Granted Patent B2

US 12,633,287 · App. 18/525,620 · Granted May 19, 2026

Inventors: Pankaj Dhoolia (Ghaziabad, IN); Daniel T O'Connor (Milton, MA); Venkat Raghavan Ganesh Sekar (Lowell, MA); Andrew James Stoneberg (Clarksburg, MD); Muhtar Burak Akbulut (Waban, MA)

Assignee: International Business Machines Corporation

G10L15/1815G06F40/30G06F40/35G10L15/1822G10L15/183G10L15/22G06F40/237G06F40/289G06F40/295G10L2015/223G10L2015/225

Abstract

Claims (50)

1 . A method comprising:

receiving a dialog system input for processing by a dialog system;

identifying a series of ambiguous intents within the dialog system input;

in response to identifying the series of the ambiguous intents, determining, based on an accessible domain model, one or more relevant annotated themes corresponding to each individual intent within the series of the ambiguous intents, wherein a first relevant annotated theme from the one or more relevant annotated themes include a temporal annotated theme that links a first individual intent with a temporary theme;

executing, using the dialog system, a plurality of conversations corresponding to the dialog system input by employing a set of generic rules, wherein the executing includes:

prioritizing the series of the ambiguous intents based on the set of generic rules that leverage the one or more relevant annotated themes corresponding to each individual intent within the series of the ambiguous intents; and

sequencing, based on the prioritizing, the plurality of conversations corresponding to the dialog system input; and

removing, when no longer applicable, the temporary theme from the first individual intent.

2 . The method of claim 1 , wherein the dialog system input comprises one or more of a textual communication, an utterance, and a non-verbal communication that is convertible into an associated textual communication.

3 . The method of claim 1 , wherein the dialog system comprises at least one of a chatbot, a virtual agent, and an algorithm or machine-learning driven software tool for generating dialog in response to the dialog system input.

4 . The method of claim 1 , wherein the determining the one or more relevant annotated themes corresponding to each individual intent within the series of the ambiguous intents comprises:

utilizing a large language model to leverage at least one of intent-clusters, intent-sequences, and frequent-intent-groups based on accessible logs of historical dialogs.

5 . The method of claim 4 , wherein the large language model further infers dependencies between Application Programming Interface (API) invoking actions.

6 . The method of claim 1 , further comprising:

executing one or more remaining intents from the one or more relevant annotated themes as part of a follow-up dialog.

7 . A computer system comprising:

a processor set;

one or more computer readable storage media; and

program instructions stored on the one or more computer readable storage media to cause the processor set to perform operations comprising:

receiving a dialog system input for processing by a dialog system;

identifying a series of ambiguous intents within the dialog system input;

in response to identifying the series of the ambiguous intents, determining, based on an accessible domain model, one or more relevant annotated themes corresponding to each individual intent within the series of the ambiguous intents, wherein a first relevant annotated theme from the one or more relevant annotated themes include a temporal annotated theme that links a first individual intent with a temporary theme;

executing, using the dialog system, a plurality of conversations corresponding to the dialog system input by employing a set of generic rules, wherein the executing includes:

prioritizing the series of the ambiguous intents based on the set of generic rules that leverage the one or more relevant annotated themes corresponding to each individual intent within the series of the ambiguous intents; and

sequencing, based on the prioritizing, the plurality of conversations corresponding to the dialog system input; and

removing, when no longer applicable, the temporary theme from the first individual intent.

8 . The computer system of claim 7 , wherein the dialog system input comprises one or more of a textual communication, an utterance, and a non-verbal communication that is convertible into an associated textual communication.

9 . The computer system of claim 7 , wherein the dialog system comprises wherein the dialog system comprises at least one of a chatbot, a virtual agent, and an algorithm or machine-learning driven software tool for generating dialog in response to the dialog system input.

10 . The computer system of claim 7 , wherein the determining the one or more relevant annotated themes corresponding to each individual intent within the series of the ambiguous intents comprises:

utilizing a large language model to leverage at least one of intent-clusters, intent-sequences, and frequent-intent-groups based on accessible logs of historical dialogs.

11 . The computer system of claim 10 , wherein the large language model further infers dependencies between Application Programming Interface (API) invoking actions.

12 . The computer system of claim 7 , wherein the operations further comprise:

executing one or more remaining intents from the one or more relevant annotated themes as part of a follow-up dialog.

13 . A computer program product comprising:

one or more computer readable storage media; and

program instructions stored on the one or more computer readable storage media to perform operations comprising;

receiving a dialog system input for processing by a dialog system;

identifying a series of ambiguous intents within the dialog system input;

in response to identifying the series of the ambiguous intents, determining, based on an accessible domain model, one or more relevant annotated themes corresponding to each individual intent within the series of the ambiguous intents, wherein a first relevant annotated theme from the one or more relevant annotated themes include a temporal annotated theme that links a first individual intent with a temporary theme;

executing, using the dialog system, a plurality of conversations corresponding to the dialog system input by employing a set of generic rules, wherein the executing includes:

prioritizing the series of the ambiguous intents based on the set of generic rules that leverage the one or more relevant annotated themes corresponding to each individual intent within the series of the ambiguous intents; and

sequencing, based on the prioritizing, the plurality of conversations corresponding to the dialog system input; and

removing, when no longer applicable, the temporary theme from the first individual intent.

14 . The computer program product of claim 13 , wherein the dialog system input comprises one or more of a textual communication, an utterance, and a non-verbal communication that is convertible into an associated textual communication.

15 . The computer program product of claim 13 , wherein the dialog system comprises wherein the dialog system comprises at least one of a chatbot, a virtual agent, and an algorithm or machine-learning driven software tool for generating dialog in response to the dialog system input.

16 . The computer program product of claim 13 , wherein the determining the one or more relevant annotated themes corresponding to each individual intent within the series of the ambiguous intents comprises:

utilizing a large language model to leverage at least one of intent-clusters, intent-sequences, and frequent-intent-groups based on accessible logs of historical dialogs.

17 . The computer program product of claim 16 , wherein the large language model further infers dependencies between Application Programming Interface (API) invoking actions.

18 . The computer program product of claim 13 , wherein wherein the operations further comprise:

executing one or more remaining intents from the one or more relevant annotated themes as part of a follow-up dialog.

Assignments (1)

Continuity (1)

References Cited (36)