CN106847305A - A kind of method and device of the recording data for processing service calls - Google Patents
A kind of method and device of the recording data for processing service calls Download PDFInfo
- Publication number
- CN106847305A CN106847305A CN201510883478.8A CN201510883478A CN106847305A CN 106847305 A CN106847305 A CN 106847305A CN 201510883478 A CN201510883478 A CN 201510883478A CN 106847305 A CN106847305 A CN 106847305A
- Authority
- CN
- China
- Prior art keywords
- recording data
- rhythm characteristic
- customer service
- dialogue
- evaluated
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Abstract
The invention provides a kind of method and device of the recording data for processing service calls, wherein, record has the dialogue that user and customer service are represented in the recording data of service calls, and the method includes：According to recording data to be evaluated, the user session rhythm characteristic and customer service for obtaining recording data to be evaluated represent dialogue rhythm characteristic；User session rhythm characteristic and customer service according to recording data to be evaluated represent dialogue rhythm characteristic, using predetermined satisfaction evaluation model, determine the satisfaction evaluation result of recording data to be evaluated.The solution of the present invention, can expand the feature space of satisfaction evaluation, from new angle preferably for customer service hot line operation provides technical support in the user satisfaction situation of dialogue rhythm aspect automatic Evaluation customer service recording.
Description
Technical field
The present invention relates to communication technical field, more particularly to a kind of method of the recording data for processing service calls
And device.
Background technology
Mainly pass through interactive voice answering (Interactive before customer service hot line recording satisfaction evaluation
Voice Response, IVR) or transmitting short message mode directly from user side obtain, but due to user ginseng
It is not very high with response rate with degree, most hotline does not complete satisfaction evaluation.Therefore
Automatic satisfaction evaluation is generated, but automatic satisfaction evaluation is currently mostly to be incited somebody to action by speech recognition engine
Recording data is converted into the text after identification, then is modeled by text emotion word correlation circumstance, and it is right to complete
The automatic satisfaction evaluation of the recording.Wherein, speech analysis techniques and machine learning techniques have been related generally to.
Speech analysis techniques are, by voice process technology, input speech signal to be decomposed into different changes
The characteristic vector changed in dimension.Mainly include speech signal pre-processing technology, Speech processing conversion, language
The speech feature extraction technology and encoding and decoding speech technology in each domain of sound etc..
Machine learning is that the important branch of artificial intelligence field, especially speech recognition technology are artificial intelligence necks
One important research field in domain.By feature extraction, statistical model is set up, model optimization, pattern-recognition,
Data mining etc. completes the concrete practice of artificial intelligence.
Therefore, the satisfaction for mainly being obtained the recording of customer service hot line by following three kinds of schemes in the prior art is commented
Valency result.
Scheme one：By speech recognition engine by hot line recording data " translation " be content of text, Ran Houtong
Cross natural language understanding technology and obtain the satisfaction correlation analysis of text, so as to judge the use of this recording of taking on the telephone
Family satisfaction situation.
Scheme two：Emotional change in the call for obtaining user is analyzed by speech emotional, so that analyze determination should
The user satisfaction situation taken on the telephone.
Scheme three：By combine phonetic feature and identification after text feature constitute union feature come analysis modeling from
It is dynamic to obtain the user satisfaction taken on the telephone.
However, in above-mentioned three kinds of schemes, scheme one due to being identified rear text using speech recognition engine,
This text message is processed so as to obtain user satisfaction information, the voice that actually lost in recording originally is expired
Meaning degree information, many unsatisfied situations of user are not reflected completely, have impact on satisfaction evaluation accurate
True property.Scheme two is simple to judge that the satisfaction of the user particularly preferred can not be evaluated from speech emotional
Effect, on the one hand because inherently one fuzzy subjective concept of emotion, evaluation criterion varies with each individual；
On the other hand because semantic information contained in having lost text, union feature is needed to use more preferably
Completion automatic Evaluation task.
But, although scheme three is automatic full by text affective characteristics after union language sound affective characteristics and identification
Meaning degree evaluation achieves larger lifting, but due to fuzzy subjective of emotion and satisfaction inherently
Concept, current effect still has larger room for promotion, it is necessary to be improved with reference to further types of feature existing
Automatic Evaluation effect.
The content of the invention
In order to overcome the above-mentioned problems in the prior art, The embodiment provides one kind treatment visitor
The method and device of the recording data of phone is taken, can be in the use of dialogue rhythm aspect automatic Evaluation customer service recording
Family satisfaction situation, expands the feature space of satisfaction evaluation, is preferably customer service hot line from new angle
Operation provides technical support.
In order to solve the above-mentioned technical problem, the present invention is adopted the following technical scheme that：
According to the one side of the embodiment of the present invention, there is provided a kind of side of the recording data for the treatment of service calls
Method, record has the dialogue that user and customer service are represented in the recording data of the service calls, and methods described includes：
According to recording data to be evaluated, user session rhythm characteristic and the visitor of the recording data to be evaluated are obtained
Clothes represent dialogue rhythm characteristic；
User session rhythm characteristic and customer service according to the recording data to be evaluated represent dialogue rhythm characteristic,
Using predetermined satisfaction evaluation model, the satisfaction evaluation result of the recording data to be evaluated is determined.
Wherein, it is described according to recording data to be evaluated in such scheme, obtain the recording data to be evaluated
User session rhythm characteristic and customer service represent dialogue rhythm characteristic, including：
The recording data of the first preset duration is obtained from the recording data to be evaluated；
Engage in the dialogue both sides' phonetic segmentation to the recording data of first preset duration, obtains the first of user
The first voice segments collection that voice segments collection and customer service are represented；
The first voice segments collection according to the user, obtains the user session rhythm of the recording data to be evaluated
Feature, and the first voice segments collection represented according to the customer service, obtain the visitor of the recording data to be evaluated
Clothes represent dialogue rhythm characteristic.
Wherein, in such scheme, the user session rhythm characteristic at least include first preset duration,
Jing Yin total duration, the recording of first preset duration that the recording data of first preset duration includes
Jing Yin hop count that language total duration that data include, the recording data of first preset duration include,
Average duration, first preset duration of include Jing Yin section of the recording data of first preset duration
The hop count of all speech segments for including of recording data, the recording data of first preset duration includes
Speech segment average duration, the language total duration of the user, the speech segment number of the user, the use
The language total duration that the average duration of the speech segment at family, the language total duration of the user are represented with the customer service
Ratio, the ratio of speech segment number that is represented with the customer service of the speech segment number of the user in one kind or many
Kind；
The customer service represents dialogue rhythm characteristic at least includes language total duration, the visitor that the customer service is represented
Take one or more in speech segment number, the average duration of the speech segment of customer service representative for representing.
