US 11,736,421 B1
Artificial intelligence (AI)-powered conversational messaging within a mobile application
Ramakrishna R. Yannam, Plano, TX (US); Prejish Thomas, Plano, TX (US); Steven Zhao, Plano, TX (US); Saahithi Chillara, Plano, TX (US); Rajan Jhaveri, Plano, TX (US); Ryan Strug, Plano, TX (US); Kurt R. Schultz, Westlake Village, CA (US); and Priyank Shah, Plano, TX (US)
Assigned to Bank of America Corporation, Charlotte, NC (US)
Filed by Bank of America Corporation, Charlotte, NC (US)
Filed on Apr. 18, 2022, as Appl. No. 17/722,889.
Int. Cl. H04L 51/02 (2022.01); H04L 51/216 (2022.01); G06N 3/08 (2023.01); G06N 3/04 (2023.01); H04L 51/224 (2022.01)
CPC H04L 51/02 (2013.01) [G06N 3/04 (2013.01); G06N 3/08 (2013.01); H04L 51/216 (2022.05); H04L 51/224 (2022.05)] 20 Claims
OG exemplary drawing
 
1. A method for condensing user communications relating to a first topic, the method for use with a system, the system comprising a processor and a non-transitory memory with instructions stored thereon, wherein the instructions upon execution by the processor, cause the processor to execute the method, said method comprising:
designating a topic of user interest, said topic of user interest being designated by receiving a selection of one of the 100 most common nouns appearing in a database;
retrieving in real-time legacy e-mail communications stored in the database regarding the topic of user interest;
determining, using the processor, whether duplicative e-mail communications are included among the retrieved legacy e-mail communications and, to the extent that duplicative e-mail communications are included in the retrieved legacy e-mail communications, removing said duplicative e-mail communications from the retrieved legacy e-mail communications;
retrieving in real-time legacy intelligence stored in the database relating to historical user selections regarding the topic of user interest;
retrieving in real-time a plurality of outcomes stored in the database based on the legacy intelligence;
with respect to the topic of user interest, forming, said forming comprising mining information that relates to the retrieved legacy e-mail communications, the retrieved legacy intelligence, and the retrieved plurality of outcomes, a topic-centric training set for a neural network, said topic-centric training set being based on the retrieved legacy e-mail communications, the retrieved legacy intelligence, and the retrieved plurality of outcomes, said topic-centric training set being delimited by an analysis of the database, said neural network comprising a plurality of nodes;
synthesizing the neural network, said synthesizing using the topic-centric training set to assign individual weights to each of the plurality of nodes based on the retrieved legacy e-mail communications, the retrieved legacy intelligence, and the retrieved plurality of outcomes;
in response to a selection of the topic of user interest, generating, for display to a user, a plurality of topic-related user options based on the neural network; and
prompting the user to select one of the plurality of topic-related user options.