Systems and methods for providing automated natural language dialogue with customers

• US 10,332,505 B2
• Filed: 08/01/2017
• Issued: 06/25/2019
• Est. Priority Date: 03/09/2017
• Status: Active Grant

• Current Assignee: Capital One Services LLC (Capital One Financial Corporation)
• Original Assignee: Capital One Services LLC (Capital One Financial Corporation)
• Inventors: Zoller, Gregory W., Karp, Scott, Jacob, Sujay Eliphaz, Mueller, Erik, Hay, Stephanie, Paynter, Adam Roy
• Primary Examiner(s): Colucci, Michael C
• Application Number:
  US15/665,960
• Publication Number:
  US 20180261203A1
• Time in Patent Office: 693 Days
• Field of Search: 704 9, 704235, 704233, 704254, 704244, 7042701, 333 9, 705 41, 705 1417, 379 8818
• CPC Class Codes: G06F 40/284 Lexical analysis, e.g. tokenisation or collocates
• CPC Class Codes: G06F 40/35 Discourse or dialogue representation
• CPC Class Codes: G06F 9/542 Event management; Broadcasting; Multicasting; Notifications
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 5/046 Forward inferencing; Production systems
• CPC Class Codes: G06Q 30/016 After-sales
• CPC Class Codes: G06Q 50/01 Social networking
• CPC Class Codes: G10L 13/00 Speech synthesis; Text to speech systems
• CPC Class Codes: G10L 13/027 Concept to speech synthesisers; Generation of natural phrases from machine-based concepts generation of parameters for speech synthesis out of text G10L13/08
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: G10L 15/26 Speech to text systems G10L15/08 takes precedence
• CPC Class Codes: G10L 2015/223 Execution procedure of a spoken command
• CPC Class Codes: H04L 51/02 using automatic reactions or user delegation, e.g. automatic replies or chatbot-generated messages

Abstract

A system includes one or more memory devices storing instructions, and one or more processors configured to execute the instructions to perform steps of a providing automated natural dialog with a customer. The system may generate one or more events and commands temporarily stored in queues to be processed by one or more of a dialog management device, an API server, and an NLP device. The dialog management device may create adaptive responses to customer communications using a customer context, a rules-based platform, and a trained-machine learning model.

Claims

1. A system for automating natural language dialogue with a customer, comprising:

one or more processors; and

memory in communication with the one or more processors and storing instructions that, when executed by the one or more processors, are configured to cause the system to;

responsive to receiving an incoming customer dialogue message in an automated customer service interaction, generate a first event to be placed in an event queue, the event queue being monitored by a dialogue management device, the dialogue management device comprising;

a rules-based platform;

a trained machine learning model; and

a customer context derived from financial customer information comprising customer conversation history;

responsive to detecting the first event in the event queue, receive, at the dialogue management device, the first event from the event queue;

responsive to processing the first event, generate, by the dialogue management device and based on one or more of the rules-based platform, the trained machine learning model, and the customer context, a first command to be placed in a command queue, the first command representing a command to the natural language processing device to determine the meaning of the incoming customer dialogue message based on the customer context, wherein the command queue comprises one or more commands for execution by one or more of a natural language processing device, an API server, and a communication interface;

responsive to the execution of the first command by the natural language processing device, generate a second event to be placed in the event queue, the second event representing a determined meaning of the incoming customer dialogue message;

responsive to detecting the second event in the event queue, receive, at the dialogue management device, the second event from the event queue;

responsive to processing the second event, generate, by the dialogue management device and based on one or more of the rules-based platform, the trained machine learning model, and the customer context, a second command to be placed in the command queue, the second command being a command to the API server to retrieve customer data, wherein customer data includes opt-out status and credential authentication;

responsive to the execution of the second command by the API server, generate a third event to be placed in the event queue, the third event representing retrieved customer data;

responsive to detecting the third event in the event queue, receive, at the dialogue management device, the third event from the event queue; and

responsive to processing the third event, generate, by the dialogue management device and based on one or more of the rules-based platform, the trained machine learning model, and the customer context, a response dialogue message and a third command to be placed in the command queue, the third command providing an instruction to the communication interface to transmit the response dialogue message, wherein the response dialogue message represents a conversational response to the incoming customer dialogue message based at least in part on the retrieved customer data as processed by the dialogue management device.

2. The system of claim 1, wherein the response dialogue message comprises the retrieved customer data.

3. The system of claim 1 wherein the first event further comprises an indication that the incoming customer dialogue message was received via one of the following types of channels:

SMS, an instant messaging program, a website based chat program, a mobile application, a voice-to-text device, or an email.

4. The system of claim 3 wherein the dialogue management device generates the response dialogue message based on the type of channel the incoming dialogue message was received on.

5. The system of claim 1 wherein the customer context is further derived from customer information associated with a particular customer stored in a database, the customer information comprising one or more of:

account types, account statuses, transaction history, people models, an estimate of customer sentiment, customer goals, and customer social media information.

6. The system of claim 5 wherein the customer context is updated each time the dialogue management device receives an event.

