US 12,242,905 B2
Automatic application program interface (API) selector for unsupervised natural language processing (NLP) intent classification
Alston Ghafourifar, Los Altos Hills, CA (US); Mehdi Ghafourifar, Los Altos Hills, CA (US); and Brienne Ghafourifar, Los Altos Hills, CA (US)
Assigned to Entefy Inc., Palo Alto, CA (US)
Filed by Entefy Inc., Palo Alto, CA (US)
Filed on Feb. 27, 2024, as Appl. No. 18/589,285.
Application 18/589,285 is a continuation of application No. 16/889,613, filed on Jun. 1, 2020, granted, now 11,948,023.
Application 16/889,613 is a continuation of application No. 15/859,183, filed on Dec. 29, 2017, abandoned.
Prior Publication US 2024/0281312 A1, Aug. 22, 2024
This patent is subject to a terminal disclaimer.
Int. Cl. G06F 9/54 (2006.01); G06F 9/451 (2018.01); G06F 40/30 (2020.01)
CPC G06F 9/547 (2013.01) [G06F 9/453 (2018.02); G06F 9/543 (2013.01); G06F 40/30 (2020.01)] 14 Claims
OG exemplary drawing
 
1. A computer-implemented method, comprising:
monitoring application program interface (API) calls resulting from interactions between two or more APIs each corresponding to one of an application layer of an application on a plurality of devices and an operating system layer of the plurality of devices;
determining API call sequences for the interactions from the monitored API calls;
accessing a graphical representation of relationships between the two or more APIs, wherein the relationships between the two or more APIs are associated with call paths executable to respond to one of a user command or a user query for a natural language processing (NLP) intent classification system of the application;
adjusting, for the graphical representation, the call paths associated with the relationships between the two or more APIs in the graphical representation, wherein the adjusting comprises:
generating a potential call path using an artificial intelligence (AI) model based on the monitored API calls and the API call sequences,
executing a test program of the potential call path, and
determining, based on the executing, whether the potential call path results in a finished sequence of events or an error condition for one of the user command or the user query;
updating the relationships in the graphical representation to include the potential call path with the call paths;
configuring the NLP intent classification system to execute the potential call path to respond to the one of the user command or the user query;
executing one of the API call sequences for the potential call path;
determining a response to the user command or the user query based on the executing the one of the API call sequences, wherein the response includes at least a portion of the graphical representation and corresponding path metrics for the API calls in the one of the API call sequences corresponding to the potential call path;
storing the response that enables responding to the user command or the user query on demand when received; and
providing the response to an end user device based on the user command or the user request.