Heterogeneous analysis of communication records using large language models

Patent metadata

Field	Value
Publication number	US-2025298986-A1
Application number	US-202418610914-A
Country	US
Kind code	A1
Filing date	Mar 20, 2024
Priority date	Mar 20, 2024
Publication date	Sep 25, 2025
Grant date	—

Abstract

One example method includes receiving a request to generate an analysis of communication records, the communication records associated with a plurality of types of communication records; accessing a plurality of communication records associated with the request, each communication record of the plurality of communication records corresponding to one type of the plurality of types of communication records; for the communication records of a respective type of communication records, generating, using a trained large language model (“LLM”), one or more analyses of the respective communication records; for each type of communication record, generating, using the trained LLM, a homogeneous analysis of the one or more analyses of the respective communication records corresponding to the respective type of communication records; generating, using the trained LLM, a heterogeneous analysis of the homogeneous analyses of the types of communication records; and providing the heterogeneous analysis in response to the request.

First claim

That which is claimed is:

1 . A method comprising:
receiving a request to generate an analysis of communication records, the communication records associated with a plurality of types of communication records; accessing a plurality of communication records associated with the request, each communication record of the plurality of communication records corresponding to one type of the plurality of types of communication records; for the communication records of a respective type of communication records, generating, using a trained large language model (“LLM”), one or more analyses of the respective communication records; for each type of communication record, generating, using the trained LLM, a homogeneous analysis of the one or more analyses of the respective communication records corresponding to the respective type of communication records; generating, using the trained LLM, a heterogeneous analysis of the homogeneous analyses of the types of communication records; and providing the heterogeneous analysis in response to the request.

2 . The method of claim 1 , wherein the plurality of types of communication records comprises meeting transcripts, chat logs, emails, meeting or calendar invitations, text messages, or documents.

3 . The method of claim 1 , further comprising, for each communication record of a respective type of communication records, generating an LLM prompt based on the respective type of communication records.

4 . The method of claim 3 , wherein generating the LLM prompt is based on metadata corresponding to the respective communication records corresponding to the respective type of communication records.

5 . The method of claim 1 , further comprising, for each type of communication records, responsive to determining that a size of the respective homogeneous analysis satisfies a threshold, using the LLM to re-analyze the respective homogeneous analysis, and wherein generating the homogeneous analysis employs the respective re-analysis of the homogeneous analysis.

6 . The method of claim 1 , wherein generating the heterogeneous analysis comprises providing one or more instructions to the LLM indicating information about the one or more types of communication records.

7 . The method of claim 1 , wherein generating the heterogeneous analysis comprises providing one or more instructions to the LLM indicating a weight for one or more types of communication records.

8 . The method of claim 1 , wherein generating the heterogeneous analysis comprises providing one or more instructions to the LLM indicating a prioritization of the one or more types of communication records.

9 . A system comprising:
a non-transitory computer-readable medium; and one or more processors communicatively connected to the non-transitory computer-readable medium, the one or more processors configured to execute processor-executable instructions stored in the non-transitory computer-readable medium to cause the one or more processors to:
receive a request to generate an analysis of communication records, the communication records associated with a plurality of types of communication records;
access a plurality of communication records associated with the request, each communication record of the plurality of communication records corresponding to one type of the plurality of types of communication records;
for the communication records of a respective type of communication records, generate, using a trained large language model (“LLM”), one or more analyses of the respective communication records;
for each type of communication record, generate, using the trained LLM, a homogeneous analysis of the one or more analyses of the respective communication records corresponding to the respective type of communication records;
generate, using the trained LLM, a heterogeneous analysis of the homogeneous analyses of the types of communication records; and
provide the heterogeneous analysis in response to the request.

10 . The system of claim 9 , wherein the plurality of types of communication records comprises meeting transcripts, chat logs, emails, meeting or calendar invitations, text messages, or documents.

11 . The system of claim 9 , wherein the one or more processors are configured to execute further processor-executable instructions stored in the non-transitory computer-readable medium to, for each communication record of a respective type of communication records, generate an LLM prompt based on the respective type of communication records.

12 . The system of claim 11 , wherein generating the LLM prompt is based on metadata corresponding to the respective communication records corresponding to the respective type of communication records.

13 . The system of claim 9 , wherein the one or more processors are configured to execute further processor-executable instructions stored in the non-transitory computer-readable medium to, for each type of communication records, responsive to determining that a size of the respective homogeneous analysis satisfies a threshold, use the LLM to re-analyze the respective homogeneous analysis, and wherein generating the homogeneous analysis employs the respective re-analysis of the homogeneous analysis.

14 . The system of claim 9 , wherein the one or more processors are configured to execute further processor-executable instructions stored in the non-transitory computer-readable medium to provide one or more instructions to the LLM indicating information about the one or more types of communication records.

15 . The system of claim 9 , wherein the one or more processors are configured to execute further processor-executable instructions stored in the non-transitory computer-readable medium to provide one or more instructions to the LLM indicating a weight for one or more types of communication records.

16 . The system of claim 9 , wherein the one or more processors are configured to execute further processor-executable instructions stored in the non-transitory computer-readable medium to provide one or more instructions to the LLM indicating a prioritization of the one or more types of communication records.

17 . A non-transitory computer-readable medium comprising processor-executable instructions configured to cause one or more processors to:
receive a request to generate an analysis of communication records, the communication records associated with a plurality of types of communication records; access a plurality of communication records associated with the request, each communication record of the plurality of communication records corresponding to one type of the plurality of types of communication records; for the communication records of a respective type of communication records, generate, using a trained large language model (“LLM”), one or more analyses of the respective communication records; for each type of communication record, generate, using the trained LLM, a homogeneous analysis of the one or more analyses of the respective communication records corresponding to the respective type of communication records; generate, using the trained LLM, a heterogeneous analysis of the homogeneous analyses of the types of communication records; and provide the heterogeneous analysis in response to the request.

18 . The non-transitory computer-readable medium of claim 17 , further comprising processor-executable instructions configured to cause the one or more processors to, for each communication record of a respective type of communication records, generate an LLM prompt based on the res

Assignees

• Zoom Video Communications Inc

Inventors

• Giovanardi Davide
• Lee Bilung
• Ting Chih-Kai

Classifications

• G06F40/40

  Processing or translation of natural language (natural language analysis G06F40/20; semantic analysis G06F40/30) · CPC title

• G06N20/00

  Machine learning · CPC title

• G06F40/35 Primary

  Discourse or dialogue representation · CPC title

• G06F16/345 Primary

  Summarisation for human users · CPC title