Description:
The Federal Motor Carrier Safety Administration (FMCSA) is responsible for regulating the safety of interstate truck and bus travel in the United States. The primary mission of FMCSA is to reduce crashes, injuries and fatalities involving large trucks and buses.FMCSA’s strategic framework is built upon three core principles:
- Raise the bar to enter the industry;
- Require operators to maintain high safety standards to remain in the industry; and
- Remove high-risk operators from our roads and highways.
The vetting process implemented within the FMCSA’s Office of Registration and Safety Information supports all of these initiatives by assuring that new applicants meet FMCSA’s standards for fitness, willingness, and ability to comply with all applicable federal statutes and regulations by checking for signs that a new applicant is not a reincarnated version of an existing high-risk operator. These initiatives set a high bar to obtain operating authority and close loopholes for those high risk operators to reincarnate themselves with a clean slate and, hence, keep them off public highways.
FMCSA already employs a proprietary risk-based screening process which uses a sophisticated matching algorithm to screen and assign risk to an applicant using primarily federal sources of data. This solicitation is seeking innovative approaches, alternate methods and public/private data sources to confirm or further expand robust automation methods that are part of its screening process.
The primary purpose of this topic is for the Offeror to use operating authority application information specified on the application form (See References 1 and 2) and compare it to the similar information on file for a list of motor carriers and identify the probability of potential affiliation between the applicant and each of the carriers of interest (i.e. development of a robust affiliation strength model with use of publicly available data sources).
FMCSA is primarily interested in
- Surveying of publicly available data sources (such as States’ data) that can be automatically cross-checked against that can validate submitted information or hint for potential affiliations;
- Surveying of affiliation strength/risk models that may be used in other business models or by other Federal or State Agencies;
- Identification of private data that could provide incremental benefits;
- Development and use of complex matching algorithms that may take into account typos, different abbreviations, use of short names, text order differences;
- Confirmation of application data validity to the extent possible such as business address;
- Use of web-search algorithms that can be automatically assimilated into useful measures;
- Development and use of probability measures for assessing affiliation strength; and
- Development of a self-learning framework and adaptive methods to automatically update the model parameters based on application disposition decisions.
The Contractor will be required to sign a non-disclosure agreement to receive sample data which can be used to develop and test out proposed methods. There are about 50,000 applications per year, each of which would need to be automatically processed for affiliation strength assessment with respect to a list of other motor carriers of interest which may be a subset of the ~725,000 motor carriers to be specified by FMCSA. Each application would not need to be checked against all motor carriers of interest and the Offeror would have latitude to further scope down the screening methodologies intelligently based on the research conducted within this project.
The entire solution would need to be fully automated. It would need to input a set of text fields from an applicant and a set of text fields from an existing company and use the underlying company information and the identified public sources of information to output a probability measure of affiliation strength between the two companies. The algorithm must run reasonably fast such that one application can be batch processed against a large number of potential other companies and the entire automatic assessment process can be completed in reasonable time (reasonable level to be defined jointly between the Contractor and FMCSA during Phase I).
Expected Phase I Outcomes:
Outcomes expected from the Phase 1 include surveying and documentation of all available public and private data sources and uses of other affiliation strength/risk models. In addition, a detailed concept that demonstrates the viability of developing complex affiliation risk model that would work within the context of FMCSA’s needs is expected to be delivered. Computational needs and processing time assessments will have to be quantified. Expected ranges of effectiveness measures would need to be developed.
Expected Phase II Outcomes:
Phase 2 efforts would prototype the Contractor’s approach to validate the affiliation risk model. Furthermore, a detailed experimental plan for assessing the efficacy of the solution would be formulated along with updated cost-benefit projections based on development activities.
