# US 2025/0159082 A1 · System and method for providing personalized customer experience in interactive communications

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2025/0159082 A1
- Document ID: US20250159082A1
- Status: Published application
- Kind: A1
- Application number: 19/021,351
- Filed: 2025-01-15
- Published: 2025-05-15
- Earliest claimed: 2022-08-31
- Assignees: Capital One Services, LLC
- Applicants: Capital One Services, LLC
- Inventors: Aysu Ezen CAN
- Web page: https://patentgrep.com/patent/US20250159082A1

## CPC

- H04M 3/5235: {Dependent on call type or called number [DNIS]}
- G10L 15/26: Speech to text systems (G10L15/08 takes precedence)
- G10L 25/27: characterised by the analysis technique
- H04M 3/5233: {Operator skill based call distribution}
- G06F 40/30: Semantic analysis
- H04M: TELEPHONIC COMMUNICATION (circuits for controlling other apparatus via a telephone cable and not involving telephone switching apparatus G08)
- G10L: SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

Disclosed herein are system, method, and computer program product embodiments for machine learning systems to process incoming call-center calls based on caller preferences. Text of historical interactive communications of a set of first callers is used to train one or more machine learning models to extract current caller preferences. A first sentiment score of a current caller may be labeled as a complaint and, based on subsequent utterances of the current caller, a first and second sentiment score trend of the current caller are detected relative to the complaint. For a second sentiment score above a complaint threshold, phrasing is generated for a call center agent interacting with the current caller and, for a second sentiment score below the complaint threshold, utterances are labelled as non-complaint utterances, and identified as a call resolution.

## Claims

### Claim 1

A system comprising:
a machine learning system configured to:
convert, from speech to text, historical interactive communications to generate training data to train one or more machine learning models;
extract, using the trained one or more machine learning models, caller preferences of a current caller;
extract, from a first series of utterances of the current caller, a first sentiment score of the current caller, wherein the first sentiment score is labeled as a complaint;
detect, based on a second series of subsequent utterances of the current caller, a sentiment score trend of the current caller relative to the complaint;
in response to the sentiment score trend, select a third series of subsequent utterances of the current caller;
detect, from the third series of subsequent utterances of the current caller, a second sentiment score of the current caller;
for a second sentiment score above a complaint threshold, generate phrasing for a call center agent interacting with the current caller;
for a second sentiment score below the complaint threshold, label the utterances as non-complaint utterances; and
identify the non-complaint utterances as a call resolution.

### Claim 2 (depends on 1)

The system of claim 1, wherein the caller preferences comprise any of: word choices, a call length, an average call length, an average call resolution time, an average satisfaction level, or a call reason.

### Claim 3 (depends on 2)

The system of claim 2, further comprising an automated call center assistance system configured to:
estimate, based on the call length, the call length of the current call.

### Claim 4 (depends on 3)

The system of claim 3, wherein the automated call center assistance system is further configured to:
communicate the estimate of the call length of the current call to the call center agent.

### Claim 5 (depends on 3)

The system of claim 3, wherein the automated call center assistance system is further configured to:
dynamically determine, based on the average call length, a number of available call center agents.

### Claim 6 (depends on 3)

The system of claim 3, wherein the automated call center assistance system is further configured to:
determine, based on the average call resolution time for a given time period, a customer experience trend; and
communicate the customer experience trend to the call center agent.

### Claim 7 (depends on 3)

The system of claim 3, wherein the automated call center assistance system is further configured to:
determine, based on an average number of one or more historical interactive communications of the caller for the call reason, a distribution of potential call reasons for the current call; and
communicate the distribution of potential call reasons to the call center agent.

### Claim 8 (depends on 1)

The system of claim 1, wherein the machine learning system is further configured to:
infer, based on one or more machine learning models and one or more historical interactive communications of the caller, a plurality of the complaints; and
communicate the plurality of complaints to the call center agent.

### Claim 9 (depends on 8)

The system of claim 8, wherein the automated call center assistance system is further configured to:
selectively route, based on the plurality of complaints from previous interactive communications, the current call to the call center agent.