Wherein, in such scheme, in the user session rhythm characteristic according to the recording data to be evaluated and visitor
Clothes represent dialogue rhythm characteristic, using predetermined satisfaction evaluation model, determine the recording to be evaluated
Before the satisfaction evaluation result of data, methods described also includes：
According to the recording data of known satisfaction evaluation result, the record of the known satisfaction evaluation result is obtained
The user session rhythm characteristic of sound data and customer service represent dialogue rhythm characteristic；
The user session rhythm characteristic of the recording data according to the known satisfaction evaluation result and customer service generation
Table talks with rhythm characteristic, and the satisfaction evaluation model is determined using SVMs (SVM) algorithm.
Wherein, in such scheme, the satisfaction evaluation model include representing it is satisfied evaluate first support to
Amount and dissatisfied the second support vector evaluated of representative；
The user session rhythm characteristic of the recording data according to the known satisfaction evaluation result and visitor
Clothes represent dialogue rhythm characteristic, and the satisfaction evaluation model is determined using SVM algorithm, including：
The user session rhythm characteristic of the recording data according to the known satisfaction evaluation result and customer service generation
Table talks with rhythm characteristic, obtains first and talks with rhythm characteristic vector, the first dialogue rhythm characteristic vector
Dimension is equal to user session rhythm characteristic and the customer service generation of the recording data of the known satisfaction evaluation result
Table talks with the number sum of rhythm characteristic；
According to multiple the first dialogue rhythm characteristic vector, using the polynomial kernel letter of the SVM algorithm
Number, determines first support vector and second support vector.
Wherein, in such scheme, the user session rhythm characteristic according to the recording data to be evaluated and
Customer service represents dialogue rhythm characteristic, using predetermined satisfaction evaluation model, determines the record to be evaluated
The satisfaction evaluation result of sound data, including：
User session rhythm characteristic and customer service according to the recording data to be evaluated represent dialogue rhythm characteristic,
Obtain second and talk with rhythm characteristic vector, the vectorial dimension of the second dialogue rhythm characteristic is equal to described to be evaluated
The user session rhythm characteristic of valency recording data and customer service represent the number sum of dialogue rhythm characteristic；
Obtain institute second and talk with that rhythm characteristic is vectorial and the first distance between first support vector, and
The second dialogue rhythm characteristic is vectorial and second distance between second support vector；
Whether first distance is judged less than the first predetermined threshold value, if so, the then recording data to be evaluated
Satisfaction evaluation result for satisfaction；
If it is not, whether the second distance is then judged less than the second predetermined threshold value, if the second distance is less than
Second predetermined threshold value, then the satisfaction evaluation result of the recording data to be evaluated is dissatisfied.
Wherein, in such scheme, the recording data of satisfaction evaluation result known to the basis is obtained
The user session rhythm characteristic of the recording data of the known satisfaction evaluation result and customer service represent dialogue section
Feature is played, including：
The recording data of the second preset duration is obtained from the recording data of the known evaluation result；
Engage in the dialogue both sides' phonetic segmentation to the recording data of second preset duration, obtains the second of user
The second voice segments collection that voice segments collection and customer service are represented；
The second voice segments collection according to the user, obtains the recording data of the known satisfaction evaluation result
User session rhythm characteristic, and the second voice segments collection represented according to the customer service obtains described known
The customer service of the recording data of satisfaction evaluation result represents dialogue rhythm characteristic.
According to the other side of the embodiment of the present invention, a kind of recording data for processing service calls is additionally provided
Device, record has the dialogue that user and customer service are represented, described device in the recording data of the service calls
Including：
First extraction module, for according to recording data to be evaluated, obtaining the use of the recording data to be evaluated
Rhythm characteristic is talked with family and customer service represents dialogue rhythm characteristic；
Evaluation module, represents for the user session rhythm characteristic according to the recording data to be evaluated and customer service
Dialogue rhythm characteristic, using predetermined satisfaction evaluation model, determines the recording data to be evaluated
Satisfaction evaluation result.
Wherein, in such scheme, first extraction module includes：
First acquisition unit, the recording number for obtaining the first preset duration from the recording data to be evaluated
According to；
First cutting unit, cuts for the both sides' voice that engages in the dialogue of the recording data to first preset duration
Point, the first voice segments collection that the first voice segments collection and customer service for obtaining user are represented；
Feature acquiring unit, for the first voice segments collection according to the user, obtains the recording to be evaluated
The user session rhythm characteristic of data, and the first voice segments collection represented according to the customer service, obtain described
The customer service of recording data to be evaluated represents dialogue rhythm characteristic.
Wherein, in such scheme, the user session rhythm characteristic at least include first preset duration,
Jing Yin total duration, the recording of first preset duration that the recording data of first preset duration includes
Jing Yin hop count that language total duration that data include, the recording data of first preset duration include,
Average duration, first preset duration of include Jing Yin section of the recording data of first preset duration
The hop count of all speech segments for including of recording data, the recording data of first preset duration includes
Speech segment average duration, the language total duration of the user, the speech segment number of the user, the use
The language total duration that the average duration of the speech segment at family, the language total duration of the user are represented with the customer service
Ratio, the ratio of speech segment number that is represented with the customer service of the speech segment number of the user in one kind or many
Kind；
The customer service represents dialogue rhythm characteristic at least includes language total duration, the visitor that the customer service is represented
Take one or more in speech segment number, the average duration of the speech segment of customer service representative for representing.
Wherein, in such scheme, described device also includes：
Second extraction module, for the recording data according to known satisfaction evaluation result, obtains described known
The user session rhythm characteristic of the recording data of satisfaction evaluation result and customer service represent dialogue rhythm characteristic；
Model building module, for the user session of the recording data according to the known satisfaction evaluation result
Rhythm characteristic and customer service represent dialogue rhythm characteristic, and the satisfaction evaluation model is determined using SVM algorithm.
