Systems and methods for generating dynamic conversational responses based on predicted user intents using artificial intelligence models
AssigneeCapital One Services, LLC
InventorsMinh Le, Arturo Hernandez Zeledon, Md Arafat Hossain Khan
Life of this patent
Estimating the term…
Abstract
Described are methods and systems are for generating dynamic conversational queries. For example, as opposed to being a simply reactive system, the methods and systems herein provide means for actively determining a user's intent and generating a dynamic query based on the determined user intent. Moreover, these methods and systems generate these queries in a conversational environment.
Claims20
A system for generating dynamic conversational responses based on predicted user intents using artificial intelligence models that compensate for training data featuring a high proportion of missing data points, the system comprising:
- cloud-based storage circuitry configured to store an artificial intelligence model, wherein the artificial intelligence model comprises a neural network that is trained to predict user intent based on a first feature input and a second feature input, wherein the first feature input is based on an embedding of a first collection of user record data for a plurality of users and for a plurality of domains, and wherein the second feature input is based on a subset of the first collection comprising a group of domains selected from the plurality of domains;
- cloud-based control circuitry configured to:
- receive first user record data, wherein the first user record data comprises user transaction data comprising a plurality of fields related to an electronic communication of a user;
- generate a user feature input based on the first user record data, wherein the user feature input comprises a vectorized representation of the plurality of fields related to the electronic communication;
- input the user feature input into the artificial intelligence model;
- receive an output from the artificial intelligence model, wherein the output corresponds to a predicted intent of the user;
- and
- cloud-based input/output circuitry configured to generate for display, on a user interface, a dynamic conversational response based on the output, wherein the dynamic conversational response comprises a text-string message based on the predicted intent of the user.
A method for generating dynamic conversational responses based on predicted user intents using artificial intelligence models, the method comprising:
- receiving first user record data;
- inputting the first user record data into an artificial intelligence model, wherein the artificial intelligence model comprises a neural network that is trained to predict user intent based on a first feature input and a second feature input, wherein the first feature input is based on an embedding of a first collection of user record data for a plurality of users and for a plurality of domains, and wherein the second feature input is based on a subset of the first collection comprising a group of domains selected from the plurality of domains;
- receiving an output from the artificial intelligence model;
- and
- generating for display, in a user interface, a dynamic conversational response based on the output.
A non-transitory, computer readable medium for generating dynamic conversational responses based on predicted user intents using artificial intelligence models comprising instructions that when executed by one or more processors, cause operations comprising:
- receiving first user record data;
- inputting the first user record data into an artificial intelligence model, wherein the artificial intelligence model comprises a neural network that is trained to predict user intent based on a first feature input and a second feature input, wherein the first feature input is based on an embedding of a first collection of user record data for a plurality of users and for a plurality of domains, and wherein the second feature input is based on a subset of the first collection comprising a group of domains selected from the plurality of domains;
- receiving an output from the artificial intelligence model;
- and
- generating for display, in a user interface, a dynamic conversational response based on the output.
Dependencies are inferred from the claim text.