# US 2017/0364506 A1 · System and method for generating phrase based categories of interactions

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

- Number: US 2017/0364506 A1
- Document ID: US20170364506A1
- Status: Published application
- Kind: A1
- Application number: 15/182,897
- Filed: 2016-06-15
- Published: 2017-12-21
- Earliest claimed: 2016-06-15
- Assignees: Nice Ltd.
- Applicants: Nice Ltd.
- Inventors: Jeffrey STERN; Nimrod COHEN
- Web page: https://patentgrep.com/patent/US20170364506A1

## CPC

- G06F 40/289: Phrasal analysis, e.g. finite state techniques or chunking
- G06F 16/285: {Clustering or classification}
- G06F 16/35: Clustering; Classification
- G10L 15/26: Speech to text systems (G10L15/08 takes precedence)
- G10L 25/06: the extracted parameters being correlation coefficients
- G10L 25/51: for comparison or discrimination
- H04M 3/42221: {Conversation recording systems (at the subscriber's set H04M1/656)}
- H04M 3/5175: {Call or contact centers supervision arrangements}
- H04M 3/5191: {interacting with the Internet}
- H04M 2201/12: Counting circuits
- H04M 2201/18: Comparators
- H04M 2201/40: using speech recognition
- H04M 2203/556: Statistical analysis and interpretation
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G10L: SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- H04M: TELEPHONIC COMMUNICATION (circuits for controlling other apparatus via a telephone cable and not involving telephone switching apparatus G08)

## Abstract

In some embodiments, a system and method for generating phrase based categories of interactions may include obtaining a base category, the base category including one or more phrases. A base trend for the base category may be generated based on a frequency of appearance of at least one of the one or more phrases in a set of recorded interactions; a candidate phrase may be obtained and a candidate trend may be generated for the candidate phrase based on a frequency of appearance of the candidate phrase in the set of recorded interactions; a correlation level may be determined by relating the candidate trend to the base trend; and, if the correlation level is greater than a threshold level then the candidate phrase may be included the in the base category.

## Claims

### Claim 1

A computer-implemented method of generating phrase based categories, the method comprising:
obtaining a base category, the base category including one or more phrases;
generating a base trend for the base category based on a frequency of appearance of at least one of the one or more phrases in a set of recorded interactions;
obtaining a candidate phrase and generating a candidate trend for the candidate phrase based on a frequency of appearance of the candidate phrase in the set of recorded interactions;
calculating a correlation level for the candidate trend based on the candidate trend and the base trend, wherein the correlation level quantifies a difference in a behavior of the candidate trend and the base trend; and
if the correlation level is greater than a threshold level then including the candidate phrase in the base category.

### Claim 2 (depends on 1)

The method of claim 1, comprising:
iteratively selecting a candidate phrase from a set of candidate phrases, and, for each selected candidate phrase:
generating a candidate trend for the selected candidate phrase,
determining a correlation level for the selected candidate phrase, and
if the correlation level is greater than a threshold level then including the selected candidate phrase in the base category.

### Claim 3 (depends on 1)

The method of claim 1, comprising:
presenting the candidate phrase to a user; and
including the candidate phrase in the base category based on input received from the user.

### Claim 4 (depends on 1)

The method of claim 1, wherein generating the candidate trend includes:
a. determining a frequency of appearance of the candidate phrase, in the set of recorded interactions, by searching for the candidate phrase according to an accuracy value,
b. generating a first candidate trend based on the frequency of appearance determined using the accuracy value,
c. decreasing the accuracy value to produce a decreased accuracy value,
d. generating a second candidate trend using the decreased accuracy value,
e. determining a correlation level by relating the first candidate trend to the second candidate trend, and if the correlation level is greater than a threshold level then repeating steps a, b, c, d and e; and
if the correlation level is less than a threshold then selecting, for generating the candidate trend, an accuracy value that is larger than the last decreased accuracy value used.

### Claim 5 (depends on 1)

The method of claim 1, wherein generating the candidate trend includes ascertaining that the candidate trend is different from a trend of the number of interactions per time period.

### Claim 6 (depends on 1)

The method of claim 1, comprising, ascertaining that the candidate trend is statistically more correlated to the trend of the base category than to a trend of the number of interactions in a time period.

### Claim 7 (depends on 1)

The method of claim 1, wherein obtaining the candidate phrase includes at least one of: receiving the candidate phrase from a user and identifying the candidate phrase by analyzing the set of recorded interactions.