Wherein, in such scheme, the satisfaction evaluation model include representing it is satisfied evaluate first support to
Amount and dissatisfied the second support vector evaluated of representative；
The model building module includes：
Second acquisition unit, for the user session of the recording data according to the known satisfaction evaluation result
Rhythm characteristic and customer service represent dialogue rhythm characteristic, obtain first and talk with rhythm characteristic vector, described first pair
The dimension of words rhythm characteristic vector is equal to the user session of the recording data of the known satisfaction evaluation result
Rhythm characteristic and customer service represent the number sum of dialogue rhythm characteristic；
Model sets up unit, for according to multiple the first dialogue rhythm characteristic vector, using the SVM
The Polynomial kernel function of algorithm, determines first support vector and second support vector.
Wherein, in such scheme, the evaluation module includes：
3rd acquiring unit, for the user session rhythm characteristic according to the recording data to be evaluated and customer service
Dialogue rhythm characteristic is represented, second is obtained and is talked with rhythm characteristic vector, the second dialogue rhythm characteristic vector
Dimension be equal to the user session rhythm characteristic of the recording data to be evaluated and customer service to represent dialogue rhythm special
The number sum levied；
4th acquiring unit, for obtain institute second talk with rhythm characteristic it is vectorial with first support vector it
Between the first distance, and the second dialogue rhythm characteristic it is vectorial and second support vector between the
Two distances；
First judging unit, for whether judging first distance less than the first predetermined threshold value, if so, then
The satisfaction evaluation result of the recording data to be evaluated is satisfaction；
Second judging unit, for when first judging unit judged result for it is no when, judge described the
Whether two distances are less than the second predetermined threshold value, if the second distance is less than second predetermined threshold value, institute
It is dissatisfied to state the satisfaction evaluation result of recording data to be evaluated.
Wherein, in such scheme, second extraction module includes：
5th acquiring unit, for obtaining the second preset duration from the recording data of the known evaluation result
Recording data；
Second cutting unit, cuts for the both sides' voice that engages in the dialogue of the recording data to second preset duration
Point, the second voice segments collection that the second voice segments collection and customer service for obtaining user are represented；
6th acquiring unit, for the second voice segments collection according to the user, obtains the known satisfaction
The user session rhythm characteristic of the recording data of evaluation result, and the second voice represented according to the customer service
Section collection, the customer service for obtaining the recording data of the known satisfaction evaluation result represents dialogue rhythm characteristic.
The beneficial effect of the embodiment of the present invention is：
The method and device of the recording data of the treatment service calls of the embodiment of the present invention, from recording number to be evaluated
Dialogue rhythm characteristic is represented according to middle acquisition user session rhythm characteristic and customer service, and according to the user session for obtaining
Rhythm characteristic and customer service represent dialogue rhythm characteristic, using predetermined satisfaction evaluation model, it is determined that treating
Evaluate the satisfaction evaluation result of recording data.Therefore, embodiments of the invention will talk with rhythm characteristic conduct
The foundation of satisfaction evaluation, the user satisfaction situation that can be recorded in the customer service of dialogue rhythm aspect automatic Evaluation,
The feature space of satisfaction evaluation is expanded, from new angle preferably for customer service hot line operation provides technology
Support.
Brief description of the drawings
The method that Fig. 1 represents the recording data of the treatment service calls of the first embodiment of the embodiment of the present invention
Flow chart；
Fig. 2 represents the device of the recording data of the treatment service calls of the second embodiment of the embodiment of the present invention
Structured flowchart；
Fig. 3 represents the device of the recording data of the treatment service calls of the 3rd embodiment of the embodiment of the present invention
Structured flowchart.
Specific embodiment
The exemplary embodiment of the disclosure is more fully described below with reference to accompanying drawings.Although being shown in accompanying drawing
The exemplary embodiment of the disclosure, it being understood, however, that may be realized in various forms the disclosure without should be by
Embodiments set forth here is limited.Conversely, there is provided these embodiments are able to be best understood from this
It is open, and can by the scope of the present disclosure it is complete convey to those skilled in the art.
First embodiment
According to the one side of the embodiment of the present invention, there is provided a kind of side of the recording data for the treatment of service calls
Method, the method first, according to recording data to be evaluated, obtains the user session of the recording data to be evaluated
Rhythm characteristic and customer service represent dialogue rhythm characteristic；Then, according to the user couple of the recording data to be evaluated
Words rhythm characteristic and customer service represent dialogue rhythm characteristic, using predetermined satisfaction evaluation model, it is determined that
The satisfaction evaluation result of the recording data to be evaluated.
Therefore, the method for the recording data of the treatment service calls of the embodiment of the present invention, there is provided new
Satisfaction evaluation angle, expands the feature space of satisfaction evaluation, is preferably customer service heat from new angle
Line operation provides technical support.
As shown in figure 1, the method includes：
Step S11, according to recording data to be evaluated, obtain the user session section of the recording data to be evaluated
Play feature and customer service represents dialogue rhythm characteristic.
What is recorded in the recording data of service calls is the dialogue that user represents with customer service.Wherein, by this
Dialogue between the user for planting the storage of recording form and customer service representative, not only have recorded conversation content, and
It is also recorded for the dialogue rhythm that user represents with customer service.In the dialogue between user and customer service represent, for
Attitude of the user retained in dialog procedure, can typically influence the rhythm of dialogue.Therefore, the present invention is implemented
In the method for the recording data of the treatment service calls of example, will dialogue rhythm as judging customer satisfaction evaluation
Foundation.
Therefore, it needs to be determined that before the satisfaction evaluation result of a certain bar recording data to be evaluated, it is necessary to from
User session rhythm characteristic and the customer service generation of the recording data to be evaluated are obtained in the recording data to be evaluated
Table talks with rhythm characteristic.
Specifically, step S11 includes：
The recording data of the first preset duration is obtained from the recording data to be evaluated；
Engage in the dialogue both sides' phonetic segmentation to the recording data of first preset duration, obtains the first of user
The first voice segments collection that voice segments collection and customer service are represented；
The first voice segments collection according to the user, obtains the user session rhythm of the recording data to be evaluated
Feature, and the first voice segments collection represented according to the customer service, obtain the visitor of the recording data to be evaluated
Clothes represent dialogue rhythm characteristic.
