US 12,292,908 B1
Attribute extraction from text sources
Ankush Chopra, Bangalore (IN); Aravind Chandramouli, Bangalore (IN); Mohammed Sameer Khan, Pune (IN); Paarth Ajay Shrivastava, Bangalore (IN); Shubham Sharma, Jalandhar (IN); and Soumendra Mohanty, Bangalore (IN)
Assigned to Tredence, Inc., San Jose, CA (US)
Filed by Tredence Inc., San Jose, CA (US)
Filed on Dec. 12, 2024, as Appl. No. 18/978,862.
Int. Cl. G06F 16/3329 (2025.01); G06F 9/54 (2006.01); G06F 16/334 (2025.01); G06F 16/338 (2019.01); G06F 40/58 (2020.01); G06F 40/284 (2020.01)
CPC G06F 16/3329 (2019.01) [G06F 9/54 (2013.01); G06F 16/3344 (2019.01); G06F 16/338 (2019.01); G06F 40/58 (2020.01); G06F 40/284 (2020.01)] 20 Claims
OG exemplary drawing
 
1. A method implemented by a computer system hosting a plurality of artificial intelligence (AI) models, the method comprising:
causing a first user interface to be presented on a device;
receiving, by the computer system and from the device, a first input of (i) an arbitrary language and (ii) arbitrary values representing an attribute of the first input;
generating, by at least using a first application programming interface (API) for a language detection model of the plurality of AI models, a set of predicted languages by using the first input as input to the language detection model;
generating, by at least using respective APIs for different translation models of the plurality of AI models, a different language translation of the first input for each language in the set of predicted languages by using the first input as input to the different translation models;
generating for each different language translation, by at least using a same second API for a same first question-answer model of the plurality of AI models, a respective answer based at least in part on a same input question prompt and the respective language translation and creating a set of answers that include each respective answer;
storing the set of answers in a database;
generating, by at least using the same second API for the same first question-answer model, an additional answer based at least in part on the same input question prompt and the first input;
updating the set of answers in the database to include the additional answer;
performing a distribution-based correction, by at least using a distribution source and at least one answer included in the set of answers, to determine the attribute represented by the arbitrary values included in the first input; and
causing, based at least in part on the distribution-based correction, the device or a different device to present a second user interface that presents a recommendation based at least in part on the attribute represented by the arbitrary values included in the first input.