Skip to content
Syntasa

Customer story · Federal Regulation

98% of duplicate comments found, live in six weeks.

A U.S. federal regulatory agency expected a flood of public comments on a proposed rule. With Syntasa, Google Public Sector and T&T Consulting Services, its reviewers saw over 98% of duplicate and near-duplicate comments identified, on a system live in six weeks.

Talk with our teamDownload the case study

Results

A faster review, live in six weeks.

  • Duplicates found98%+of duplicate and near-duplicate comments identified
  • Time to live6 weeksfrom start to a system in production, with results right away

Context

Every comment read, however many arrive.

The agency's main job is to create or change the rules that guide the industry it regulates. Once in place, those rules have the force of law and are enforced by the agency.

Part of the process is asking the public for comments on a proposed rule, which keeps the agency transparent and accountable. In late 2023 the agency proposed a rule that, like earlier ones, was likely to draw a lot of attention.

The constraint

A flood of comments, on a deadline.

  1. Every comment reviewedLegal experts had to review each public comment to reduce the risk of legal challenges.
  2. A volume that could overwhelmThe expected number of comments threatened to swamp the team in a short review window.
  3. Duplicates everywhereEarlier reviews showed many comments were duplicates or minor variations of others.
  4. Compliant and secureAny solution had to meet regulatory standards and keep the data confidential.

Deployment

From public website to reviewer dashboard.

Syntasa, Google Public Sector and T&T Consulting Services used Syntasa's pre-built Behavioral Sentiment Analytics solution on Google Cloud.

  1. Step 1: Bring the comments in securely

    Comments are transferred securely from a public website to Google Cloud Storage.

  2. Step 2: Find exact and near duplicates

    An advanced deduplication process, using Google Cloud large language models, separates exact and near duplicates from original comments.

  3. Step 3: Read sentiment, summarize, search

    A pre-trained model scores each comment for or against the rule, long documents get brief summaries, and search finds similar comments by keyword.

  4. Step 4: Review in Looker

    Custom Looker dashboards, refined with reviewers, assign comments and let reviewers make changes and add notes, behind multi-factor authentication.

Outcome

What changed for the mission.

  • Duplicates out of the way

    Over 98% of duplicate and near-duplicate comments identified, which dramatically streamlined the review.

  • Live in six weeks

    Under T&T Consulting Services' guidance, the whole system went live within six weeks, with results available straight away.

  • Secure from end to end

    Led by T&T's security and compliance experts, multi-factor authentication protected the processed data, keeping it confidential at every stage.

  • Welcomed across the agency

    Executive leadership, information security, attorneys and reviewers all praised the efficiency and security of the review.

FAQ

Questions teams ask first.

Case study · PDF · 4 pages

Read the full Federal regulatory agency story.

The full story, with the architecture, how it works, a timeline and the detailed results.

Download the case study

Related.

Your constraint won't be our first.

Tell us what you are working under. We will bring the closest story and the engineers who delivered it.

Talk with our teamDownload the case studyA working session with our engineers, on your data and constraints.