RedShred receives $745K to use AI to find the key points in long, boring documents

There’s plenty of valuable information in the long, complex documents that the federal government releases. For businesses seeking out government work, they’re especially important when it comes to seeking out new contracts or grants.
The task of poring over the documents for that key information, however, can be cumbersome. In past work for a contractor, Jeehye Yun found this firsthand. She had to comb documents like requests for proposals, where they could find potential business opportunities. While the right information was there, there was plenty more that wasn’t relevant. In the process of providing independent verification and analysis, Yun also read lots of policy documents, which often contained similar ideas, just repeated with slightly different variations.
In seeking ways to make the process easier, Yun began to consider the documents from a data perspective, and how technology could help distill it down to what a person needed. That’s where artificial intelligence entered the equation. In 2014, Yun teamed with cofounder Jim Kukla to create RedShred. Working out of bwtech@UMBC, the cofounders teamed with UMBC researchers including Dr. Tim Finin and others in the Department of Computer Science and Electrical Engineering as it developed the platform.
The startup seeks to eliminate the need to read long documents to find the data they’re seeking. After a user supplies what they’re looking for, the platform applies machine learning and natural language processing to produce an “at-a-glance” summary of documents where that info might be found. It can also identify key topic areas and facts, as well as tag documents with metadata.
Overall, Yun said, the system is designed to “help people find answers to specific questions and information that would be relevant to what they’re doing.” When it comes to helping institutions with federal procurement or grants, the goal is to identify the right information that will help pre-qualify opportunities.

A grant from the National Science Foundation through the U.S. Small Business Innovation Research (SBIR) program provided $225,000 helped to develop the application for federal documents. Along with contractors, the system has also been used by universities.
The company recently received $745,000 in funding. The Phase II grant will help the startup expand the platform to additional capabilities over the next 18 months to help more kinds of businesses.
Working with federal documents offered a large initial dataset that was publicly available. The technology that underlies the platform is designed to be applied for different kinds of complex documents.
“The goal is to make that more useful across the board,” Yun said. “…We just happened to start with government documents.”