### Claim 10 (depends on 8)

The system of claim 8, wherein the automated call center assistance system is further configured to:
selectively route the current call to the call center agent based on a percentage of the plurality of complaints within the one or more historical interactive communications of the caller exceeding a predetermined threshold.

### Claim 11 (depends on 8)

The system of claim 8, wherein the automated call center assistance system is further configured to:
determine an average amount of time before the plurality of complaints is detected in the one or more historical interactive communications of the caller; and
communicate the average amount of time to the call center agent.

### Claim 12 (depends on 8)

The system of claim 8, wherein the automated call center assistance system is further configured to:
determine a frequency of word choices, within the one or more historical interactive communications of the caller, occurring during the plurality of complaints;
rank the frequency of word choices; and
communicate the rank of the frequency of word choices associated with the plurality of complaints to the call center agent.

### Claim 13 (depends on 8)

The system of claim 8, wherein the automated call center assistance system is further configured to:
determine an average number of words before the plurality of the complaints are detected in the one or more historical interactive communications of the caller;
predict when a potential one of the plurality of complaints will occur; and
communicate the prediction to the call center agent.

### Claim 14

A computer-implemented method for processing a call, comprising:
convert, from speech to text, historical interactive communications to generate training data to train one or more machine learning models;
extract, using the trained one or more machine learning models, caller preferences of a current caller;
extract, from a first series of utterances of the current caller, a first sentiment score of the current caller, wherein the first sentiment score is labeled as a complaint;
detect, based on a second series of subsequent utterances of the current caller, a sentiment score trend of the current caller relative to the complaint;
in response to the sentiment score trend, select a third series of subsequent utterances of the current caller;
detect, from the third series of subsequent utterances of the current caller, a second sentiment score of the current caller;
for a second sentiment score above a complaint threshold, generate phrasing for a call center agent interacting with the current caller;
for a second sentiment score below the complaint threshold, label the utterances as non-complaint utterances; and
identify the non-complaint utterances as a call resolution.

### Claim 15 (depends on 14)

The computer-implemented method of claim 14, wherein the caller preferences comprise any of: word choices, a call length, an average call length, an average call resolution time, an average satisfaction level, or a call reason.

### Claim 16 (depends on 14)

The computer-implemented method of claim 14, further comprising:
inferring, based on the one or more trained machine learning models and one or more historical interactive communications of the caller, a plurality of complaints; and
communicating the plurality of complaints to the call center agent.

### Claim 17 (depends on 16)

The computer-implemented method of claim 16, further comprising:
selectively routing the current call to the call center agent based on a percentage of the plurality of complaints, within the one or more historical interactive communications of the caller, exceeding a predetermined threshold.

### Claim 18 (depends on 16)

The computer-implemented method of claim 16, further comprising:
determining a frequency of word choices, within the one or more historical interactive communications of the caller, occurring during the plurality of complaints;
ranking the frequency of word choices; and
communicating the rank of the frequency of word choices associated with the plurality of complaints to the call center agent.

### Claim 19

A non-transitory computer-readable device having instructions stored thereon that, when executed by at least one computing device, cause the at least one computing device to perform operations comprising:
converting, from speech to text, historical interactive communications to generate training data to train one or more machine learning models;
extracting, using the trained one or more machine learning models, caller preferences of a current caller;
extracting, from a first series of utterances of the current caller, a first sentiment score of the current caller, wherein the first sentiment score is labeled as a complaint;
detecting, based on a second series of subsequent utterances of the current caller, a sentiment score trend of the current caller relative to the complaint;
in response to the sentiment score trend, selecting a third series of subsequent utterances of the current caller;
detecting, from the third series of subsequent utterances of the current caller, a second sentiment score of the current caller;
for a second sentiment score above a complaint threshold, generate phrasing for a call center agent interacting with the current caller;
for a second sentiment score below the complaint threshold, label the utterances as non-complaint utterances; and
identify the non-complaint utterances as a call resolution.

### Claim 20 (depends on 19)

The non-transitory computer-readable device of claim 19 further configured to perform operations comprising:
inferring, based on another of the one or more trained machine learning models and one or more historical interactive communications of the caller, a plurality of complaints from the one or more historical interactive communications of the caller; and
communicating the plurality of complaints to the call center agent.
