User travel behavior tag identification method and device
By setting a rule-prioritizing mechanism for key feature subsets of different travel behavior labels and combining it with model prediction, efficient user travel behavior identification is achieved when the user scale is large. This solves the problem of high computational overhead in existing technologies and maintains the accuracy and efficiency of identification.
CN122114996AActive Publication Date: 2026-05-29ZHEJIANG YUNTONG SHUDA TECH CO LTD
7 Cites 0 Cited by
Patent Information
- Application Number
- CN202610553917.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-04-24
- Publication Date
- 2026-05-29
- Estimated Expiration
- 2046-04-24
AI Technical Summary
Technical Problem
Existing methods for recognizing user travel behavior, while ensuring accuracy, suffer from high computational overhead, making it difficult to meet real-time requirements, especially when dealing with a large user base.
Method used
It adopts a two-stage architecture of rule priority and model fallback. By setting different key feature subsets for different travel behavior labels, the results are directly output in the rule matching stage, and the prediction model is only called for users who cannot be covered by the rules, thus reducing the number of model calls.
Benefits of technology
While reducing computational overhead, it maintains high recognition coverage and accuracy, thereby improving overall recognition efficiency.
✦ Generated by Eureka AI based on patent content.
Abstract
The invention relates to the field of data processing, in particular to a user travel behavior tag identification method and device, and the method comprises the steps: obtaining target user information, and converting the target user information into a user feature vector; based on judgment conditions corresponding to a plurality of preset travel behavior tags and a plurality of travel behavior tag prediction models, determining a travel behavior tag corresponding to the user feature vector; if a preset association relationship exists between the user feature vector and any judgment condition, taking the travel behavior tag corresponding to the judgment condition as a result; otherwise, inputting the user feature vector into a plurality of travel behavior label prediction models, and outputting a determination result according to the models. According to the method, differentiated key feature subsets are set for different travel behaviors to serve as judgment conditions, and a two-stage framework of rule priority and model understanding is combined, so that the calling frequency of a complex model is reduced while the recognition accuracy is ensured, and the overall recognition efficiency is improved.
Need to check novelty before this filing date? Find Prior Art