The practice and application of user portrait

User portrait is the tagging of user information, and in essence, user portrait is the tagging of data. There are three common user portrait systems: structured system, unstructured system and semi-structured system. Unstructured systems have no obvious hierarchy and are independent. Semi-structured hierarchies have some concept of hierarchy, but do not have strict dependencies. Structured systems have a strong hierarchy. Take a simple three-level structured tag as an example. The first-level tag has basic attributes and interest preferences, and can be extended to the second-level tag and the third-level tag to be specific to different attributes and interests.

In the field of Internet and e-commerce, user portrait is often used as the basic work of precision marketing and recommendation system. Its functions generally include:

(1) Precision marketing: Based on the characteristics of historical users, the operation personnel can analyze the potential users of the product and the potential needs of users, and then carry out marketing for specific groups through corresponding means.

(2) User analysis: after classifying users according to their attributes and behavioral characteristics, the number and distribution of users with different characteristics can be counted and the distribution characteristics of different user portrait groups can be analyzed.

(3) Data mining: Based on user portraits, developers can build recommendation systems, search engines and advertising delivery systems to improve service accuracy.

(4) Service products: Draw user portraits of products, conduct audience analysis of products, better understand the psychological motivation and behavior habits of users, improve product operation and improve service quality.

(5) Industry report & User research: through user portrait analysis, operators can have a better understanding of industry dynamics, such as consumer habits, consumer preference analysis, consumption differences analysis of different regional categories, etc.

Relying on years of push service accumulation and strong big data analysis ability, Getui launched Getui Portrait SDK (Getui Portrait) to provide APP developers with rich user portrait data and real-time scene recognition ability.

Getuo.com’s unique cold, hot and temperature data tags can effectively analyze users’ online and offline behaviors, dig into user characteristics and help APP operators fully understand user attributes. Among them, “cold data” refers to the user’s basic attributes, such as gender and age level, which are less likely to change. “Warm data” can trace the user’s recent active applications and scenarios, with a certain timeliness; “Hot data” refers to the user’s current scene and real-time user behavior, helping APP operators seize fleeting marketing opportunities.

Getui not only has a rich general label system, but also can jointly model according to the specific needs of customers and output customized labels to meet the operation of APP in different scenarios.

The construction of user portraits requires the participation of both technical and business people to avoid formal user portraits. Twitter also has a few practices for developers to follow.

(1) Label system design. Developers need to know their own data first and determine the form of tags they need to design.

(2) Basic data collection and multi-data source data fusion. To build user profiles, twitter integrates data from twitter and the APP itself.

(3) Achieve unified identification of users. In most cases, many users of the APP are distributed in different account systems, and each tweet will identify them uniformly.

(4) Construction of user portrait feature layer. Characterize each piece of data.

(5) Portrait label rule + algorithm modeling. Both are indispensable, in practical application, the algorithm is difficult to solve the problem, using simple rules can also achieve good results.

(6) Use algorithms to label all users.

(7) Portrait quality monitoring. In the actual application, the user’s portrait will produce certain fluctuation. In order to solve this problem, We set up a corresponding monitoring system to monitor the quality of the portrait.

The overall process of user portrait construction can be divided into three parts. First, basic data processing. Basic data includes user device information, user’s online APP preference and offline scene data, etc.

Second, data processing in portrait. The processing results include online APP preference features and offline scene features.

Third, portrait information table. There should be four types of information in the table: device base attributes; User base portrait, including the user’s gender, age level, relevant consumption level, etc.; User interest portrait, that is, the direction in which users are more interested, for example, users prefer price comparison apps or overseas shopping apps; Other portraits of users, etc.

Machine learning plays an important role in the process of user portrait construction. Machine learning is mainly a process of continuous updating, data cleaning and data storage of massive data. It makes more use of machine learning platform for prediction analysis and model output.

There are two focuses on portrait quality. First, how to optimize the quality. The user portrait model is modified and optimized periodically. Second, pay attention to the fluctuation of portrait quality and give a timely warning of abnormal changes.

The integration of individual portrait SDK can enrich the user analysis dimension of APP, which is mainly reflected in two aspects: First, accurate recommendation. APP operators can recommend different content for different users through various labels such as gender, age level, interests and scenes provided by the image, so as to achieve more refined operation and improve user activity and retention rate.

Second, user clustering. The extrapolation can help APP process user data, complete user portraits and build user clustering models. Meanwhile, through the analysis of user characteristics, Getuan can also map the old users of the APP to a certain cluster, so as to produce the target cluster of the APP, and ultimately help the APP operators to formulate more accurate operation strategies for different user groups.

As the saying goes, “It is not as good as one person to know you when thousands of people tease you.” When the Internet gradually steps into the era of big data, only by truly understanding users can APP obtain and retain users. Based on the complete big data computing architecture of Getuan, the access of GEtuan portrait SDK can not only help developers improve the efficiency of development decisions, but also help APP operators carry out fine operation, thus improving the marketing efficiency and market competitiveness of enterprises.