US 11,915,157 B2
Computerized method of training a computer executed model for recognizing numerical quantities
Jamal Zabihi, Toronto (CA); and Alexander Karl Hudek, Toronto (CA)
Assigned to KIRA INC., Toronto (CA); and ZUVA INC., Toronto (CA)
Filed by KIRA INC., Toronto (CA); and Zuva, Inc., Toronto (CA)
Filed on Oct. 21, 2022, as Appl. No. 17/971,525.
Application 17/971,525 is a continuation of application No. 16/883,183, filed on May 26, 2020, granted, now 11,507,864.
Prior Publication US 2023/0040388 A1, Feb. 9, 2023
This patent is subject to a terminal disclaimer.
Int. Cl. G06F 7/00 (2006.01); G06N 5/04 (2023.01); G06N 20/00 (2019.01)
CPC G06N 5/04 (2013.01) [G06N 20/00 (2019.01)] 14 Claims
OG exemplary drawing
 
1. A computerized method of training a computer executed model for recognizing at least one numerical quantity, the method carried out by one of more processors, the method comprising:
receiving, as input, at least one unit expression;
searching for numeric values and the at least one unit expression in a text corpus, the text corpus comprising sets of words and frequency of occurrence of each of the sets, the search resulting in identification of sets that comprise a combination of a numeric value and the at least one unit expression;
generating sentences from the text corpus by applying the identified sets as input;
generating a training dataset by auto labelling the identified sets within the generated sentences based on the at least one numerical quantity; and
training the model by providing input based on the training dataset.