Real time adaptive control of transaction review rate score curve
OwnerHung-Tzaw Hu, Jason Eaton +2inventors
InventorsHung-Tzaw Hu, Jason Eaton, Scott Boding, Jianji Ma
Life of this patent
Estimating the term…
Abstract
Modern technologies of signal processing and real-time feedback adaptive control are applied to a fraud screen model to enable the fraud screen model to adapt to changes in shopping behavior to effectively maintain the model's performance on a review rate score curve continuously. Multiple techniques disclosed herein integrate control technologies with modeling technology to control the review rate score curve in real time.
Claims18
A computer-implemented method for automatically adjusting a review rate score curve for detecting fraudulent transactions, the review rate score curve having a plurality of inflection points, the method comprises:
- performing, by a computer system, for each of the inflection points:
- receiving a time series representing a set of review rates at the inflection point over a predetermined duration;
- generating a set of output data comprising a trend review rate output, an integral delta output, and an derivative delta output based at least on the time series for the inflection point;
- and
- generating an updated inflection point based at least on the set of output data;
- and
- generating an updated review rate score curve based at least on the updated inflection points.
A system comprising:
- one or more processors;
- and
- one or more memories coupled to the one or more processors, the one or more memories storing machine executable code, which when executed by the one or more processors causes the one or more processors to perform a method comprising:
- for each of a plurality of inflection points in a review rate score curve:
- receiving a time series representing a set of review rates at the inflection point over a predetermined duration;
- generating a set of output data comprising a trend review rate output, an integral delta output, and an derivative delta output based at least on the time series for the inflection point;
- and
- generating an updated inflection point based at least on the set of output data;
- and
- generating an updated review rate score curve based at least on the updated inflection points.
Dependencies are inferred from the claim text.