Creating dialogs for a problem-solution data set

Patent metadata

Field	Value
Publication number	US-12499317-B2
Application number	US-202117483275-A
Country	US
Kind code	B2
Filing date	Sep 23, 2021
Priority date	Sep 23, 2021
Publication date	Dec 16, 2025
Grant date	Dec 16, 2025

Abstract

Metadata can be generated for documentation accessed from at least one knowledge source. Intents for main topics of the documentation can be generated using content structure information for the documentation. Dialogs for the intents can be generated. Sub-dialogs can be created for each of the dialogs based, at least in part, on the metadata for the documentation. The dialogs can be configured to be used to modify existing dialogs of a problem-solution data set.

First claim

What is claimed is:

1 . A method, comprising:
accessing a plurality of pages of documentation from at least one knowledge source; generating first metadata describing content of respective ones of the plurality of pages of the documentation; generating, using a processor, a set of first intents for main topics of the documentation using content structure information for the documentation, wherein the generating the set of first intents comprises identifying, using an artificial intelligence (AI) system, intents and entities based on information provided in the respective ones of the plurality of pages of the documentation, and wherein the AI system is trained using machine learning to identify the intents and the entities; generating dialogs for the set of first intents; creating sub-dialogs for each of the generated dialogs for the set of first intents based, at least in part, on the first metadata; determining, for a first intent of the set of first intents, whether there is a second intent, from an existing problem-solution data set used as a chatbot workspace, that is a same as or similar to the first intent; responsive to determining that the second intent, from the existing problem-solution data set used as the chatbot workspace, is not the same as or similar to the first intent, adding to existing dialogs of the existing problem-solution data set the generated dialogs for the set of first intents and the sub-dialogs created for each of the generated dialogs for the set of first intents; and generating a user conversation based on the existing problem-solution set using a chatbot, wherein the chatbot is trained using the AI system based at least in part on the intents.

2 . The method of claim 1 , further comprising:
creating examples for the intents, wherein the examples represent questions to which answers are to be provided by the sub-dialogs; wherein the creating the sub-dialogs further is based, at least in part, on the examples created for the intents.

3 . The method of claim 2 , wherein the sub-dialogs specify as output the answers to the questions.

4 . The method of claim 3 , wherein at least a portion of the sub-dialogs further specify conditions that provide context to the answers.

5 . The method of claim 4 , wherein the sub-dialogs inherit conditions from sub-dialog groups to which the sub-dialogs are assigned.

6 . The method of claim 3 , further comprising:
adding to the respective ones of the plurality of pages of the documentation a corresponding second metadata, the sub-dialogs comprising the questions and the answers addressed by the respective ones of the plurality of pages of the documentation.

7 . The method of claim 1 , further comprising:
identifying a first level of the content structure information as the main topics of the documentation.

8 . A system, comprising:
one or more processors; and
one or more memory devices coupled to the one or more processors, wherein the one or more processors are configured to:
access a plurality of pages of documentation from at least one knowledge source;
generate first metadata describing content of respective ones of the plurality of pages of the documentation;
generate a set of first intents for main topics of the documentation using content structure information for the documentation, wherein the generating the set of first intents comprises identifying, using an artificial intelligence (AI) system, intents and entities based on information provided in the respective ones of the plurality of pages of the documentation, and wherein the AI system is trained using machine learning to identify the intents and the entities;
generate dialogs for the set of first intents;
create sub-dialogs for each of the generated dialogs for the set of first intents based, at least in part, on the first metadata;
determine, for a first intent of the set of first intents, whether there is a second intent, from an existing problem-solution data set used as a chatbot workspace, that is a same or similar to the first intent;
responsive to determining that the second intent, from the existing problem-solution data set used as the chatbot workspace, is not the same as or similar to the first intent, add to existing dialogs of the existing problem-solution data set the generated dialogs for the set of first intents and the sub-dialogs created for each of the generated dialogs for the set of first intents; and
generate a user conversation based on the existing problem-solution set using a chatbot, wherein the chatbot is trained using the AI system based at least in part on the intents.

9 . The system of claim 8 , wherein the one or more processors are further configured to:
create examples for the intents, wherein the examples represent questions to which answers are to be provided by the sub-dialogs, wherein the creating the sub-dialogs further is based, at least in part, on the examples created for the intents.

10 . The system of claim 9 , wherein the sub-dialogs specify as output the answers to the questions.

11 . The system of claim 10 , wherein at least a portion of the sub-dialogs further specify conditions that provide context to the answers.

12 . The system of claim 11 , wherein the sub-dialogs inherit conditions from sub-dialog groups to which the sub-dialogs are assigned.

13 . The system of claim 10 , wherein the one or more processors are further configured to:
add to the respective ones of the plurality of pages of the documentation second metadata, the sub-dialogs comprising the questions and the answers addressed by the respective ones of the plurality of pages of the documentation.

14 . The system of claim 8 , wherein the one or more processors are further configured to:
identify a first level of the content structure information as the main topics of the documentation.

15 . A non-transitory computer-readable medium storing a set of instructions for wireless communication, the set of instructions comprising:
one or more instructions that, when executed by one or more processors of a device, cause the device to:
access a plurality of pages of documentation from at least one knowledge source;
generate first metadata describing content of respective ones of the plurality of pages of the documentation;
generate a set of first intents for main topics of the documentation using content structure information for the documentation, wherein the generating the set of first intents comprises identifying, using an artificial intelligence (AI) system, intents and entities based on information provided in the respective ones of the plurality of pages of the documentation, and wherein the AI system is trained using machine learning to identify the intents and the entities;
generate dialogs for the set of first intents;
creating sub-dialogs for each of the generated dialogs for the set of first intents based, at least in part, on the first metadata;
determine, for a first intent of the set of first intents, whether there is a second intent, from an existing problem-solution data set used as a chatbot workspace, that is a same or similar to the first intent;
responsive to determining that the second intent, from the existing problem-solution data set used as the chatbot workspace, is not the same as or similar to the first intent, add to existing dialogs of the existing problem-solution data set the generated dialogs for the set of first intents and the sub-dialogs created for each of the generated dialogs for the set of first intents; and
generate a user conversation based on the existing problem-solution se

Assignees

• IBM

Inventors

• Kalandyk Piotr
• Kompanowski Hubert
• Szczepanik Grzegorz Piotr
• Januszek Pawel Tadeusz

Classifications

• G06F16/3329 Primary

  Natural language query formulation · CPC title

• G06F40/166

  Editing, e.g. inserting or deleting · CPC title

• G10L15/16

  using artificial neural networks · CPC title

• G06F40/35 Primary

  Discourse or dialogue representation · CPC title