Text Analytics Data Science Sentence Similarity Model

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

High-impactSource: a) High-impact

Impact justification

Not reported

Start date

2019-11-01 (day precision)Source: 2019-11-01T00:00:00

Withheld from public reporting

Not reported

Mission

Topic area

Law Enforcement

Operational functions

Fraud, waste and improper payments Derived by keyword rules; see methodology

Problem the AI is intended to solve

The Text Analytics capability employs machine learning and data graphing techniques to identify patterns that may indicate potential fraud, national security, and/or public safety concerns by scanning the digitized narrative sections of the associated applications and looking for common language patterns.

Expected benefits

Text Analytics augments the tedious and time-consuming manual process to identify potential fraud, national security, and/or public safety concerns and enables the identification of such concerns across jurisdictional boundaries. It increases the integrity of immigration programs, strengthens officers’ confidence in their work, and contributes to the reduction in customer wait times.

System outputs

Text Analytics does not make predictions, recommendations, or decisions. It is merely a research tool that identifies potential patterns, while remaining agnostic as to whether those patterns identify potential fraud, national security, and/or public safety concerns. Instead, trained staff evaluate the patterns to determine whether they identify potential concerns and then validate and/or invalidate those potential concerns through the course of their investigations or adjudications.

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)

Text Analytics

Custom-developed code

Yes

Public source code

Not reported

Sourcing

How it was built

Contract and in-houseSource: c) Developed with both contracting and in-house resources

Vendor (as reported)

Inadev

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

Yes

Privacy Impact Assessment

www.dhs.gov ↗

Authorization to Operate

Yes

Demographic features

Not reported

Training and evaluation data

Text Analytics stores information extracted from benefit forms and supporting documents, focusing on the narrative portions of those documents.

Federal Data Catalog

Not reported

Governance

8 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

There is small risk of false positives or false negatives due to the model. The risk is mitigated through a manual review of any information produced from the tool. Text Analytics is a decision support tool. Text Analytics does not make recommendations of fraud or benefit / adjudication decisions, any decisions made from information stored in the tool is conducted through a manual review by a USCIS employee.