In for any one recording data of service calls, wherein, in the front portion time that recording starts
Usually user and customer service represent the communication to essential information, the satisfaction evaluation knot with required determination
Fruit universal relation is little.So, in order to reduce the statistics to unnecessary information, accelerate satisfaction evaluation result
Acquisition, the user session rhythm characteristic of the recording data to be evaluated is being obtained from recording data to be evaluated
, it is necessary to intercept the record of the first preset duration from recording data to be evaluated first when representing session features with customer service
Sound data, for example, obtaining the recording data of the second half section duration of recording data to be evaluated.Of course, it is possible to manage
Solution, the percentage of shared recording data whole duration to be evaluated specific for the first preset duration, not
Be limited to 50 percent, can also be recording data to be evaluated whole durations rear three/it is second-class.
Wherein, the user session rhythm characteristic at least includes that first preset duration, described first are preset
Jing Yin total duration that the recording data of duration includes, the recording data of first preset duration include
It is Jing Yin hop count that language total duration, the recording data of first preset duration include, described first pre-
If average duration, the recording data of first preset duration of include Jing Yin section of the recording data of duration
The speech segment that the hop count of all speech segments for including, the recording data of first preset duration include
Average duration, the language total duration of the user, the speech segment number of the user, the speech segment of the user
Average duration, the language total duration of the user ratio, the institute of language total duration that are represented with the customer service
State one or more in the ratio of the speech segment number that the speech segment number of user is represented with the customer service；
The customer service represents dialogue rhythm characteristic at least includes language total duration, the visitor that the customer service is represented
Take one or more in speech segment number, the average duration of the speech segment of customer service representative for representing.
Wherein, user session rhythm characteristic and customer service represents dialogue rhythm characteristic and can be collectively referred to as talking with rhythm characteristic,
These dialogue rhythm characteristics are the roots after to the statistics of the recording data of satisfaction evaluation results known to a large amount of
Result is therefrom chosen according to statistics.
Specifically, recording data is engaged in the dialogue after both sides' phonetic segmentation using voice cutting techniques, is first united
Difference of the statistic of target between customer service representative and user data is counted, and object statistics amount is before dialogue
The difference of section and back segment, such as using equation below：
X1=(B1+B2+B3+ ...)/(A1+A2+A3+ ...)
The ratio X1 that language total duration is represented in the second half section user spoken utterances total duration of dialogue and customer service is calculated,
Wherein, B1 is first language duration of latter segment user, and A1 is first that latter segmentation customer service is represented
Individual language duration；
Again or utilize equation below：
X2=max (B)/max (A)
The maximum duration and second half section customer service for calculating the single cutting section of second half section user represent single cutting section
The ratio X2 of maximum duration.Wherein, B represents the duration of the single cutting section of second half section user, after A is represented
Half section of customer service represents the duration of single cutting section.Max functions characterize the maximum found in all goal sets.
Therefore, by above-mentioned similar statistical method, about nearly thousand such rhythm characteristics can be obtained.This
When, the above-mentioned multiple rhythm characteristics that can be obtained according to user and customer service represent between difference from big to small
Arrangement, selection ranks the feature of first some as final feature.
Step S13, the user session rhythm characteristic according to the recording data to be evaluated and customer service represent dialogue
Rhythm characteristic, using predetermined satisfaction evaluation model, determines the satisfaction of the recording data to be evaluated
Degree evaluation result.
Wherein, satisfaction evaluation module is true previously according to the recording data of a plurality of known satisfaction evaluation result
Fixed.Specifically, first, according to the recording data of known satisfaction evaluation result, obtain described known full
The user session rhythm characteristic of the recording data of meaning degree evaluation result and customer service represent dialogue rhythm characteristic；Then,
It is right that the user session rhythm characteristic of the recording data according to the known satisfaction evaluation result and customer service are represented
Words rhythm characteristic, the satisfaction evaluation model is determined using SVM algorithm.
Wherein, in the recording data according to known satisfaction evaluation result, the known satisfaction is obtained
When the user session rhythm characteristic of the recording data of evaluation result and customer service represent dialogue rhythm characteristic, specifically,
Firstly, it is necessary to the recording data of the second preset duration is obtained from the recording data of the known evaluation result,
Latter section of recording data in whole durations of the recording data for obtaining known evaluation result, to reduce to not
The treatment of necessary data, improves data processing speed；Then, to the recording data of second preset duration
Engage in the dialogue both sides' phonetic segmentation, the second voice segments collection that the second voice segments collection and customer service for obtaining user are represented；
Finally, the second voice segments collection according to the user, obtains the recording number of the known satisfaction evaluation result
According to user session rhythm characteristic, and the second voice segments collection represented according to the customer service, obtain it is described
Know that the customer service of the recording data of satisfaction evaluation result represents dialogue rhythm characteristic.
User session rhythm characteristic and the visitor of the recording data of known evaluation result are obtained by the above method
After clothes represent dialogue rhythm characteristic, then using SVM algorithm, the satisfaction evaluation result according to known to a large amount of
Recording data user session rhythm characteristic and customer service represent dialogue rhythm characteristic, obtain satisfaction evaluation mould
Type.
Specifically, in the user session rhythm characteristic of the recording data according to the known satisfaction evaluation result
Dialogue rhythm characteristic is represented with customer service, when determining the satisfaction evaluation model using SVM algorithm, can be first
First, the user session rhythm characteristic of the recording data according to the known satisfaction evaluation result and customer service is represented
Dialogue rhythm characteristic, obtains first and talks with rhythm characteristic vector, the dimension of the first dialogue rhythm characteristic vector
Number is represented equal to the user session rhythm characteristic of the recording data of the known satisfaction evaluation result and customer service
Talk with the number sum of rhythm characteristic；Then, according to multiple the first dialogue rhythm characteristic vector, utilize
The Polynomial kernel function of the SVM algorithm, determines first support vector and second support vector.
Wherein, the learning model of SVM to be one have supervision, in machine learning field, is commonly used to carry out
Pattern-recognition, classification and regression analysis.SVM methods are by a Nonlinear Mapping p, sample
This space reflection is in a higher-dimension or even infinite dimensional feature space so that non-in original sample space
The problem of linear separability is converted into the problem of the linear separability in feature space.
In the method for the recording data of the treatment service calls of the embodiment of the present invention, it is assumed that full known to each
The user session rhythm characteristic of the recording data of meaning degree evaluation result and customer service represent dialogue rhythm characteristic and have
13, then recording data 13 characteristic values of correspondence of each known satisfaction evaluation result.At this moment, root
One 13 can be obtained according to corresponding 13 characteristic values of the recording data of each known satisfaction evaluation result
The dialogue rhythm characteristic vector Xt of dimension.So, each dialogue rhythm characteristic vector Xt one satisfaction of correspondence
Degree evaluation result, to be satisfied with or being dissatisfied.
