A user portrait accurate delivery targeting method based on label dynamic analysis

By dynamically analyzing tags, assessing tag shift trends and solidification levels, and dynamically adjusting recommendation strategies, the problem of difficulty in reflecting changes in user profiles in real time is solved, resulting in more accurate content delivery and improved user experience.

CN121860707BActive Publication Date: 2026-06-09YUNDONG (SHANGHAI) TECH CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN Β· China
Patent Type
Patents(China)
Current Assignee / Owner
YUNDONG (SHANGHAI) TECH CO LTD
Filing Date
2026-03-12
Publication Date
2026-06-09

AI Technical Summary

Technical Problem

Existing user profiles rely heavily on historical tags and static interest features, making it difficult to reflect the dynamic changes in user interests in real time. They also lack comprehensive analysis of complex behavioral features such as tag-jumping behavior, search rollback, and interaction trigger frequency, resulting in lagging recommended content and a decline in user experience.

Method used

By dynamically analyzing tags, we can assess tag shift trends, tag solidification levels, content spillover coefficients, and tag stickiness indices, and dynamically adjust recommendation strategies, including tag monitoring, dwell time analysis, tag bounce characteristic evaluation, and interaction frequency statistics, to achieve precise content delivery.

Benefits of technology

It improves the real-time nature of user profiles and the timeliness of content distribution, avoiding content lag and user interest fatigue caused by fixed tags, and achieving more accurate and intelligent targeting.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121860707B_ABST
    Figure CN121860707B_ABST
Patent Text Reader

Abstract

The application discloses a user portrait accurate delivery orientation method based on label dynamic analysis, relates to the technical field of delivery orientation, and is used for solving the problems of lagging recommended content and decreased user experience. When a group label is newly added in a candidate label database, the newly added label is marked and a monitoring time is set. The staying time data of group users under the marked group label is collected and the label offset trend is evaluated. The label solidification degree is calculated in combination with the label jump-out feature. The content overflow coefficient is set by using the label solidification degree, and the injection proportion of the content of the marked group label in the recommended scheduling pool is controlled accordingly. The search back-off proportion and the interaction trigger frequency of the group users are counted. The label adhesion index of the marked group label is fused and generated. Whether the marked group label is stored to the label database is judged. The real-time performance of the group users and the timeliness of content distribution are improved. More accurate and intelligent delivery orientation is realized.
Need to check novelty before this filing date? Find Prior Art