US 11,868,879 B2
United states patent application for neural network systems and methods for email parameter extraction
Eric Glyman, Brooklyn, NY (US); William James Russell Locke, Old Westbury, NY (US); Kathleen Zasada, Brooklyn, NY (US); Philippe Tyan, New York, NY (US); Jae In Lee, Brooklyn, NY (US); and Karim Atiyeh, Brooklyn, NY (US)
Assigned to Capital One Services, LLC, McLean, VA (US)
Filed by Capital One Services, LLC, McLean, VA (US)
Filed on Jun. 15, 2018, as Appl. No. 16/010,354.
Application 16/010,354 is a continuation of application No. 15/669,874, filed on Aug. 4, 2017, granted, now 10,318,865.
Claims priority of provisional application 62/528,030, filed on Jun. 30, 2017.
Prior Publication US 2019/0005389 A1, Jan. 3, 2019
Int. Cl. G06N 3/08 (2023.01); G06Q 10/02 (2012.01); G06F 9/455 (2018.01); G06F 9/445 (2018.01); G06N 3/045 (2023.01)
CPC G06N 3/08 (2013.01) [G06F 9/44505 (2013.01); G06F 9/45558 (2013.01); G06N 3/045 (2023.01); G06Q 10/02 (2013.01); G06F 2009/45562 (2013.01); G06F 2009/45579 (2013.01); G06F 2009/45595 (2013.01)] 11 Claims
OG exemplary drawing
 
1. An email parsing system, comprising:
at least one memory storing instructions; and
at least one processor operatively connected to the at least one memory and configured to execute the instructions to perform operations, including:
automatically monitoring an email address for receipt of a reservation confirmation email;
in response to identifying receipt of the reservation confirmation email based on the monitoring of the email address:
filtering computer code included in the reservation confirmation email to identify a portion of the reservation confirmation email likely to correspond to one or more parameters of a reservation associated with the reservation confirmation email;
converting the identified portion of the reservation confirmation email into one or more feature vectors; and
determining the one or more parameters for the reservation as well as a respective value for each of the one or more parameters by inputting the one or more feature vectors into a trained neural network;
monitoring an inventory for an alternative reservation that is equivalent to the reservation and that has a lower price than the reservation, wherein determining that the alternative reservation is equivalent to the reservation includes:
extracting one or more inventory parameters associated with the alternative reservation and a respective value for each parameter;
converting the one or more inventory parameters and the respective value for each parameter into one or more inventory feature vectors; and
performing a vector comparison between the one or more feature vectors and the one or more inventory feature vectors to determine a similarity value, the similarity value being above a predetermined threshold indicating that the alternative reservation is equivalent to the reservation;
in response to detecting, via the monitoring, the alternative reservation equivalent to the reservation but with a lower price:
generating, based on the determined one or more parameters and respective value for each parameter associated with the alternative reservation, a request for the alternative reservation that is executable to cause a booking entity associated with the alternative reservation to automatically book the alternative reservation; and
executing the request, such that the alternative reservation is automatically booked; and
in response to receiving a confirmation of the booking of the alternative reservation from the booking entity associated with the alternative reservation:
generating, based on the determined one or more parameters and respective value for each parameter associated with the reservation, a cancellation request that is executable to cause a booking entity associated with the reservation; and
executing the cancellation request, such that the reservation is automatically cancelled.