Now, using Polynomial kernel function k (x, y)=[(the ax*y)+c] of SVMd, to known full
The dialogue rhythm characteristic vector of meaning degree evaluation result is trained, and satisfied first for evaluating is represented so as to find
Support vector Xa and dissatisfied the second support vector Xb for evaluating of representative.Wherein, a, c and d are constant,
Can sets itself according to the actual requirements.
Therefore, step S13 includes：
User session rhythm characteristic and customer service according to the recording data to be evaluated represent dialogue rhythm characteristic,
Obtain second and talk with rhythm characteristic vector, the vectorial dimension of the second dialogue rhythm characteristic is equal to described to be evaluated
The user session rhythm characteristic of valency recording data and customer service represent the number sum of dialogue rhythm characteristic；
Obtain institute second and talk with that rhythm characteristic is vectorial and the first distance between first support vector, and
The second dialogue rhythm characteristic is vectorial and second distance between second support vector；
Whether first distance is judged less than the first predetermined threshold value, if so, the then recording data to be evaluated
Satisfaction evaluation result for satisfaction；
If it is not, whether the second distance is then judged less than the second predetermined threshold value, if the second distance is less than
Second predetermined threshold value, then the satisfaction evaluation result of the recording data to be evaluated is dissatisfied.
Wherein, first apart from Da=sqrt (∑ (Xt_i-Xa_i) ^2), second distance Db=sqrt (∑s
(Xt_i-X_b i) ^2, wherein, i=1,2..n, n are characteristic dimensions.Xt_i be the second dialogue rhythm characteristic to
I-th coordinate representation of amount, Xa_i is i-th coordinate representation of the first support vector, and Xb_i is second
Support i-th coordinate representation of vector.
In addition, can also determine to obtain the dissatisfied data of how many target from assessment data by calculating recall rate,
So that it is determined that the accuracy of correspondence system.
In sum, when it needs to be determined that a certain bar recording data to be evaluated satisfaction evaluation result when, it is necessary to
User session feature is obtained from this recording data to be evaluated and customer service represents session features, that is, talk with rhythm
Feature, then obtains a multidimensional dialogue rhythm characteristic vector Xt according to these dialogue rhythm characteristics；Then,
Calculate respectively this dialogue rhythm characteristic vector and the first support vector Xa and the second support vector Xb away from
From；Finally, judge dialogue rhythm characteristic vector with the distance of which support vector in certain threshold range
Within, then which branch is the evaluation result to the corresponding recording data to be evaluated of dialogue rhythm characteristic vector be
Support vectorial corresponding satisfaction evaluation result.
It follows that the method for the recording data of the treatment service calls of the embodiment of the present invention is one new expiring
Meaning degree evaluates angle, and its calculating process is very simple, can quickly assess satisfaction, and can reach
Almost original identical evaluates performance, so as to be adapted to quickly process the substantial amounts of customer service recording data of each company.
Second embodiment
According to the other side of the embodiment of the present invention, a kind of recording data for processing service calls is additionally provided
Device, as shown in Fig. 2 the device 200 includes：
First extraction module 205, for according to recording data to be evaluated, obtaining the recording data to be evaluated
User session rhythm characteristic and customer service represent dialogue rhythm characteristic；
Evaluation module 207, for the user session rhythm characteristic according to the recording data to be evaluated and customer service
Dialogue rhythm characteristic is represented, using predetermined satisfaction evaluation model, the recording number to be evaluated is determined
According to satisfaction evaluation result.
3rd embodiment
According to the other side of the embodiment of the present invention, a kind of recording data for processing service calls is additionally provided
Device, as shown in figure 3, the device 200 includes：
First extraction module 205, for according to recording data to be evaluated, obtaining the recording data to be evaluated
User session rhythm characteristic and customer service represent dialogue rhythm characteristic；
Evaluation module 207, for the user session rhythm characteristic according to the recording data to be evaluated and customer service
Dialogue rhythm characteristic is represented, using predetermined satisfaction evaluation model, the recording number to be evaluated is determined
According to satisfaction evaluation result.
Alternatively, first extraction module 205 includes：
First acquisition unit 2051, for obtaining the first preset duration from the recording data to be evaluated
Recording data；
First cutting unit 2052, engages in the dialogue both sides for the recording data to first preset duration
Phonetic segmentation, the first voice segments collection that the first voice segments collection and customer service for obtaining user are represented；
Feature acquiring unit 2053, for the first voice segments collection according to the user, obtains described to be evaluated
The user session rhythm characteristic of valency recording data, and the first voice segments collection represented according to the customer service, obtain
Customer service to the recording data to be evaluated represents dialogue rhythm characteristic.
Alternatively, the user session rhythm characteristic at least includes first preset duration, described first pre-
If the Jing Yin total duration that the recording data of duration includes, the recording data of first preset duration include
Language total duration, the recording data of first the preset duration Jing Yin hop count, described first that include
Average duration, the recording number of first preset duration of include Jing Yin section of the recording data of preset duration
According to the speech segment that the recording data of the hop count of all speech segments for including, first preset duration includes
Average duration, the language total duration of the user, the speech segment number of the user, the language of the user
The ratio of the language total duration that the average duration of section, the language total duration of the user are represented with the customer service,
One or more in the ratio of the speech segment number that the speech segment number of the user is represented with the customer service；
The customer service represents dialogue rhythm characteristic at least includes language total duration, the visitor that the customer service is represented
Take one or more in speech segment number, the average duration of the speech segment of customer service representative for representing.
Alternatively, the device 200 also includes：
Second extraction module 201, for the recording data according to known satisfaction evaluation result, obtains described
It is special that the user session rhythm characteristic of the recording data of known satisfaction evaluation result and customer service represent dialogue rhythm
Levy；
Model building module 203, for the user of the recording data according to the known satisfaction evaluation result
Dialogue rhythm characteristic and customer service represent dialogue rhythm characteristic, and the satisfaction evaluation is determined using SVM algorithm
Model.
Alternatively, the satisfaction evaluation model includes representing satisfied the first support vector evaluated and represents not
Satisfied the second support vector evaluated；
The model building module 203 includes：
Second acquisition unit 2031, for the use of the recording data according to the known satisfaction evaluation result
Rhythm characteristic is talked with family and customer service represents dialogue rhythm characteristic, is obtained first and is talked with rhythm characteristic vector, described
The dimension of the first dialogue rhythm characteristic vector is equal to the use of the recording data of the known satisfaction evaluation result
Rhythm characteristic is talked with family and customer service represents the number sum of dialogue rhythm characteristic；
Model sets up unit 2032, for according to multiple the first dialogue rhythm characteristic vector, using institute
The Polynomial kernel function of SVM algorithm is stated, first support vector and second support vector is determined.
