US 11,900,071 B2
Generating customized digital documents using artificial intelligence
Arindam Paul, Bangalore (IN); Angela Kontos, Malden, MA (US); Rachna Saxena, Bangalore (IN); Santhosh Kolloju, Suryapet (IN); Arijit Saha, Bangalore (IN); Aaditya Mathur, Bangalore (IN); Pavan Mohan, Rajarajeshwari Nagar (IN); and Mohamed Asif Khan, Bangalore (IN)
Assigned to FMR LLC, Boston, MA (US)
Filed by FMR LLC, Boston, MA (US)
Filed on May 28, 2021, as Appl. No. 17/333,387.
Claims priority of provisional application 63/032,390, filed on May 29, 2020.
Prior Publication US 2021/0374360 A1, Dec. 2, 2021
Int. Cl. G06F 40/56 (2020.01); G06N 3/08 (2023.01); G06F 40/284 (2020.01)
CPC G06F 40/56 (2020.01) [G06F 40/284 (2020.01); G06N 3/08 (2013.01)] 24 Claims
OG exemplary drawing
 
1. A system used in a computing environment in which unstructured computer text is analyzed for generation of customized digital documents, the system comprising:
a computer data store including (i) a plurality of historical user interactions each associated with a user, each historical user interaction comprising a plurality of data fields, and (ii) a plurality of historical digital documents corresponding to the plurality of historical user interactions, each historical digital document comprising a corpus of unstructured computer text, and
a server computing device in communication with the computer data store, the server computing device comprising a memory to store computer-executable instructions and a processor that executes the computer-executable instructions to:
tokenize each historical user interaction and each historical digital document into a set of tokens using a byte pair encoder;
encode each set of tokens for the historical user interaction and the historical digital document into a multidimensional vector;
train an interaction classification model using the multidimensional vectors as input, the trained interaction classification model configured to generate a digital document classification for an input user interaction;
train a language generation model using the multidimensional vectors as input, the trained language generation model configured to generate a customized digital document based upon an input user interaction;
receive a new user interaction associated with a user of a client computing device;
tokenize the new user interaction into a new set of tokens using a byte-pair encoder and encode the new set of tokens into a new multidimensional vector;
execute the trained interaction classification model using the new multidimensional vector as input to generate a digital document classification for the new multidimensional vector;
execute the trained language generation model using the new multidimensional vector and the digital document classification for the new multidimensional vector as input to generate a customized digital document for the user of the client computing device; and
transmit the customized digital document to the client computing device for display to the user of the client computing device.