Tagging telecommunication interactions

Patent metadata

Field	Value
Publication number	US-2025190476-A1
Application number	US-202318537150-A
Country	US
Kind code	A1
Filing date	Dec 12, 2023
Priority date	Dec 12, 2023
Publication date	Jun 12, 2025
Grant date	—

Abstract

In some implementations, a device may obtain a data set that indicates scores for first terms based on association scores between the first terms and second terms. The association scores may be based on occurrences of the first terms in electronic documents that provide transcripts relating to telecommunication interactions, and occurrences of the second terms in electronic records that provide summaries relating to the telecommunication interactions. The device may process an electronic document, that provides a transcript relating to a new telecommunication interaction involving an agent device and a user device, to extract candidate key phrases from the transcript. The device may determine a relevance score for each candidate key phrase based on the score(s), of the data set, associated with the first term(s) in the candidate key phrase. The device may generate tag(s) for the new telecommunication interaction based candidate key phrase(s) associated with highest relevance score(s).

First claim

What is claimed is:

1 . A system for tagging telecommunication interactions, the system comprising:
one or more memories; and one or more processors, communicatively coupled to the one or more memories, configured to:
retrieve a plurality of electronic documents that provide transcripts relating to a plurality of telecommunication interactions involving one or more agent devices and one or more user devices, and a plurality of electronic records that provide agent-prepared summaries relating to the plurality of telecommunication interactions;
scan the plurality of electronic documents and the plurality of electronic records to generate a first index of first terms present in the transcripts and a second index of second terms present in the agent-prepared summaries, wherein the first index and the second index define a plurality of term pairs each including a first term of the first index and a second term of the second index;
determine respective association scores for the plurality of term pairs;
generate a data set that indicates a plurality of scores for the first terms based on respective highest association scores, of the respective association scores, for the first terms;
process an electronic document, that provides a transcript relating to a new telecommunication interaction involving an agent device and a user device, using an artificial intelligence model, to extract a plurality of candidate key phrases from the transcript;
determine a relevance score for each candidate key phrase, of the plurality of candidate key phrases, based on one or more of the plurality of scores, of the data set, associated with one or more of the first terms in the candidate key phrase; and
generate one or more tags for the new telecommunication interaction based on one or more candidate key phrases, of the plurality of candidate key phrases, associated with highest relevance scores.

2 . The system of claim 1 , wherein the one or more processors are further configured to:
process media content containing audio of the new telecommunication interaction using a speech-to-text technique to generate text associated with the new telecommunication interaction; and generate, based on the text, the electronic document that provides the transcript.

3 . The system of claim 1 , wherein the one or more processors, to determine respective association scores for the plurality of term pairs, are configured to:
generate a matrix indicating co-occurrences of the plurality of term pairs in the transcripts and the agent-prepared summaries; and determine, based on the matrix, respective pointwise mutual information (PMI) values for the plurality of term pairs, wherein the respective association scores are the respective PMI values.

4 . The system of claim 1 , wherein the one or more processors, to scan the plurality of electronic documents and the plurality of electronic records, are configured to:
extract respective sets of terms from each of the transcripts, provided in the plurality of electronic documents and each of the agent-prepared summaries provided in the plurality of electronic records; process the respective sets of terms to remove numbers, to correct misspellings, and to combine phrases into single terms; and generate the first index and the second index based on the respective sets of terms.

5 . The system of claim 1 , wherein the one or more processors are further configured to:
identify, based on the one or more tags, information relevant to the new telecommunication interaction; and cause transmission of a communication for a user of the user device that indicates the information.

6 . The system of claim 1 , wherein the one or more processors are further configured to:
determine, based on the respective association scores for the plurality of term pairs, one or more topics absent from the new telecommunication interaction; and perform an action based on the one or more topics being absent from the new telecommunication interaction, the action being one or more of:
causing transmission of a notification for the agent indicating the one or more topics,
causing transmission of a communication for a user of the user device indicating information relating to the one or more topics, or
generating a telecommunication event for an additional telecommunication interaction for the user.

7 . The system of claim 1 , wherein the one or more processors are further configured to:
obtain an electronic record, that provides an agent-prepared summary, associated with the electronic document that provides the transcript; identify an undefined term in the agent-prepared summary; identify, based on the respective association scores, one first term, of the first terms, having a highest association score with the undefined term; and update the electronic record to replace or supplement the undefined term with the one first term.

8 . A method of tagging telecommunication interactions, comprising:
obtaining, by a device, a data set that indicates a plurality of scores for first terms based on association scores between the first terms and second terms,
wherein the association scores are based on occurrences of the first terms in a plurality of electronic documents that provide transcripts relating to a plurality of telecommunication interactions involving one or more agent devices and one or more user devices, and occurrences of the second terms in a plurality of electronic records that provide agent-prepared summaries relating to the plurality of telecommunication interactions;
processing, by the device, an electronic document, that provides a transcript relating to a new telecommunication interaction involving an agent device and a user device, using an artificial intelligence model, to extract a plurality of candidate key phrases from the transcript; determining, by the device, a relevance score for each candidate key phrase, of the plurality of candidate key phrases, based on one or more of the plurality of scores, of the data set, associated with one or more of the first terms in the candidate key phrase; and generating, by the device, one or more tags for the new telecommunication interaction based on one or more candidate key phrases, of the plurality of candidate key phrases, associated with highest relevance scores.

9 . The method of claim 8 , further comprising:
processing media content containing audio of the new telecommunication interaction using a speech-to-text technique to generate text associated with the new telecommunication interaction; and generating, based on the text, the electronic document that provides the transcript.

10 . The method of claim 8 , wherein the artificial intelligence model is configured to identify phrases in unstructured text based on parts of speech.

11 . The method of claim 8 , wherein the artificial intelligence model is a natural language processing model or a neural network model.

12 . The method of claim 8 , further comprising:
storing information associating the one or more tags with the new telecommunication interaction.

13 . The method of claim 8 , wherein determining the relevance score for each candidate key phrase comprises:
identifying the one or more of the first terms in the candidate key phrase; identifying, based on the data set, respective scores, of the plurality of scores, associated with the one or more of the first terms; and combining the respective scores to obtain the relevance score for the candidate key phrase.

14 . The method of claim 8 , further comprising:
identifying, based on the one or more tags, information relevant

Assignees

• Capital One Services Llc

Inventors

• Ramamohan Deepak
• Regati Purna Prasanna Srikar Raju

Classifications

• G06F16/355 Primary

  Creation or modification of classes or clusters · CPC title