Methods and apparatus for performing machine learning to improve capabilities of an artificial intelligence (AI) entity used for online communications

Patent metadata

Field	Value
Publication number	US-10474673-B2
Application number	US-201715471838-A
Country	US
Kind code	B2
Filing date	Mar 28, 2017
Priority date	Mar 28, 2017
Publication date	Nov 12, 2019
Grant date	Nov 12, 2019

Abstract

A method for providing query responses to a user via online chat establishes a first communication connection for online chat between a user interface and an artificial intelligence (AI) entity comprising a processor and a memory element configured to store a database of query answers; receives a user input query transmitted via the first communication connection; performs a lookup in the database of query answers, to locate a query answer corresponding to the user input query; when unable to locate a query answer, establishes a second communication connection for online chat between the user interface and a live agent interface that transmits responses dynamically provided by a human operator; evaluates a chat between the user interface and the live agent interface; identifies an answer to the user input query, based on evaluating the chat; and stores the answer to be provided by the AI entity in the future.

First claim

What is claimed is:

1. A method for providing query responses to a user via online chat, the method comprising:
establishing a first communication connection for online chat between a user interface and an artificial intelligence (AI) entity comprising at least one processor and a memory element configured to store a database comprising query answers and a plurality of candidate user input queries;
receiving, by the at least one processor, a user input query transmitted via the first communication connection and from the user interface;
performing a lookup, by the at least one processor, in the database of query answers, to locate at least one of the plurality of candidate user input queries within a confidence threshold and determine that the at least one processor is unable to locate a query answer corresponding to the user input query when the plurality of candidate user input queries are not within the confidence threshold;
when the at least one processor is unable to locate a query answer corresponding to the user input query, establishing a second communication connection for online chat between the user interface and a live agent interface, wherein the live agent interface transmits responses that are dynamically provided by a human operator;
evaluating a chat between the user interface and the live agent interface, the chat transmitted via the second communication connection;
identifying an answer to the user input query, based on evaluating the chat; and
storing the answer to be provided by the AI entity during future chat instances of receiving the user input query.

2. The method of claim 1 , wherein the first communication connection is established by connecting to an instant messaging platform using a computer network;
wherein the user interface comprises a graphical user interface (GUI) presented by the instant messaging platform; and
wherein the user input query is received via the GUI and transmitted by the instant messaging platform.

3. The method of claim 1 , wherein the first communication connection is established by connecting to a social media instant messaging platform using a computer network;
wherein the user interface comprises a social media chat interface presented by the social media instant messaging platform; and
wherein the user input query is received via the social media chat interface and transmitted by the social media instant messaging platform.

4. The method of claim 1 , wherein locating a query answer, by the at least one processor, further comprises:
performing the lookup to locate a match for the user input query in the database; and
determining that the at least one processor is unable to locate a query answer when there is no match for the user input query and the plurality of candidate user input queries are not within the confidence threshold.

5. The method of claim 1 , wherein evaluating the chat further comprises:
identifying contents of the chat, wherein the contents comprise the user input query and a response provided by the live agent interface, wherein the answer comprises the response; and
wherein storing the answer further comprises:
associating the response to the user input query, to generate an associated response; and
storing the user input query and the associated response.

6. The method of claim 1 , further comprising:
using an application programming interface (API) to establish the second communication connection, by the at least one processor;
wherein the API comprises at least one of a text API, an audio call API, a video call API, and a live agent text chat API.

7. The method of claim 1 , wherein performing the lookup to locate at least one of the plurality of candidate user input queries within the confidence threshold comprises:
determining an understanding of the user input query using natural language processing; and
evaluating the understanding of the user input query to determine that the plurality of candidate user input queries are not within the confidence threshold.

8. The method of claim 1 , wherein performing the lookup to locate at least one of the plurality of candidate user input queries within the confidence threshold comprises:
determining an understanding of the user input query using machine learning; and
evaluating the understanding of the user input query to determine that the plurality of candidate user input queries are not within the confidence threshold.

9. A system for providing query responses to a user via online chat, the system comprising an artificial intelligence (AI) entity, and the system comprising:
a system memory element configured to store a database of query answers and a plurality of candidate user input queries;
a communication device, configured to establish communication connections using a computer network; and
at least one processor communicatively coupled to the system memory element and the communication device, the at least one processor configured to:
establish a first communication connection, via the communication device, for online chat between a user interface and the AI entity;
receive a user input query transmitted via the first communication connection and from the user interface;
perform a lookup in the database of query answers, to locate at least one of the plurality of candidate user input queries within a confidence threshold and determine that the at least one processor is unable to locate a query answer corresponding to the user input query when the plurality of candidate user input queries are not within the confidence threshold;
when the database does not include a query answer corresponding to the user input query, establish a second communication connection for online chat between the user interface and a live agent interface, wherein the live agent interface transmits responses that are dynamically provided by a human operator;
evaluate a chat between the user interface and the live agent interface, the chat transmitted via the second communication connection;
identify an answer to the user input query, based on evaluating the chat; and
store the answer to be provided by the AI entity during future chat instances of receiving the user input query.

10. The system of claim 9 , wherein the communication device is further configured to establish the first communication connection by connecting to an instant messaging platform using a computer network;
wherein the user interface comprises a graphical user interface (GUI) presented by the instant messaging platform; and
wherein the user input query is received via the GUI and transmitted by the instant messaging platform.

11. The system of claim 9 , wherein the communication device is further configured to establish the first communication connection by connecting to a social media instant messaging platform using a computer network;
wherein the user interface comprises a social media chat interface presented by the social media instant messaging platform; and
wherein the user input query is received via the social media chat interface and transmitted by the social media instant messaging platform.

12. The system of claim 9 , wherein the at least one processor is further configured to locate a query answer, by:
performing the lookup to locate a match for the user input query in the database; and
determining that the at least one processor is unable to locate a query answer when there is no match for the user input query and the plurality of candidate user input queries are not within the confidence threshold.

13. The system of claim 9 , wherein the at least one processor is further configured to evaluate the chat, by:
ident

Assignees

• Salesforce Com Inc

Inventors

• Vora Darshil Vipul
• Kwak Youngjun

Classifications

• G06N5/041

  Abduction · CPC title

• G06N20/00

  Machine learning · CPC title

• G06F16/245 Primary

  Query processing · CPC title

• G06N3/006

  based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO] · CPC title

• G06F16/22

  Indexing; Data structures therefor; Storage structures · CPC title