Fireside’s NLP Solution Helps Congress Respond to Constituents

  • Developed advanced Natural Language Processing (NLP) and text analytics models to classify and extract information from communications sent to members of the House of Representatives
  • Organized and grouped communications by issues, themes, and keywords to enable elected officials to effectively respond to constituent’s inquiries
  • Enabled Fireside to bring state-of-the-art language processing and communication analytics capabilities to the House of Representatives

“Miner & Kasch gave us the tools to bring next generation analytics to an ecosystem that still relies on fax machines.”

– Scott Sadlo, Product Manager, Fireside

Company Overview
Fireside provides constituent relations solutions to over one-third of the members of the United States House of Representatives. Fireside facilitates communication services between representatives and their constituents and provides tools to congressional offices to organize and respond to incoming messages.
Each congressional office receives hundreds of pieces of mail and email per day. Identifying themes and sentiment in this communication is a time consuming and tedious task. Too often, delays in responses and form letter replies leave constituents feeling that their concerns are not given attention by the office. Due to the volume of communication, elected officials tend to focus only on broader issues from their emails to get a sense of their constituents concerns or where they stand.
Fireside along with Miner & Kasch, a machine learning (ML) and artificial intelligence (AI) company, started an in-depth investigation into how congressional offices currently communicate with their constituents, and what information elected officials wish they could obtain. The biggest asks included:

  • Obtaining an overview of the issues that constituents care about
  • Organizing the emails to drive better engagement through personalized responses
  • Routing messages to the correct specialists in the office (e.g., casework requests)

To address these asks, our data scientists first designed, developed, and implemented natural language processing pipelines, models, and tools to extract entities from emails. These entities include references to proposed legislation and bills, people, and places. This extracted information is used in the grouping and organization of messages by content.

We implemented state-of-the-art methods to capture aspect-based sentiment in emails. This capability enables users of the solution to identify the stance that a constituent has towards a specific topic mentioned in their email.

Lastly, we applied transfer learning and model augmentation techniques to pre-trained NLP models to perform classification of messages into widely accepted policy areas. This capability gives representatives the ability to organize messages effectively by policies including education, economics, and environmental issues, and allows for a more in-depth understanding of micro-issues in the district.

Firesides’ Communication Analysis solution saves elected officials time and allows them to engage with their constituents more effectively:

  • Representatives increase their insights into the issues that constituents care about and can more accurately represent their constituents
  • Representatives can process more responses and personalize replies based on issues, policy areas, and communication content

We’d love to hear from you

Contact us for more information or to discuss what problem we may be able to help you solve.