US 12,271,919 B2
Performance and quality improvements for a market research platform
Nicolas Gauchat, Glen Ridge, NJ (US); Bryan Gregg Silverman, New York, NY (US); William Shawn Mansfield, Wilmington, NC (US); Matthew Britton, Brooklyn, NY (US); Albert Avi Savar, New York, NY (US); Chee Seng Sim, Brooklyn, NY (US); Kirby Beth Horwitz, Brooklyn, NY (US); Brit Carthy Hudgins, Exton, PA (US); Dennis Yip, Brooklyn, NY (US); McKenzie Ann Lawton, Brooklyn, NY (US); Alyssa Bender, Sylvan Springs, AL (US); and Erika Tierney Bailey, New York, NY (US)
Assigned to SUZY, INC., New York, NY (US)
Filed by SUZY, INC., New York, NY (US)
Filed on Aug. 17, 2021, as Appl. No. 17/404,162.
Claims priority of provisional application 63/153,991, filed on Feb. 26, 2021.
Prior Publication US 2022/0277326 A1, Sep. 1, 2022
Int. Cl. G06Q 30/02 (2023.01); G06N 20/00 (2019.01); G06Q 30/0203 (2023.01)
CPC G06Q 30/0203 (2013.01) [G06N 20/00 (2019.01)] 15 Claims
OG exemplary drawing
 
1. A method performed by processing circuitry of a computing system, the method comprising:
training a machine learning model (MLM) to perform a first filtering operation based on first usage data of a first plurality of survey respondents to a first survey and second usage data of a second plurality of survey respondents to a second survey, the first survey being associated with a first tenant of a multi-tenant survey platform and conforming to a survey template provided by the multi-tenant survey platform, and the second survey being associated with a second tenant of the multi-tenant survey platform and conforming to the survey template provided by the multi-tenant survey platform, wherein training the MLM to perform the first filtering operation includes inputting digital fingerprint attributes and behavioral metadata into the MLM over a plurality of training cycles to identify responses from bots, the behavioral metadata including a timestamp and the digital fingerprint attributes including one or more of:
(1) an IP address,
(2) an autonomous system number,
(3) a hosting provider, and
(4) an e-mail address;
receiving a first set of survey responses to the first and second surveys;
in response to determining that the MLM is not enabled, performing a second filtering operation on the first set of survey responses by a codified rule-based filter, the second filtering operation identifying a first subset of the first set of survey responses as being responses from bots;
detecting that a threshold condition has been met, the threshold condition being that a threshold number of training cycles have been performed;
in response to determining that the threshold condition has been met, enabling the MLM;
receiving a second set of survey responses to the first and second surveys;
in response to determining that the MLM has been enabled, performing the first filtering operation by the MLM on the second set of survey responses, the first filtering operation identifying a second subset of the second set of survey responses as being responses from bots; and
automatically, via the processing circuitry, restricting, in real-time, usage of the first and second subsets of the computing system by users associated with the survey responses, including quarantining users associated with the survey responses of the first and second subsets to limit access by those users of the computing system for a period of time;
wherein quarantining users associated with the survey responses of the first and second subsets to limit access by those users of the computing system includes one of:
preventing those users from responding to any questions and
preventing those users from responding to particular types of questions.