4Paradigm Released Intelligent Recommendation System First Click Accelerating the Intelligent Upgrade of Media Business.


On November 6th, the 2018 Artificial Intelligence and New Media Forum organized by 4Paradigm was held in Beijing. At the meeting, the 4Paradigm released AI products empowering media - the First Click  recommendation intelligent system (hereinafter referred to as First Click ). The product incorporates advanced artificial intelligence algorithms to help media quickly and accurately match the user's reading needs, improve media operation efficiency and user retention time, and realize user and traffic incremental monetization, and then accelerates the intelligenization of media.

AI recommendations become the foundation for media An insight into real user needs

With the beginning of the mobile Internet era, competition in the media industry has intensified, and problems such as the continuous homogenization of reading client content and the fragmentation of user browsing time have become prominent. How to retain users and satisfy users' personalized high-quality content demand becomes the key to user operations. Previously, the industry generally regarded technology as an important productivity measure to improve content and user operational efficiency. The recommendation system based on emerging technologies such as artificial intelligence has become a weapon for media's intelligent transformation. Some workers in this industry pointed out that AI-based algorithms are no longer the core competitiveness of news clients, but a basic capability.

In fact, the intelligent recommendations, where everyone sees different content, is one of the important breakthrough points in the game of user attention deficit and information overload. However, due to the high threshold of artificial intelligence technology and the lack of relevant talents, it is difficult for media to build a recommendation system by in-house technical team. The release of First Click  provides a good platform for them.

First Click  is a one-stop recommendation service visualization platform that integrates content uploading, content management, content distribution, recommendation intervention, and front-end rendering. It supports all platform access such as PC, WAP, and APP, which helps media build recommendation systems from 0 to 1. It significantly improves important business metrics such as user activity, retention, and watch time, and significantly improves the efficiency of media operations while reducing cost. Through the above steps, it realizes artificial transformation. At present, First Clickhas carried out in-depth cooperation with more than 300 media such as Xinhua News Agency, People's Daily, Titanium Media, Tiger Sniff, Yiou, and CSDN. Among them, in the CSDN information flow scenario and the WAP station advertisement recommendation and other projects, First Click achieved a 110% increase in click-through rate, while the traffic and revenue increased by 187% and 49%, respectively.

Through low threshold, advanced algorithms and multi-scenario applications, First Click can enhance media capabilities.

First Click supports real-time and comprehensive analysis of user search habits, scenarios and contexts, enabling accurate content distribution, PV increase of 30-200%, as well as significantly enhanced user stickiness and click depth. Millisecond data updates provide real-time recommendations for high-quality content. At the same time, Xianjian is also equipped with search engine optimization to achieve intelligent optimization of SEO.

Currently, First Click provides three kinds of services: personalized recommendations, related recommendations, and popular recommendations. Based on the intent recognition ability of machine learning, First Click can recommend content that best suits the current scenario for each user at different times in different places, providing a truly personalized recommendation service that enables everyone to see different, tailored contents. Related recommendations are based on current content semantic analysis, recommending information similar to the current content, in order to improve the length of stay and the number of materials read per person. While popular recommendations are based on the reading statistics of the whole channel content to make global, sub-channel or sub-regional rankings.