US 11,704,499 B2
Generating questions using a resource-efficient neural network
Shubham Agrawal, Bellevue, WA (US); Owais Khan Mohammed, Bellevue, WA (US); and Weiming Wen, Bellevue, WA (US)
Assigned to Microsoft Technology Licensing, LLC, Redmond, WA (US)
Filed by Microsoft Technology Licensing, LLC, Redmond, WA (US)
Filed on Apr. 13, 2021, as Appl. No. 17/228,728.
Prior Publication US 2022/0327287 A1, Oct. 13, 2022
Int. Cl. G06F 40/30 (2020.01); G06F 16/93 (2019.01); G06F 40/126 (2020.01); G06F 40/284 (2020.01); G06N 3/08 (2023.01); G06N 3/045 (2023.01)
CPC G06F 40/30 (2020.01) [G06F 16/93 (2019.01); G06F 40/126 (2020.01); G06F 40/284 (2020.01); G06N 3/045 (2023.01); G06N 3/08 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A method for generating questions using a neural network, comprising:
obtaining an electronic document having a plurality of textual passages;
selecting, using a first neural network, a subset of textual passages for use in generating questions regarding the electronic document;
generating, using a second neural network, one or more candidate answers for each textual passage selected by the first neural network, to produce a plurality of candidate passage-answer pairs;
selecting, using a third neural network, a subset of the plurality of candidate passage-answer pairs produced by the second neural network; and
generating an output result to provide to a user that includes one or more output questions chosen from the subset of the plurality of candidate passage-answer pairs identified by the third neural network,
the use of the first neural network reducing processing load placed on the second neural network and the third neural network by reducing an amount of data to be processed by the second neural network and the third neural network.