US 11,776,291 B1
Document analysis architecture
Samuel Cameron Fleming, Spokane, WA (US); David Craig Andrews, Carnation, WA (US); John E. Bradley, III, Duvall, WA (US); Lewis C. Lee, Atherton, CA (US); Jared Dirk Sol, Spokane, WA (US); Timothy Seegan, Spokane, WA (US); and Scott Buzan, Spokane Valley, WA (US)
Assigned to AON RISK SERVICES, INC. OF MARYLAND, New York, NY (US)
Filed by AON RISK SERVICES, INC. OF MARYLAND, New York, NY (US)
Filed on Jun. 10, 2020, as Appl. No. 16/897,606.
Int. Cl. G06V 30/414 (2022.01); G06N 3/08 (2023.01); G06F 40/284 (2020.01); G06V 30/416 (2022.01); G06F 18/214 (2023.01); G06F 18/2413 (2023.01)
CPC G06V 30/414 (2022.01) [G06F 18/214 (2023.01); G06F 18/2413 (2023.01); G06F 40/284 (2020.01); G06N 3/08 (2013.01); G06V 30/416 (2022.01)] 18 Claims
OG exemplary drawing
 
1. A method, comprising:
generating machine-learning models configured to identify a document as in class or out of class, the machine-learning models trained utilizing user input data indicating a first portion of documents as in class and a second portion of documents as out of class, wherein the documents are intellectual property documents;
determining, for individual ones of the machine-learning models, a category associated with individual ones of the machine-learning models, the category associated with a classification system associated with the documents;
generating a taxonomy of the machine-learning models, the taxonomy indicating categorical relationships between the machine-learning models, wherein generating the taxonomy is based at least in part on the category associated with the individual ones of the machine-learning models;
receiving, for a machine-learning model of the machine-learning models, a training dataset configured to train the machine-learning model to determine which of the documents are in class to the machine-learning model;
receiving an indication that a portion of the training dataset includes confidential information;
generating a modified machine-learning model corresponding to the machine-learning model trained without the portion of the training dataset that includes the confidential information; and
wherein generating the taxonomy comprises generating the taxonomy based at least in part on the modified machine-learning model.