the new media integration platform.
Editor-in-chief: Liu Yangge