Graph-based natural language generation for conversational systems

Patent metadata

Field	Value
Publication number	US-11706164-B2
Application number	US-202117315726-A
Country	US
Kind code	B2
Filing date	May 10, 2021
Priority date	May 10, 2021
Publication date	Jul 18, 2023
Grant date	Jul 18, 2023

Abstract

A computing device may represent dialog for output by a chatbot as a response graph. The graph may consist of nodes and edges, both of which may have attributes. The graph may be linked with another graph. There may be one or more traversal paths through the graph. The computing device may represent each traversal path as a row in a database table. Each column in the table row may correspond to: one or more nodes in the traversal path, a condition value, weight, and/or other filtering condition. A computing device may display a graphical user interface that allows a user to add, edit, and/or delete nodes and/or edges of the graph. The user interface may also allow the user to compare two or more graphs. The user interface may generate visualizations of traversal paths of a response graph.

First claim

What is claimed is:

1. A method comprising:
causing, by a computing device, a first graph to be displayed via a graphical user interface (GUI), wherein:
the first graph comprises one or more response nodes,
the first graph indicates a structure of a response,
each response node, of the one or more response nodes, indicates a portion of a possible response to an inquiry to be received by a trained chatbot;

receiving, via the GUI, a selection of a first response node;
causing, based on the selection of the first response node, a second graph to be displayed via the GUI, wherein:
the second graph comprises a first portion of the response to the inquiry;
the first portion of the response comprises a plurality of words, wherein the plurality of words is represented by at least one word node in the second graph;

receiving one or more inputs to modify the first portion of the response, resulting in a modified first portion of the response, wherein the one or more inputs comprises a selection of one or more word nodes of the second graph, and wherein the modified first portion of the response corresponds to a modified version of the second graph;
determining one or more differences between the second graph and the modified version of the second graph;
causing the one or more differences to be displayed; and
causing, based on the one or more inputs, the modified first portion of the response to be a trained output by the chatbot, when deployed, in response to receiving the inquiry.

2. The method of claim 1 , further comprising:
causing, based on the selection of the first response node, a route overlay to be displayed on the second graph, wherein the route overlay indicates each word node that is used to generate the first portion of the response.

3. The method of claim 1 , further comprising:
generating, based on the first graph, a table comprising a plurality of responses for the chatbot;
receiving, via the chatbot, a second inquiry;
querying, based on the second inquiry, the table to determine a second response to the second inquiry; and
causing the second response to be outputted by the chatbot.

4. The method of claim 3 , further comprising:
determining, using natural language processing, a context of the second inquiry prior to querying the table.

5. The method of claim 1 , wherein the first graph comprises a content determination graph.

6. The method of claim 1 , wherein the second graph comprises a grammar graph.

7. The method of claim 1 , wherein the receiving one or more inputs to modify the first portion of the response comprises:
receiving a selection of a path of the second graph; and
generating, based on receiving the selection, a preview of a first portion of the response.

8. The method of claim 1 , wherein a node of the first graph is associated with the second graph.

9. The method of claim 1 , wherein the second graph comprises one or more variable nodes, wherein a first variable node of the one or more variable nodes causes additional information to be inserted into the response.

10. A computing device comprising:
one or more processors; and
memory storing instructions that, when executed by the one or more processors, cause the computing device to:
cause a first graph to be displayed via a graphical user interface (GUI), wherein:
the first graph comprises one or more response nodes,
the first graph indicates a structure of the response, and
each response node, of the one or more response nodes, indicates at least a portion of a possible response to an inquiry to be received by a trained chatbot;

receive, via the GUI, a selection of a first response node;
based on the selection of the first response node:
cause a second graph to be displayed via the GUI, wherein:
the second graph comprises a first portion of the response to the inquiry received by the chatbot; and
the first portion of the response comprises a plurality of words, wherein the plurality of words are represented by at least one word node in the second graph;

cause a route overlay to be displayed on the second graph, wherein the route overlay indicates each word node that is used to generate the portion of the response;
receive one or more inputs to modify the first portion of the response, resulting in a modified first portion of the response, wherein the one or more inputs comprises a selection of one or more word nodes of the second graph, and wherein the modified first portion of the response corresponds to a modified version of the second graph;
determine one or more differences between the second graph and the modified version of the second graph;
cause the one or more differences to be displayed; and
cause, based on the one or more inputs, the modified first portion of the response to be a trained output by the chatbot, when deployed, in response to receiving the inquiry.

11. The computing device of claim 10 , wherein the instructions, when executed by the one or more processors, cause the computing device to:
generate, based on the first graph, a table comprising a plurality of responses for the chatbot;
receive, via the chatbot, a second inquiry;
determine, using natural language processing, a context of the second inquiry;
query, based on the context, the table to determine a second response to the second inquiry; and
cause the second response to be outputted by the chatbot.

12. The computing device of claim 10 , wherein the instructions, when executed by the one or more processors, cause the computing device to:
receive a selection of a path of the second graph; and
generate, based on receiving the selection, a preview of the portion of the response.

13. The computing device of claim 10 , wherein the instructions, when executed by the one or more processors, cause the computing device to:
receive a selection of a path of the first graph; and
generate, based on receiving the selection, a preview of the response.

14. The computing device of claim 10 , wherein:
the first graph further comprises one or more conditional nodes;
a first conditional node of the one or more conditional nodes is associated with a value of a variable; and
the response is outputted based on the value of the variable.

15. A non-transitory computer readable medium comprising instructions, that when executed by one or more processors, cause a computing device to:
cause, for display via a graphical user interface (GUI), a first graph wherein:
the first graph comprises one or more response nodes, and
each response node, of the one or more response nodes, comprises a possible response to an inquiry to be received by a trained chatbot;

receive, via the GUI, a selection of a first response node;
based on the selection of the first response node:
cause a second graph to be displayed via the GUI, wherein:
the second graph comprises a portion of the response to the inquiry expected to be received by the chatbot; and
the response comprises a plurality of words, wherein the plurality of words is represented by at least one word node in the second graph;

cause a route overlay to be displayed on the second graph, wherein the route overlay indicates each word node that is used to generate the portion of the response;
receive one or more inputs to modify the portion of the response, resulting in a modified first portion of the response, wherein the one or more inputs comprises a selection of one or more word nodes of the second graph, and wherein the modified first portion of the response corresponds to a modified version of the second graph;
determine one or more differences bet

Assignees

• Capital One Services Llc

Inventors

• Mingels Diana
• Bakshi Amandeep
• Cowell Robert
• Doll Jeremy
• Dunn Jed
• Menasi Mathias

Classifications

• G06T11/26

  Drawing of charts or graphs · CPC title

• H04L51/02 Primary

  using automatic reactions or user delegation, e.g. automatic replies or chatbot-generated messages · CPC title

• G06F16/9024

  Graphs; Linked lists (G06F16/9027 takes precedence) · CPC title

• G06F40/20

  Natural language analysis (semantic analysis of natural language G06F40/30) · CPC title

• G06T11/206

  Physics · mapped topic