American Recovery Act Classification Tool
Description
This project will use AI/ML to better understand how contracts and agreements funded through the American Relief Act support disaster response themes across the country, such as road repair, timber salvage, and recreation site rehabilitation.
Detailed example
The outputs are probabilities for each need category identified in the American Recovery Act. A highly confident prediction will categorize a document into one or more categories while a less-confident result will flag the document for human review.
AI / analytics pattern
Natural Language Processing: AI that processes, interprets, and shares information in human language.
Automation level / stage
b) Pilot – The use case has been deployed in a limited test or pilot capacity.
Expected benefit
This project will promote accountability and enable public reporting of expenditures.
Controls / human review
ATO: No; PIA: Not published
Data needed
This uses word2vec word embeddings pre-trained on Google News (this is not a LLM) and TF-IDF (term frequency/inverse document frequency), a statistical measure of word counts in contracts and statements of work (SOWs) for disaster-recovery projects.