Alternatively, the evaluation module 207 includes：
3rd acquiring unit 2071, for the user session rhythm characteristic according to the recording data to be evaluated
Dialogue rhythm characteristic is represented with customer service, second is obtained and is talked with rhythm characteristic vector, the second dialogue rhythm is special
The dimension for levying vector is equal to user session rhythm characteristic and the customer service representative dialogue of the recording data to be evaluated
The number sum of rhythm characteristic；
4th acquiring unit 2072, for obtaining, the second dialogue rhythm characteristic is vectorial supports with described first
The first distance between vector, and the second dialogue rhythm characteristic it is vectorial with second support vector it
Between second distance；
First judging unit 2073, for whether judging first distance less than the first predetermined threshold value, if
It is that then the satisfaction evaluation result of the recording data to be evaluated is satisfaction；
Second judging unit 2074, for when first judging unit 2073 judged result for it is no when,
Whether the second distance is judged less than the second predetermined threshold value, if the second distance is preset less than described second
Threshold value, then the satisfaction evaluation result of the recording data to be evaluated is dissatisfied.
Alternatively, second extraction module 201 includes：
5th acquiring unit 2011, it is pre- for obtaining second from the recording data of the known evaluation result
If the recording data of duration；
Second cutting unit 2012, engages in the dialogue both sides for the recording data to second preset duration
Phonetic segmentation, the second voice segments collection that the second voice segments collection and customer service for obtaining user are represented；
6th acquiring unit 2013, for the second voice segments collection according to the user, obtains described known
The user session rhythm characteristic of the recording data of satisfaction evaluation result, and the represented according to the customer service
Two voice segments collection, the customer service for obtaining the recording data of the known satisfaction evaluation result represents dialogue rhythm spy
Levy.
Above-described is the preferred embodiment of the present invention, it should be pointed out that for the ordinary people of the art
For member, some improvements and modifications can also be made under the premise of principle of the present invention is not departed from, these
Improvements and modifications are also within the scope of the present invention.
Claims (14)
1. a kind of method of the recording data for processing service calls, records in the recording data of the service calls
There is the dialogue that user and customer service are represented, it is characterised in that methods described includes：
According to recording data to be evaluated, user session rhythm characteristic and the visitor of the recording data to be evaluated are obtained
Clothes represent dialogue rhythm characteristic；
User session rhythm characteristic and customer service according to the recording data to be evaluated represent dialogue rhythm characteristic,
Using predetermined satisfaction evaluation model, the satisfaction evaluation result of the recording data to be evaluated is determined.
2. method according to claim 1, it is characterised in that described according to recording data to be evaluated,
The user session rhythm characteristic and customer service for obtaining the recording data to be evaluated represent dialogue rhythm characteristic, including：
The recording data of the first preset duration is obtained from the recording data to be evaluated；
Engage in the dialogue both sides' phonetic segmentation to the recording data of first preset duration, obtains the first of user
The first voice segments collection that voice segments collection and customer service are represented；
The first voice segments collection according to the user, obtains the user session rhythm of the recording data to be evaluated
Feature, and the first voice segments collection represented according to the customer service, obtain the visitor of the recording data to be evaluated
Clothes represent dialogue rhythm characteristic.
3. method according to claim 2, it is characterised in that the user session rhythm characteristic is at least
Jing Yin total duration that recording data including first preset duration, first preset duration includes,
Language total duration, the recording of first preset duration that the recording data of first preset duration includes
Include Jing Yin section of Jing Yin hop count that data include, the recording data of first preset duration it is flat
Hop count, described first of all speech segments that equal duration, the recording data of first preset duration include
The average duration of the speech segment that the recording data of preset duration includes, the language total duration of the user, institute
State the speech segment number of user, the average duration of the speech segment of the user, the language total duration of the user with
If the ratio of language total duration of the customer service representative, the speech segment number of the user are represented with the customer service
One or more in the ratio of paragraph number；
The customer service represents dialogue rhythm characteristic at least includes language total duration, the visitor that the customer service is represented
Take one or more in speech segment number, the average duration of the speech segment of customer service representative for representing.
4. method according to claim 1, it is characterised in that according to the recording data to be evaluated
User session rhythm characteristic and customer service represent dialogue rhythm characteristic, using predetermined satisfaction evaluation mould
Type, before determining the satisfaction evaluation result of the recording data to be evaluated, methods described also includes：
According to the recording data of known satisfaction evaluation result, the record of the known satisfaction evaluation result is obtained
The user session rhythm characteristic of sound data and customer service represent dialogue rhythm characteristic；
The user session rhythm characteristic of the recording data according to the known satisfaction evaluation result and customer service generation
Table talks with rhythm characteristic, and the satisfaction evaluation model is determined using support vector machines algorithm.
5. method according to claim 4, it is characterised in that the satisfaction evaluation model includes generation
Be satisfied with evaluation the first support vector and represent dissatisfied the second support vector evaluated；
The user session rhythm characteristic of the recording data according to the known satisfaction evaluation result and visitor
Clothes represent dialogue rhythm characteristic, and the satisfaction evaluation model is determined using support vector machines algorithm,
Including：
The user session rhythm characteristic of the recording data according to the known satisfaction evaluation result and customer service generation
Table talks with rhythm characteristic, obtains first and talks with rhythm characteristic vector, the first dialogue rhythm characteristic vector
Dimension is equal to user session rhythm characteristic and the customer service generation of the recording data of the known satisfaction evaluation result
Table talks with the number sum of rhythm characteristic；
According to multiple the first dialogue rhythm characteristic vector, using the polynomial kernel letter of the SVM algorithm
Number, determines first support vector and second support vector.
