US 12,450,569 B2
Data ingestion and understanding for natural language processing systems
Luu Tran, Sunnyvale, CA (US); Omer Luzzatti, Portola Valley, CA (US); Nathanael Joe Hayashi, Piedmont, CA (US); and Abhinav Athreya, Dublin, CA (US)
Assigned to Amazon Technologies, Inc., Seattle, WA (US)
Filed by Amazon Technologies, Inc., Seattle, WA (US)
Filed on Jun. 28, 2023, as Appl. No. 18/342,908.
Application 18/342,908 is a continuation of application No. 16/836,027, filed on Mar. 31, 2020, granted, now 11,748,713.
Prior Publication US 2024/0062164 A1, Feb. 22, 2024
This patent is subject to a terminal disclaimer.
Int. Cl. G10L 15/22 (2006.01); G06N 5/02 (2023.01); G06Q 10/1093 (2023.01); G10L 13/02 (2013.01); G10L 15/08 (2006.01)
CPC G06Q 10/1093 (2013.01) [G06N 5/02 (2013.01); G10L 13/02 (2013.01); G10L 15/08 (2013.01); G10L 2015/088 (2013.01)] 18 Claims
OG exemplary drawing
 
1. A computer-implemented method, comprising:
receiving first data representing a first natural language input provided to a user input component, wherein the first natural language input is associated with a first profile;
determining that the first data corresponds to a first request to perform a first action;
causing a first software application component to perform the first action so that at least one computing device associated with the first profile provides at least one of a first audible or a first visible output corresponding to the first action;
generating first activity data corresponding to the first action; and
after the first software application component has performed the first action:
processing the first activity data using at least one trained machine learning natural language processing component to determine that a first element represented in the first activity data is associated with a first category of a plurality of categories;
determining second data corresponding to the first action, the second data associating the first element and the first category;
storing the second data, in association with the first profile, in a data storage component;
receiving third data representing a second natural language input, wherein the third data is associated with the first profile;
based at least in part on the third data being associated with the first profile, retrieving the second data from the data storage component;
processing the third data and the second data retrieved from the data storage component to determine a second action to perform in response to the second natural language input; and
causing a second software application component to perform the second action so that at least one computing device associated with the first profile provides at least one of a second audible or a second visible output corresponding to the second action.