Internet Market Disparate-Impact Assessment With Robotic Users

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing research has not adequately addressed the issue of disparate impact in online retailing, particularly in the context of necessary goods like groceries, due to the lack of data sharing by online retailers and the differences between online and offline shopping environments, making it difficult to assess how pricing algorithms may discriminate against different consumer groups.

Innovation Solution

A methodology using internet crawlers to collect unbiased offering data across a large number of products and geographic areas, mimicking consumer behaviors to assess disparate impact without collecting or storing demographic data, and matching it with public data sources for analysis.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If online retailers collect and store demographic data of consumers to assess pricing fairness, then they can directly analyze disparate impact, but they face increased legal liability and data privacy risks

Engineering Contradiction:

Improveassessment accuracyVSAvoidlegal liability

Core Design Contradiction:

Measurement precisionVSObject-affected harmful factors

Solution Approach 1:

The patent introduces an intermediary research platform that acts as a mediator between retailers and demographic data. The platform collects demographic data from public sources and conducts fairness assessments independently, allowing retailers to obtain assessment results without directly handling sensitive demographic information, thus reducing legal liability while maintaining measurement precision

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent creates a simulated consumer environment where robotic users impersonate consumers from different demographic groups to interact with the online retailer. This copying approach allows the system to observe pricing algorithms' behavior toward different demographic groups without collecting actual consumer demographic data, thereby assessing disparate impact while avoiding legal risks

Inventive Principle:

Principle #26Copying

2Object-affected harmful factors

If online retailers do not collect demographic data, then they avoid legal liability, but they cannot assess how pricing algorithms affect different demographic groups

Engineering Contradiction:

Improvelegal liabilityVSAvoiddemographic information

Core Design Contradiction:

Object-affected harmful factorsVSLoss of information

Solution Approach 1:

The research platform serves as an intermediary that collects demographic information from public sources and matches it with pricing data, enabling demographic analysis without the retailer directly handling sensitive information. This resolves the contradiction by providing demographic insights while maintaining legal compliance

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent replaces the traditional mechanical approach of directly collecting consumer demographic data with an automated robotic user system that impersonates consumers and observes pricing behavior. This substitution allows the system to gather pricing information correlated with demographics through simulation rather than direct data collection

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Loss of information

If brick-and-mortar retailers share local pricing and demographic data with researchers, then researchers can study disparate impact, but online retailers do not share such data making research difficult

Engineering Contradiction:

Improvedata availabilityVSAvoidmethodology applicability

Core Design Contradiction:

Loss of informationVSAdaptability or versatility

Solution Approach 1:

The patent employs robotic users that copy human consumer behavior to interact with online retailers and collect pricing data. This copying mechanism enables the system to gather pricing information from online retailers without requiring them to share internal data, making the methodology adaptable to online environments where data sharing is not practiced

Inventive Principle:

Principle #26Copying

Solution Approach 2:

Instead of requiring retailers to provide data (the traditional approach), the patent inverts the approach by having robotic users directly interact with the retailer's public-facing systems to collect pricing data. This inversion makes the methodology applicable to online retailers who do not share internal data

Inventive Principle:

Principle #13The other way round (Inversion)

Data Source

PatentUS12462268B2Method for assessing disparate impact in internet markets

Publication Date: 2025.11.04 UNIVERSITY OF SOUTH CAROLINA

AI summary

Disclosed methodology assesses the existence of disparate impact in internet markets that serve geographically dispersed consumers. Implementations of the method can collect unbiased offering (e.g., pricing and/or other fees and costs) data for a large number of products and geographic areas, so that marketing decisions, such as price, recommendations, and delivery fees can be matched to consumer demographic data from established sources such as censuses and large scale surveys. The combined data can then be used to investigate the presence and nature of disparate impact and can be used by internet platforms and retailers to audit their algorithms for disparate impact without collecting or holding the demographic data of their own users. Thus, a methodology is provided for the collection of data required to study the extent to which algorithms in internet markets may induce disparities across demographic consumer groups and whether disparities can be justified by valid interests.