DHS · DHS-2727 · record R1883

Large Language Model (LLM) Guided Data Dictionary Generation

Department of Homeland Security · FEMA

Overview

Development stage: PilotSource: b) Pilot – The use case has been deployed in a limited test or pilot capacity.
High-impact designation: Not high-impactSource: c) Not high-impact
Impact justification: Not reported
Start date: 2025-09-01 (day precision)Source: 2025-09-01T00:00:00
Withheld from public reporting: Not reported

Mission

Topic area: Information Technology
Operational functions: Search and knowledge retrieval Derived by keyword rules; see methodology

Problem the AI is intended to solve

This LLM-generated data dictionary eases the burden of metadata documentation on the data stewards when integrating their data into FEMADex by creating field definitions.

Expected benefits

By utilizing this LLM-generated data dictionary, it automatically creates field definitions, saving data stewards significant time and effort during data onboarding.

System outputs

This LLM model utilizes Retrieval Augmented Generation (RAG) technique for data dictionary generation by first retrieving relevant provided metadata from 1) the source system intake form and 2) an acronym key. The LLM then uses this context to generate clear, brief dscriptions for each field.

Technology

AI classification: Generative AISource: Generative AI: AI that generates new or synthetic content (e.g., images, videos, audio, text, code).
System name(s): Not reported
Custom-developed code: Yes
Public source code: Not reported

Sourcing

How it was built: Developed in-houseSource: b) Developed in-house
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: Not reported
Training and evaluation data: This model utilizes metadata documentation provided by datastewards, known as the Source System Intake Form (SSIF), during source system intake for FEMADex. Additionally an acronym key for each source system is also used and provided by the respective data stewards. The SSIF and acronym key are unique for each source system.
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.

Pre-deployment testingNo answer reported

AI impact assessmentNo answer reported

Independent reviewNo answer reported

Ongoing monitoringNo answer reported

Operator trainingNo answer reported

Fail-safeNo answer reported

Appeal processNo answer reported

User and public consultationNo answer reported

Potential impacts and how they were identified

Not reported