### Claim 8 (depends on 1)

The method of claim 1, wherein recorded interactions include at least one of: a voice recording, a video recording and a recording of text of a textual interaction.

### Claim 9

A computer-implemented method of generating phrase categories, the method comprising:
defining a reference phrase category that includes one or more phrases;
generating a base trend of the reference category, wherein a trend of a category is defined by a set of values representing the number of recorded interactions, in a respective set of time units, that include at least one phrase included in the category;
generating a candidate trend of a candidate phrase;
calculating a correlation level based on the candidate trend and the base trend; and
if the correlation level is greater than a threshold level then including the candidate phrase in the base category.

### Claim 10 (depends on 9)

The method of claim 9, comprising:
iteratively selecting a candidate phrase from a set of candidate phrases, and, for each selected candidate phrase:
generating a candidate trend for the selected candidate phrase,
calculating a correlation value for the selected candidate phrase, and
if the correlation value is greater than a threshold level then including the selected candidate phrase in the base category.

### Claim 11 (depends on 9)

The method of claim 9, comprising:
presenting the candidate phrase to a user; and
including the candidate phrase in the base category based on input received from the user.

### Claim 12 (depends on 9)

The method of claim 9, wherein generating the candidate trend includes:
a. determining a frequency of appearance of the candidate phrase, in the set of recorded interactions, by searching for the candidate phrase according to an accuracy value,
b. generating a first candidate trend based on the frequency of appearance determined using the accuracy value,
c. decreasing the accuracy value to produce a decreased accuracy value,
d. generating a second candidate trend using the decreased accuracy value,
e. determining a correlation level by relating the first candidate trend to the second candidate trend, and if the correlation level is greater than a threshold level then repeating steps a, b, c, d and e; and
if the correlation level is less than a threshold then selecting, for generating the candidate trend, an accuracy value that is larger than the last decreased accuracy value used.

### Claim 13

A system comprising:
a memory; and
a controller configured to:
obtain a base category, the base category including one or more phrases;
generate a base trend for the base category based on a frequency of appearance of at least one of the one or more phrases in a set of recorded interactions;
obtain a candidate phrase and generating a candidate trend for the candidate phrase based on a frequency of appearance of the candidate phrase in the set of recorded interactions;
determine a correlation level by relating the candidate trend to the base trend; and
if the correlation level is greater than a threshold level then include the candidate phrase in the base category.

### Claim 14 (depends on 13)

The system of claim 13, wherein the controller is further configured to:
iteratively select a candidate phrase from a set of candidate phrases, and, for each selected candidate phrase:
generate a candidate trend for the selected candidate phrase,
determine a correlation level for the selected candidate phrase, and
if the correlation level is greater than a threshold level then include the selected candidate phrase in the base category.

### Claim 15 (depends on 13)

The system of claim 13, wherein the controller is further configured to:
Present the candidate phrase to a user; and
include the candidate phrase in the base category based on input received from the user.

### Claim 16 (depends on 13)

The system of claim 13, wherein generating the candidate trend includes:
a. determining a frequency of appearance of the candidate phrase, in the set of recorded interactions, by searching for the candidate phrase according to an accuracy value,
b. generating a first candidate trend based on the frequency of appearance determined using the accuracy value,
c. decreasing the accuracy value to produce a decreased accuracy value,
d. generating a second candidate trend using the decreased accuracy value,
e. determining a correlation level by relating the first candidate trend to the second candidate trend, and if the correlation level is greater than a threshold level then repeating steps a, b, c, d and e; and
if the correlation level is less than a threshold then selecting, for generating the candidate trend, an accuracy value that is larger than the last decreased accuracy value used.

### Claim 17 (depends on 13)

The method of claim 13, wherein generating the candidate trend includes ascertaining that the candidate trend is different from a trend of the number of interactions per time period.

### Claim 18 (depends on 13)

The system of claim 13, wherein the controller is further configured to ascertain that the candidate trend is statistically more correlated to the trend of the base category than to a trend of the number of interactions in a time period.

### Claim 19 (depends on 13)

The system of claim 13, wherein obtaining the candidate phrase includes at least one of: receiving the candidate phrase from a user and identifying the candidate phrase by analyzing the set of recorded interactions.

### Claim 20 (depends on 13)

The system of claim 13, wherein recorded interactions include at least one of: a voice recording, a video recording and a recording of text of a textual interaction.