7. A system for automating natural language dialogue with a customer, comprising:

one or more processors; and

memory in communication with the one or more processors and storing instructions that, when executed by the one or more processors, are configured to cause the system to;

responsive to receiving an incoming customer dialogue message in an automated customer service interaction, generate a first event to be placed in an event queue, the event queue being monitored by a dialogue management device, the dialogue management device comprising;

a rules-based platform;

a trained machine learning model; and

a customer context derived from financial customer information comprising customer conversation history;

responsive to detecting the first event in the event queue, receive, at the dialogue management device, the first event from the event queue, wherein the first event comprises an indication of a communication channel that was used to deliver the incoming customer dialogue message, the communication channel comprising one of the following types of channels;

SMS, an instant messaging program, a website based chat program, a mobile application, a voice-to-text device, or an email;

responsive to processing the first event, generate, by the dialogue management device and based on one or more of the rules-based platform, the trained machine learning model, and the customer context, a first command to be placed in a command queue, wherein the command queue comprises one or more commands for execution by one or more of a natural language processing device, an API server, and a communication interface;

responsive to receiving the first command, determining the meaning of the incoming customer dialogue message based on the customer context;

responsive to execution of the first command by one of a natural language processing device or the API server, generate a second event to be placed in the event queue;

responsive to detecting the second event in the event queue, receive, at the dialogue management device, the second event from the event queue; and

responsive to processing the second event and based on the indication of the communication channel, generate, by the dialogue management device and based on one or more of the rules-based platform, the trained machine learning model, and the customer context, a response dialogue message and a second command to be placed in the command queue, the second command providing an instruction to the communication interface to transmit the response dialogue message via the communication channel, wherein the response dialogue message represents a conversational response to the incoming customer dialogue message.

8. The system of claim 7, wherein the first event comprises data representative of an incoming customer dialogue message.

9. The system of claim 8, wherein the first command comprises a command to the natural language processing device to determine the meaning of the incoming customer dialogue message.

10. The system of claim 7, wherein the first command is one of a command to retrieve customer data, a command to perform an account action, a command to authenticate a customer, or an opt-in/op-out command.

11. The system of claim 7 wherein the customer context is further derived from customer information associated with a particular customer stored in a database, the customer information comprising one or more of:

account types, account statuses, transaction history, people models, an estimate of customer sentiment, customer goals, and customer social media information.

12. The system of claim 11 wherein the customer context is updated each time the dialogue management device receives an event.

13. The system of claim 12 wherein the customer context is updated by receiving updated customer information from the database.

14. The system of claim 12, further comprising outputting, from the dialogue management device to the database following the generation of each command, a record of a customer interaction relating to the generated command to the database.

15. The system of claim 7, wherein the natural language processing device is capable of determining the meaning of an incoming customer dialogue message by utilizing one or more of the following artificial intelligence techniques:

intent classification, named entity recognition, sentiment analysis, relation extraction, semantic role labeling, question analysis, rule extraction and discovery, and story understanding.

16. The system of claim 7, wherein the natural language processing device is capable of performing natural language generation by utilizing one or more of the following artificial intelligence techniques:

content determination, discourse structuring, referring expression generation, lexicalization, linguistic realization, explanation generation.

17. A method for providing automated natural language dialogue with a customer, comprising:

receiving a first event to be placed in an event queue, the event queue being monitored by a dialogue management device in an automated customer service interaction, the dialogue management device comprising;

a rules-based platform;

a trained machine learning model; and

a customer context derived from financial customer information comprising customer conversation history;

responsive to detecting the first event in the event queue, receiving, at the dialogue management device, the first event from the event queue;

responsive to processing the first event, generating, by the dialogue management device and based on one or more of the rules-based platform, the trained machine learning model, and the customer context, a first command to be placed in a command queue, wherein the command queue comprises one or more commands for execution by one or more of a natural language processing device, an API server, and a communication interface, wherein the natural language processing device, the API server, and the communication interface are configured to operate independently from one another by separately pulling commands from the command queue;

responsive to receiving the first command, determining the meaning of the incoming customer dialogue message based on the customer context;

responsive to execution of the first command by one of a natural language processing device or the API server, generating a second event to be placed in the event queue;

responsive to detecting the second event in the event queue, receiving, at the dialogue management device, the second event from the event queue; and

responsive to processing the second event, generating, by the dialogue management device and based on one or more of the rules-based platform, the trained machine learning model, and the customer context, a response dialogue message and a second command to be placed in the command queue, the second command providing an instruction to the communication interface to transmit the response dialogue message.

18. The method of claim 17, wherein the first event comprises data representative of an incoming customer dialogue message.

19. The method of claim 17, wherein the first command comprises a command to the natural language processing device to determine the meaning of the incoming customer dialogue message.

20. The method of claim 17, wherein the customer context is further derived from customer information associated with a particular customer stored in a database, the customer information comprising one or more of:

account types, account statuses, transaction history, people models, an estimate of customer sentiment, customer goals, and customer social media information.