Online User Activity Detection via Distribution Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Online advertising systems are vulnerable to fraudulent activities, such as botnets simulating user traffic, which can lead to revenue loss and trust issues, as existing methods for detecting abnormal user activity are not scalable for large-scale operations.
Innovation Solution
A system and method for detecting abnormal online user activities by establishing baseline distribution data for normal user interactions and comparing it with dynamic distribution data in real-time, using measures like Kolmogorov-Smirnov and Information Value to identify deviations, thereby flagging potential fraudulent behavior.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional user activity monitoring methods are used, then detection of fraudulent activity is possible, but the system is not scalable for large-scale operations
Solution Approach 1:
The patent transforms user activity detection from examining individual activity patterns to analyzing statistical distribution parameters (CDFs) of aggregated user behaviors. By computing measures like KS statistic and Information Value on distribution parameters rather than individual events, the system achieves both high detection accuracy and scalability to large datasets in real-time
Solution Approach 2:
The patent creates simplified statistical models (distribution functions) that replicate the essential characteristics of complex user behavior patterns. These distribution models serve as computationally efficient representations that can be rapidly compared to detect anomalies without processing every individual user interaction, enabling scalable real-time detection
2Reliability
If real-time detection of abnormal user activity is implemented, then fraudulent practices can be identified promptly, but computational complexity increases
Solution Approach 1:
The patent reduces computational complexity by transforming raw user activity data into distribution parameter comparisons. Instead of analyzing individual user interactions in real-time, the system computes statistical measures (KS statistic, Information Value) on aggregated distributions, dramatically reducing the computational burden while maintaining real-time detection capability
Solution Approach 2:
The patent pre-computes baseline distribution data representing normal user behavior patterns before real-time detection begins. This preliminary establishment of reference distributions allows the system to perform rapid comparisons during real-time operation without performing complex analyses on the fly, reducing real-time computational requirements
Data Source
AI summary
The present teaching generally relates to detecting abnormal user activity associated with an entity. In a non-limiting embodiment, baseline distribution data representing a baseline distribution characterizing normal user activities for an entity may be obtained. Information related to online user activities with respect to the entity may be received, distribution data representation a dynamic distribution may be determined based, at least in part, on the information. One or more measures characterizing a difference between the baseline distribution and the dynamic distribution may be computed, and in real-time it may be assessed whether the information indicates abnormal user activity. If the first information indicates abnormal user activity, then output data including the distribution data and the one or more measures may be generated.