Abnormal Access User Detection via Platform Overlap Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Content display platforms face challenges in accurately identifying abnormal access users, as these users can mimic normal access behaviors, leading to reduced accuracy and increased network congestion, which affects the promotion effectiveness and expense of commodities or services.
Innovation Solution
A data processing method and apparatus that calculates access user overlapping degrees between content display platforms to identify abnormally accessed platforms and users, thereby improving accuracy and reducing the complexity of identifying abnormal access users.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If abnormal access users are identified by analyzing access behaviors of each access user, then the identification process can be performed, but the accuracy of identifying abnormal access users is reduced because abnormal access users can imitate normal access behaviors
Solution Approach 1:
The patent transitions from analyzing access behaviors within a single content display platform to analyzing the overlap of access users across multiple content display platforms. This dimensional change from single-platform behavior analysis to multi-platform user overlap analysis enables the system to identify abnormal users who access multiple platforms, thereby resolving the limitation of behavior imitation within a single platform.
Solution Approach 2:
The patent introduces 'access user overlap degree' as an intermediary metric to identify abnormal access users. Instead of directly analyzing access behaviors, the system uses the overlap degree between access user sets of different content display platforms as a mediator to detect abnormal users, thereby avoiding the problem of behavior imitation.
2Measurement precision
If all access users belonging to at least two content display platforms are analyzed to identify abnormal users, then comprehensive identification can be achieved, but the complexity and time consumption of the identification process increases
Solution Approach 1:
The patent extracts only the essential information needed for identification - the access user sets of different content display platforms - and calculates their overlap degree. This extraction approach avoids the need to analyze all detailed access behaviors of all users, significantly reducing time consumption while maintaining identification comprehensiveness.
Solution Approach 2:
The patent performs partial analysis by focusing only on the overlap portion of access user sets across platforms rather than analyzing all access behaviors in detail. This partial action approach achieves effective identification of abnormal users without the excessive time cost of comprehensive behavior analysis.
3Productivity
If abnormal access users are not efficiently screened out, then all users must be processed, but network congestion occurs and promotion effectiveness is reduced
Solution Approach 1:
The patent performs preliminary identification of abnormal access users by calculating access user overlap degrees before processing all user data. This preliminary action enables the system to screen out abnormal users in advance, preventing network congestion and reducing wasted promotion expenses on fake users.
Data Source
AI summary
A data processing method includes: acquiring access users associated with at least two content display platforms, the at least two content display platforms being configured to provide business contents to the access users; generating access user overlapping degrees between pairs of content display platforms in the at least two content display platforms according to the access users; determining abnormally accessed content display platforms from the at least two content display platforms according to the access user overlapping degrees and regarding the determined abnormally accessed content display platforms; and determining abnormal access users from target access users belonging to the target content display platforms.