6. method according to claim 5, it is characterised in that described according to the recording number to be evaluated
According to user session rhythm characteristic and customer service represent dialogue rhythm characteristic, using predetermined satisfaction evaluation
Model, determines the satisfaction evaluation result of the recording data to be evaluated, including：
User session rhythm characteristic and customer service according to the recording data to be evaluated represent dialogue rhythm characteristic,
Obtain second and talk with rhythm characteristic vector, the vectorial dimension of the second dialogue rhythm characteristic is equal to described to be evaluated
The user session rhythm characteristic of valency recording data and customer service represent the number sum of dialogue rhythm characteristic；
Obtain institute second and talk with that rhythm characteristic is vectorial and the first distance between first support vector, and
The second dialogue rhythm characteristic is vectorial and second distance between second support vector；
Whether first distance is judged less than the first predetermined threshold value, if so, the then recording data to be evaluated
Satisfaction evaluation result for satisfaction；
If it is not, whether the second distance is then judged less than the second predetermined threshold value, if the second distance is less than
Second predetermined threshold value, then the satisfaction evaluation result of the recording data to be evaluated is dissatisfied.
7. method according to claim 4, it is characterised in that satisfaction evaluation knot known to the basis
The recording data of fruit, obtains the user session rhythm of the recording data of the known satisfaction evaluation result
Feature and customer service represent dialogue rhythm characteristic, including：
The recording data of the second preset duration is obtained from the recording data of the known evaluation result；
Engage in the dialogue both sides' phonetic segmentation to the recording data of second preset duration, obtains the second of user
The second voice segments collection that voice segments collection and customer service are represented；
The second voice segments collection according to the user, obtains the recording data of the known satisfaction evaluation result
User session rhythm characteristic, and the second voice segments collection represented according to the customer service obtains described known
The customer service of the recording data of satisfaction evaluation result represents dialogue rhythm characteristic.
8. a kind of device of the recording data for processing service calls, records in the recording data of the service calls
There is the dialogue that user and customer service are represented, it is characterised in that described device includes：
First extraction module, for according to recording data to be evaluated, obtaining the use of the recording data to be evaluated
Rhythm characteristic is talked with family and customer service represents dialogue rhythm characteristic；
Evaluation module, represents for the user session rhythm characteristic according to the recording data to be evaluated and customer service
Dialogue rhythm characteristic, using predetermined satisfaction evaluation model, determines the recording data to be evaluated
Satisfaction evaluation result.
9. device according to claim 8, it is characterised in that first extraction module includes：
First acquisition unit, the recording number for obtaining the first preset duration from the recording data to be evaluated
According to；
First cutting unit, cuts for the both sides' voice that engages in the dialogue of the recording data to first preset duration
Point, the first voice segments collection that the first voice segments collection and customer service for obtaining user are represented；
Feature acquiring unit, for the first voice segments collection according to the user, obtains the recording to be evaluated
The user session rhythm characteristic of data, and the first voice segments collection represented according to the customer service, obtain described
The customer service of recording data to be evaluated represents dialogue rhythm characteristic.
10. device according to claim 9, it is characterised in that the user session rhythm characteristic is extremely
Include less Jing Yin total duration that first preset duration, the recording data of first preset duration include,
Language total duration, the recording of first preset duration that the recording data of first preset duration includes
Include Jing Yin section of Jing Yin hop count that data include, the recording data of first preset duration it is flat
Hop count, described first of all speech segments that equal duration, the recording data of first preset duration include
The average duration of the speech segment that the recording data of preset duration includes, the language total duration of the user, institute
State the speech segment number of user, the average duration of the speech segment of the user, the language total duration of the user with
If the ratio of language total duration of the customer service representative, the speech segment number of the user are represented with the customer service
One or more in the ratio of paragraph number；
The customer service represents dialogue rhythm characteristic at least includes language total duration, the visitor that the customer service is represented
Take one or more in speech segment number, the average duration of the speech segment of customer service representative for representing.
11. devices according to claim 8, it is characterised in that described device also includes：
Second extraction module, for the recording data according to known satisfaction evaluation result, obtains described known
The user session rhythm characteristic of the recording data of satisfaction evaluation result and customer service represent dialogue rhythm characteristic；
Model building module, for the user session of the recording data according to the known satisfaction evaluation result
Rhythm characteristic and customer service represent dialogue rhythm characteristic, and the satisfaction is determined using support vector machines algorithm
Degree evaluation model.
12. devices according to claim 11, it is characterised in that the satisfaction evaluation model includes
Represent satisfied the first support vector evaluated and represent dissatisfied the second support vector evaluated；
The model building module includes：
Second acquisition unit, for the user session of the recording data according to the known satisfaction evaluation result
Rhythm characteristic and customer service represent dialogue rhythm characteristic, obtain first and talk with rhythm characteristic vector, described first pair
The dimension of words rhythm characteristic vector is equal to the user session of the recording data of the known satisfaction evaluation result
Rhythm characteristic and customer service represent the number sum of dialogue rhythm characteristic；
Model sets up unit, for according to multiple the first dialogue rhythm characteristic vector, using the SVM
The Polynomial kernel function of algorithm, determines first support vector and second support vector.
13. devices according to claim 12, it is characterised in that the evaluation module includes：
3rd acquiring unit, for the user session rhythm characteristic according to the recording data to be evaluated and customer service
Dialogue rhythm characteristic is represented, second is obtained and is talked with rhythm characteristic vector, the second dialogue rhythm characteristic vector
Dimension be equal to the user session rhythm characteristic of the recording data to be evaluated and customer service to represent dialogue rhythm special
The number sum levied；
4th acquiring unit, for obtain institute second talk with rhythm characteristic it is vectorial with first support vector it
Between the first distance, and the second dialogue rhythm characteristic it is vectorial and second support vector between the
Two distances；
First judging unit, for whether judging first distance less than the first predetermined threshold value, if so, then
The satisfaction evaluation result of the recording data to be evaluated is satisfaction；
Second judging unit, for when first judging unit judged result for it is no when, judge described the
Whether two distances are less than the second predetermined threshold value, if the second distance is less than second predetermined threshold value, institute
It is dissatisfied to state the satisfaction evaluation result of recording data to be evaluated.
