Fraudulent Advertiser Account Detection System Using Conditional Probability

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Website operators face significant financial losses due to fraudulent advertising accounts established using stolen or illegally obtained payment information, leading to chargebacks and potential halting of financial transactions, as existing methods lack effective real-time detection and verification mechanisms.

Innovation Solution

A fraudulent advertiser account detection system that analyzes property and behavioral attributes of advertiser accounts using statistical methods to identify suspicious accounts, determining initial probabilities and likelihoods, and assigning conditional probability values to autonomously flag and verify accounts, allowing for near-real-time identification and prevention of fraudulent transactions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If traditional advertising account verification methods are used, then account setup is simple and fast, but fraudulent accounts go undetected leading to financial losses

Engineering Contradiction:

Improvefraud detection accuracyVSAvoidverification system complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The system performs preliminary analysis of advertiser account attributes during account creation and before payment processing. By evaluating property attributes (contact information, IP address, payment information) and behavioral attributes (account creation patterns, transaction history) in advance, the system identifies fraudulent accounts before they can cause financial harm, eliminating the need for complex post-detection remediation

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The verification system autonomously evaluates advertiser accounts using automated statistical analysis and machine learning algorithms. The system self-updates its detection models by continuously analyzing new account data and fraud patterns, reducing the need for manual verification processes while improving detection accuracy over time

Inventive Principle:

Principle #25Self-service

2Loss of energy

If real-time fraud detection is implemented, then chargebacks are minimized, but processing time and computational resources increase

Engineering Contradiction:

Improvefinancial loss from chargebacksVSAvoidaccount verification time

Core Design Contradiction:

Loss of energyVSLoss of time

Solution Approach 1:

The system performs fraud detection evaluations in advance during account creation and before payment processing. By determining initial probabilities of fraud and evaluating account attributes beforehand, the system prevents chargebacks from occurring in the first place, eliminating the time and resources that would be spent on post-chargeback dispute resolution

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system replaces manual fraud review processes with automated statistical analysis and machine learning algorithms. This substitution enables real-time fraud detection without requiring human intervention, maintaining fast processing speeds while improving detection accuracy through continuous automated analysis of account patterns

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Measurement precision

If comprehensive account attribute analysis is performed, then fraud detection accuracy improves, but system complexity and processing overhead increase

Engineering Contradiction:

Improvefraud detection precisionVSAvoidanalysis system complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The system segments fraud detection into distinct evaluation components: property attribute analysis (contact information, IP address, payment information), behavioral attribute analysis (account creation patterns, transaction history), and probability calculation. Each segment is processed independently using specialized algorithms, making the overall complex system manageable and maintainable while achieving high detection precision through comprehensive attribute evaluation

Inventive Principle:

Principle #1Segmentation

Data Source

PatentUS10108968B1Apparatus, method and article to facilitate automatic detection and removal of fraudulent advertising accounts in a network environment

Publication Date: 2018.10.23 PLENTYOFFISH MEDIA ULC

AI summary

A fraudulent advertiser account detection system may be used by a Website operator to identify active advertiser accounts as at least one of either a known valid advertiser account or a known fraudulent advertiser account. Each advertiser account has a logically associated account profile. Each account profile includes a number of advertiser attributes, each attribute having one or more logically associated attribute values characterizing the respective advertiser. Using a plurality of known valid advertiser accounts and plurality of known fraudulent advertiser accounts, the system determines a number of initial probabilities and likelihoods that one or more attribute values and/or attribute value combinations present in an advertiser account profile indicate at least one of a known valid or a known fraudulent advertiser account. Using these probabilities and likelihoods, the system determines for each advertiser account a conditional probability value indicative of whether the respective account is a fraudulent advertiser account.