Predictive Fraud Model via Data Pooling

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing fraud detection systems face challenges due to inadequate volumes or numbers of business transactions, particularly fraudulent transactions, which hinder the generation of accurate predictive models for identifying fraudulent activities.

Innovation Solution

A system where multiple businesses contribute their transaction data, including fraudulent and non-fraudulent examples, to a shared pool, which is then used to generate and update predictive models using data mining algorithms, ensuring a rich dataset for improved pattern recognition and model accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Quantity of substance

If a single organization uses its own transaction data to generate predictive models, then data privacy and security are maintained, but the volume and diversity of training data are insufficient

Engineering Contradiction:

Improvevolume of transaction dataVSAvoiddata pooling system complexity

Core Design Contradiction:

Quantity of substanceVSDevice complexity

Solution Approach 1:

The patent combines transaction data from multiple organizations into a centralized pool, merging previously separate data sources to achieve sufficient volume and diversity for training accurate predictive models while maintaining individual organization ownership

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The patent introduces a data pooling system as an intermediary layer between individual organizations and the predictive model generation process, allowing data to be aggregated and processed without direct exposure between participating organizations

Inventive Principle:

Principle #24Intermediary (Mediator)

2Measurement precision

If fraudulent transactions are heavily weighted in the training dataset, then the model can learn fraud patterns effectively, but the model may become biased and perform poorly on legitimate transactions

Engineering Contradiction:

Improvefraud detection accuracyVSAvoidmodel generalization performance

Core Design Contradiction:

Measurement precisionVSEase of operation

Solution Approach 1:

The patent dynamically adjusts the weighting parameter for fraudulent transactions during model training, allowing the system to optimize between fraud detection sensitivity and overall model generalization based on the specific data distribution and performance metrics

Inventive Principle:

Principle #35Parameter changes

3Measurement precision

If the predictive model is continuously updated with new data, then model accuracy improves over time, but computational resources and processing time increase

Engineering Contradiction:

Improvemodel accuracyVSAvoidmodel training time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent implements selective model updating by retraining only when necessary conditions are met (such as performance degradation thresholds or significant data distribution changes), rather than continuously retraining with all available data, thus balancing accuracy improvement with computational efficiency

Inventive Principle:

Principle #16Partial or excessive action

Data Source

PatentUS7693767B2Method for generating predictive models for a business problem via supervised learning

Publication Date: 2010.04.06 ORACLE INT CORP

AI summary

A method for solving a business problem includes pooling transaction data received from a plurality of subscribers over a network, the transaction data including samples of fraudulent transactions. A data mining algorithm is then applied to the pooled transaction data, resulting in a predictive model that detects a fraudulent transaction. The predictive model is then provided to the subscribers in exchange for a subscription fee. It is emphasized that this abstract is provided to comply with the rules requiring an abstract that will allow a searcher or other reader to quickly ascertain the subject matter of the technical disclosure. It is submitted with the understanding that it will not be used to interpret or limit the scope or meaning of the claims. 37 CFR 1.72(b).