The browser version you are accessing is too low. To provide better experience, it is recommended that you upgrade the browser toEdgeBrowserOr, it is recommended.GoogleBrowser

Smart Profile Empowers TV Screen for Optimized Operation

Release Date:2021-01-22  Author:By He Yanfeng, Liu Zhijun  Click:

According to the Ministry of Industry and Information Technology (MIIT), the number of IPTV users in China had reached 312 million as of November 2020, covering nearly 1 billion people, and its growth rate exceeded that of cable television and OTT video. Behind the impressive data, the growth rate of IPTV users has actually slowed down. The average monthly IPTV user growth of China's three major operators was only 1.5 million from July to August in 2020, which was lower than the average monthly growth rate of 4 million in 2019, and much lower than the ultra-high growth in the past three to five years. In the future, it will be difficult to support the operation model with the main goal of increasing user base, and the optimized operation of existing users will be the only way for operators to operate their TV screen services.
Data is the basis of optimized operation. What really empowers the optimized operation is the operational profile model based on data. The smart profile built by ZTE based on the Easyk intelligent data tag system focuses on user profile and leverages the machine learning algorithm model to intelligently manage profile tags in different scenarios. The smart user profile has higher adaptability, matching, and automation over a traditional one.

Building Smart Profile  

Overall Process

A smart profile model is built through the following four steps of data processing.
Step 1 Sharing. Allow the operation-related applications and systems to share data, thereby providing a solid data basis for building a profile model.
Step 2 Atomizing. Atomize shared data to unify the data format, naming rules, and indicator systems. Atomization refers to the integrity and independence of a single data, which provides flexible guarantee for the subsequent construction of the tag system. After the atomization of shared data, a structured data dictionary will be generated to provide fast query and retrieval of atomic data. 
Step 3 Tagging. Perform statistics, analysis, and algorithm processing on the structured atomic data to form the object-oriented semantic tags that can be identified by the operation personnel. 
Step 4 Modeling. Use a specific model to build a profile model oriented to the operation scenario based on the operational object tag system, and finally generate a smart profile model.

Building Tag System
A tag system is the prerequisite and guarantee for building smart profiles, because its design and construction determine the availability and usability of the profile model. Both raw data and data dictionaries are very technical data systems for common operation personnel. They can not be directly operated and used by the operation personnel before being converted by R&D personnel into readable support data. This conversion often needs a few weeks, which reduces operation efficiency.
The creation of a tag starts with defining a tag. The definition of a tag consists of the following information and steps:
Tag classification: Tags are classified based on their objects or scenarios. 
Tag level: Tags are graded according to the affiliation of the objects or scenarios. 
Tag name: Tags usually have a name associated with the operation scenario, which is easy for ordinary operation personnel to identify and use. 
Tag value: The tag value can be automatically given by the data dictionary or manually marked. 
Tag attribute: The attributes of tags are mainly related to the features of scenarios and objects. Different attributes of the same tag determine its application mode in different scenarios.
Tags are divided into real tags and virtual tags. Real tags can be created in two ways. One is the relevant atomic fields directly from the data dictionary, called fact tags, while the other is the indicators generated through simple unification and aggregation based on the data dictionary, called statistic tags. Virtual tags can also be created in two ways. One is the tag created based on an event relationship model, called a model tag, which is related to the specific operation scenario. The other is the prediction tag created based on the AI prediction model. 
An operation scenario model is a model that correlates the objects involved in operation through the operation time. An operation object can be identified through a specific tag set. ZTE has therefore designed the object scenario tag (OST) model. The relationship among object, scenario and tag is as follows:
—An object is a set of tags that have specific physically associated attributes.
—A tag is a readable indicator that represents the features of an object or scenario.
—A scenario is a set of event tags generated when a specific object has a relationship.

Building Smart Profile Based on OST
Building a smart profile based on the OST model involves the following processes:
—Using the tag management platform to generate an OST model manually or through algorithm-based intelligent discovery and association.
—Filtering specific tags through attributes and values based on the OST model and specific operation scenarios, completing group operation of specific objects or scenarios, and generating a model that achieves the targets of operation scenarios.  
—Generating a smart profile based on the above model and specific operation data.
There are two output modes for smart profile. One is to output an indicator report, providing a profile analysis report of a specific operation scenario, and the other is to output profile data of a specific operation scenario through APIs, supporting contact operation by related operation tools. 

Optimized Operation Empowerment

System Architecture
The architecture of an optimized operations support system based on smart profile is shown in Fig. 1. The system architecture is divided into five layers that are associated through data management and orchestration. The entire system empowers the optimized operation at the top layer that consists of four functions such as target monitoring, operation analysis, decision support and closed loop optimization. Based on these four functions, the system can achieve optimized management of operation targets as well as optimized operations support. 


Closed Loop Management of Operation Targets

The operations support system based on smart profile implements closed loop management of operation targets, as shown in Fig. 2. The closed loop management empowers the operation personnel to monitor the achievement of operation targets in real time or on a regular basis and adjust the operation targets according to the analysis report. In this way, the operation targets can be more in line with the actual investment in operation.


Closed Loop Optimization of Routine Operations 
The optimized operations support system carries out data monitoring, target analysis and decision support for routine operations such as recommendations, advertisements and packages. It can also implement closed loop monitoring and optimization based on the results of support strategy implementation, as shown in Fig. 3.


Optimized TV screen operation is the key for operators to transform their business growth model based on a large number of existing users from extensive user expansion to user value exploration. The success of optimized operation depends on data empowerment. ZTE's operations support system based on smart profile transforms data to AI capabilities that support optimized operation through data sharing, atomizing, tagging and modeling, helping operators achieve a qualitative change in the mode of operation.