Combating Fraud, Waste, and Abuse (FWA)
The Small Business Innovation Research (SBIR) program is arguably the most successful, high return on investment (ROI) small business R&D program in the federal government. To maintain the integrity of the SBIR program, agencies that make SBIR contract awards are required by law to implement a system for prevention of fraud, waste and abuse (FWA), including evaluating risks of FWA in each SBIR application.
Government fraud cost taxpayers $125 billion in 2014, according to The Washington Times. With an annual budget of over $1 billion, assuring the integrity of the SBIR program by protecting against FWA is vital to the Government, to taxpayers, and to the small business community. The potential for FWA in the form of duplication can occur when a proposed research effort is substantially similar to past work by another firm, or when a company proposes work and does not disclose that it was previously funded for the same effort. As a part the Missile Defense Agency’s (MDA) FWA prevention system, all SBIR applications are reviewed to detect potential duplications. This review calls for comparing new proposals with the large body of previously funded efforts. Previously, using manual analysis and automated systems, the MDA could retrieve individual files by unique identifiers or by keyword-searches; but it could not search by file content. The resulting process for comparing studies was tedious and error prone.
The DoD SBIR program alone has generated a 22:1 ROI on the DoD’s investment, $347 billion in total economic output, and the creation of more than 1.5 million jobs, according to a study by TechLink, a national DoD partnership intermediary at Montana State University-Bozeman, in collaboration with the Business Research Division of the Leeds School of Business at the University of Colorado in Boulder.
In addition to the obvious misuse of taxpayer funds, any instance of FWA casts a negative shadow over an otherwise popular and successful program that has delivered countless innovations to federal customers and benefited thousands of hardworking small businesses.
Of the tens of thousands of awards under the SBIR program, there were only 27 FWA convictions between 2013 and 2019. However, detractors of small business R&D programs latch onto these rare incidents of FWA to attack the overall integrity of the program. This places at risk the research mission and small business R&D goals of the eleven agencies that participate in the SBIR program, as well as the livelihoods of the innovative small businesses that win awards under the program.
Our Movia Analytics for Content Similarity (MACS) solution provides MDA’s SBIR program managers with a powerful machine learning algorithm that goes beyond keyword matching. MACS rapidly and automatically compares abstracts of new proposals with past awards, instantly identifies potential duplicates, and then ranks all matches by confidence. Instances of high-confidence duplication can be retrieved (within the MACS user-interface) for side-by-side comparison with the new proposal.
MACS provides an Amazon Web Services (AWS) hosted Software as a Service (SaaS) solution for comparing new research topic ideas or submitted proposal abstracts against the entire body of past contract awards. Technical capabilities include:
- Drag and drop new documents (research topics or submitted proposal abstracts) to queue up for analysis
- Correlate similarities of SBIR abstracts, topics, and proposals using the MACS deep-search capability that applies a machine learning algorithm
- Rank potentially similar past efforts and display in order of confidence
- Select and view high confidence matches. The MACS interface displays the new and old documents side-by-side and highlights sections with matching text or concepts.
MACS supports two primary FWA use cases.
- During research topic development, MACS compares draft topic ideas to the repository of historical contract awards and identifies potential existing solutions to a research need. The earlier research can be reviewed and potentially acquired in lieu of starting a new program, thus saving time and money.
- During evaluation of submitted proposals, MACS identifies submitted proposals that are substantially similar to previous awarded contracts, again avoiding investing in duplicative research, and in rare cases, detecting potential fraud.