“When Congress passes laws, federal agencies implement those laws through regulations. These regulations vary in subject and include everything from ensuring water is safe to drink to setting health care standards. Regulations.gov is the place where users can find and comment on regulations.” - Regulation.gov Currently, all the comments are analyzed by government officials. While process works when the volume of comments is low, it fails when a particular regulation receives a large volume of comments. For example, the docket for the Consumer Financial Protection Bureau’s proposed rule on ‘Payday, Vehicle Title, and Certain High-Cost Installment Loans’ received 421,260 comments. Supposing an average of 1 minute is spent per comment, processing these comments would take approximately 878 8-hour work days. In this project, we aim to automatically summarize regulatory documents and cluster comments on those documents to aid policymakers in making better decisions based on the comments. We also hope this project will allow other interested parties to explore the documents and the comments being submitted.

Our goal was to develop an algorithm that would aid human analysts at regulation.gov in the process of summarizing and clustering comments submitted by citizens during each regulation’s mandatory notice-and-comment period. The project aims to summarize government regulations and assess large volumes of comments on proposed these regulations. It utilizes natural language processing techniques like keyword extraction, key sentences extraction, clustering(unsupervised machine learning).

This project was a group effort. My fellow teammates Kinshuk, Emily Witt and Jason Danker made this amazing initiative possible.