Synthetic Data Engine

Overview

Development stage

DeployedSource: c) Deployed – The use case is being actively authorized or utilized to support the functions or mission of an agency.

High-impact designation

Not high-impactSource: c) Not high-impact

Impact justification

The output is not presumed to be high-impact and is not used as the principal basis for significant decisions/actions

Start date

2022-03-28 (day precision)Source: 3/28/2022

Withheld from public reporting

NoSource: a) No

Mission

Topic area

Information Technology

Operational functions

Translation and transcription Derived by keyword rules; see methodology

Problem the AI is intended to solve

This AI based synthetic data generator provides synthetic data for a robust, large-scale dataset consistent with US Population Data Heuristics for testing and advanced analytics by dimensionalizing over 180 elements of socioeconomic data and over 300 dimensions, aging people, households and businesses over time, and tightly correlating the synthetic data to actual familial and socioeconomic attributes of US taxpayers.

Expected benefits

Generate synthetic data for testing IRS tax processing systems and significantly reduce the risk of exposing taxpayer information while enabling more comprehensive testing of functional test cases. This system can now use to simulate fraud cases for testing negative and positive automated test cases.

System outputs

AI is capable of outputting 3 tax years of IMF and BMF returns starting in Tax Year 2022. With each XML Schema Definition (XSD) that is released under the Modernized e-File (MeF), system implements each XSD version and provides full regression support for testing in successive years. Each XSD version has automated tests that run to ensure verifications can be run to reduce the likelihood of AI generated synthetic returns that might have data anomalies. Also, to help seed test systems with synthetic individuals and synthetic businesses, system can output the Data Master 1 (DM1) file simulating a feed from Social Security Administration (SSA) and Application for Employer Identification Number file (Form SS-4) containing transcribed form information so that the test system can be initialized with the appropriate backend reference data for validation of IMF and BMF entities, as well as the ability to validate incoming synthetic tax returns.

Technology

AI classification

Classical / predictive MLSource: Classical/Predictive Machine Learning: Models trained on data to make predictions or classifications based on identified patterns or relatio…

System name(s)

Not reported

Custom-developed code

Yes

Public source code

Not reported

Sourcing

How it was built

Purchased from vendorSource: a) Purchased from a vendor

Vendor (as reported)

Not reported

Vendors (standardized)

None on the standard list

A vendor is the supplier named by the agency. It does not identify the underlying model or AI technology.

Data and privacy

Involves PII

No

Privacy Impact Assessment

Not reported

Authorization to Operate

No

Demographic features

a) Race/Ethnicity; b) Sex; c) Age; e) Socioeconomic Status; g) Residency Status; h) Marital Status; i) Income; j) Employment Status

Training and evaluation data

Over 140 input feeds including US Census, US Bureau of Labor and many other sources, understand relations between variables and find the most relevant with regards to household income levels. divide the population into distinct groups defined by value ranges in sets of variables, generate new population for each group maintaining the frequency distribution for variables.

Federal Data Catalog

Not reported

Governance

0 of 8 minimum-practice questions answered. A blank answer means the agency reported nothing; it does not mean the practice is absent.

Potential impacts and how they were identified

Not reported