Account Clustering for Illegitimate Identification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for identifying illegitimate accounts in online services are inefficient and impractical, requiring significant manual effort and time, especially as the number of accounts increases, making them unsuitable for scalable solutions.
Innovation Solution
A system that clusters accounts based on selected features using a force-directed graph visualization process, such as the Fruchterman-Reingold algorithm, to generate groups of nodes representing accounts, determining whether these clusters consist of illegitimate or legitimate accounts by acquiring legitimacy classifications.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional manual methods are used to identify illegitimate accounts, then identification accuracy can be maintained through human review, but the process requires significant manual effort and time, making it inefficient and impractical as the number of accounts increases
Solution Approach 1:
The system enables self-service by implementing automated clustering algorithms that automatically group accounts based on feature similarities without requiring manual intervention. The algorithm autonomously processes account data, generates clusters, and identifies illegitimate accounts through pattern recognition, allowing the system to serve itself in the identification process.
Solution Approach 2:
The patent replaces the mechanical manual review process with an automated computational system. Instead of human operators manually examining accounts, the system uses clustering algorithms and machine learning models to automatically analyze account features, generate clusters, and identify illegitimate accounts, substituting mechanical human effort with automated computational processing.
2Quantity of substance
If the number of accounts to be reviewed increases, then the scale of the online service grows and more revenue can be generated, but conventional manual methods become increasingly impractical and unsustainable
Solution Approach 1:
The system implements dynamics by designing an adaptive clustering algorithm that can dynamically adjust to varying account volumes and characteristics. The algorithm automatically scales its processing capacity based on the number of accounts, adjusting cluster generation parameters and processing throughput to maintain identification effectiveness regardless of the scale of account growth.
Solution Approach 2:
The patent applies parameter changes by modifying clustering parameters such as cluster size thresholds, feature weightings, and similarity criteria based on the volume and characteristics of accounts being processed. This allows the system to optimize its identification performance across different scales, adjusting parameters dynamically as the quantity of accounts increases to maintain scalability.
3Productivity
If automated clustering algorithms are used to identify illegitimate accounts, then processing speed and scalability are improved, but the complexity of the system increases due to the need for feature selection and cluster validation
Solution Approach 1:
The system applies segmentation by dividing the complex account identification process into distinct modular components: feature extraction module, clustering algorithm module, cluster validation module, and illegitimate account identification module. Each module handles a specific aspect of the process, making the overall system more manageable and easier to implement despite the increased automation capability.
Solution Approach 2:
The patent introduces an intermediary feature selection layer that acts as a mediator between the raw account data and the clustering algorithm. This intermediary component pre-processes and selects relevant features, simplifying the input to the clustering algorithm and reducing the complexity of parameter tuning required for effective cluster generation and validation.
Data Source
AI summary
Systems, methods, and non-transitory computer-readable media can acquire a plurality of accounts associated with a set of features. Each account in the plurality of accounts can be associated with a respective set of feature values for the set of features. A selection for a subset of features out of the set of features can be received. A group of clusters can be generated based on the selection for the subset of features. Each cluster in the group of clusters can include a respective collection of nodes representing at least some of the plurality of accounts. It can be determined whether a particular collection of nodes, included in at least one cluster out of the group of clusters, represents illegitimate accounts or legitimate accounts.