14. devices according to claim 11, it is characterised in that second extraction module includes：
5th acquiring unit, for obtaining the second preset duration from the recording data of the known evaluation result
Recording data；
Second cutting unit, cuts for the both sides' voice that engages in the dialogue of the recording data to second preset duration
Point, the second voice segments collection that the second voice segments collection and customer service for obtaining user are represented；
6th acquiring unit, for the second voice segments collection according to the user, obtains the known satisfaction
The user session rhythm characteristic of the recording data of evaluation result, and the second voice represented according to the customer service
Section collection, the customer service for obtaining the recording data of the known satisfaction evaluation result represents dialogue rhythm characteristic.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510883478.8A CN106847305B (en) | 2015-12-04 | 2015-12-04 | Method and device for processing recording data of customer service telephone |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510883478.8A CN106847305B (en) | 2015-12-04 | 2015-12-04 | Method and device for processing recording data of customer service telephone |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106847305A true CN106847305A (en) | 2017-06-13 |
| CN106847305B CN106847305B (en) | 2020-09-18 |
Family
ID=59150289
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510883478.8A Active CN106847305B (en) | 2015-12-04 | 2015-12-04 | Method and device for processing recording data of customer service telephone |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106847305B (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108564968A (en) * | 2018-04-26 | 2018-09-21 | 广州势必可赢网络科技有限公司 | A kind of method and device of evaluation customer service |
| CN111049999A (en) * | 2018-10-11 | 2020-04-21 | 上海智臻智能网络科技股份有限公司 | Voice customer service quality inspection system and customer service quality inspection equipment |
| CN111049998A (en) * | 2018-10-11 | 2020-04-21 | 上海智臻智能网络科技股份有限公司 | Voice customer service quality inspection method, customer service quality inspection equipment and storage medium |
| CN111221953A (en) * | 2020-01-08 | 2020-06-02 | 南京新贝金服科技有限公司 | Online pre-sale customer service effect evaluation method |
| CN111314566A (en) * | 2020-01-20 | 2020-06-19 | 北京神州泰岳智能数据技术有限公司 | Voice quality inspection method, device and system |
| CN111783441A (en) * | 2020-08-08 | 2020-10-16 | 中国人民解放军国防科技大学 | Dialog statement correctness evaluation method based on support vector machine |
| CN111816161A (en) * | 2020-07-28 | 2020-10-23 | 深圳大学 | Customer service dialogue quality detection method and system based on NLP |
| CN112036923A (en) * | 2020-07-06 | 2020-12-04 | 北京嘀嘀无限科技发展有限公司 | Service evaluation method, system, device and storage medium |
| CN112507072A (en) * | 2020-12-07 | 2021-03-16 | 上海明略人工智能(集团)有限公司 | Sale evaluation method and system based on conversation and electronic equipment |
| CN114067842A (en) * | 2021-11-17 | 2022-02-18 | 中国银行股份有限公司 | Customer satisfaction degree identification method and device, storage medium and electronic equipment |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101662550A (en) * | 2009-09-11 | 2010-03-03 | 中兴通讯股份有限公司 | Method and system for service quality detection for call center |
| WO2012124104A1 (en) * | 2011-03-17 | 2012-09-20 | 富士通株式会社 | Operator evaluation support device, operator evaluation support method, and storage medium having operator evaluation support program recorded therein |
| CN103811009A (en) * | 2014-03-13 | 2014-05-21 | 华东理工大学 | Smart phone customer service system based on speech analysis |
| CN104301554A (en) * | 2013-07-18 | 2015-01-21 | 中兴通讯股份有限公司 | Device and method used for detecting service quality of customer service staff |
-
2015
- 2015-12-04 CN CN201510883478.8A patent/CN106847305B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101662550A (en) * | 2009-09-11 | 2010-03-03 | 中兴通讯股份有限公司 | Method and system for service quality detection for call center |
| WO2012124104A1 (en) * | 2011-03-17 | 2012-09-20 | 富士通株式会社 | Operator evaluation support device, operator evaluation support method, and storage medium having operator evaluation support program recorded therein |
| CN104301554A (en) * | 2013-07-18 | 2015-01-21 | 中兴通讯股份有限公司 | Device and method used for detecting service quality of customer service staff |
| CN103811009A (en) * | 2014-03-13 | 2014-05-21 | 华东理工大学 | Smart phone customer service system based on speech analysis |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108564968A (en) * | 2018-04-26 | 2018-09-21 | 广州势必可赢网络科技有限公司 | A kind of method and device of evaluation customer service |
| CN111049999A (en) * | 2018-10-11 | 2020-04-21 | 上海智臻智能网络科技股份有限公司 | Voice customer service quality inspection system and customer service quality inspection equipment |
| CN111049998A (en) * | 2018-10-11 | 2020-04-21 | 上海智臻智能网络科技股份有限公司 | Voice customer service quality inspection method, customer service quality inspection equipment and storage medium |
| CN111221953A (en) * | 2020-01-08 | 2020-06-02 | 南京新贝金服科技有限公司 | Online pre-sale customer service effect evaluation method |
| CN111221953B (en) * | 2020-01-08 | 2024-03-08 | 南京新贝金服科技有限公司 | Online pre-sale customer service effect evaluation method |
| CN111314566A (en) * | 2020-01-20 | 2020-06-19 | 北京神州泰岳智能数据技术有限公司 | Voice quality inspection method, device and system |
| CN112036923A (en) * | 2020-07-06 | 2020-12-04 | 北京嘀嘀无限科技发展有限公司 | Service evaluation method, system, device and storage medium |
| CN111816161A (en) * | 2020-07-28 | 2020-10-23 | 深圳大学 | Customer service dialogue quality detection method and system based on NLP |
| CN111783441A (en) * | 2020-08-08 | 2020-10-16 | 中国人民解放军国防科技大学 | Dialog statement correctness evaluation method based on support vector machine |
| CN112507072A (en) * | 2020-12-07 | 2021-03-16 | 上海明略人工智能(集团)有限公司 | Sale evaluation method and system based on conversation and electronic equipment |
| CN114067842A (en) * | 2021-11-17 | 2022-02-18 | 中国银行股份有限公司 | Customer satisfaction degree identification method and device, storage medium and electronic equipment |
| CN114067842B (en) * | 2021-11-17 | 2024-01-30 | 中国银行股份有限公司 | Customer satisfaction degree identification method and device, storage medium and electronic equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106847305B (en) | 2020-09-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106847305A (en) | A kind of method and device of the recording data for processing service calls | |
| CN107818798B (en) | Customer service quality evaluation method, device, equipment and storage medium | |
| US10665253B2 (en) | Voice activity detection using a soft decision mechanism | |
| CN108962282B (en) | Voice detection analysis method and device, computer equipment and storage medium | |
| US9875739B2 (en) | Speaker separation in diarization | |
| US9672825B2 (en) | Speech analytics system and methodology with accurate statistics | |
| WO2021128741A1 (en) | Voice emotion fluctuation analysis method and apparatus, and computer device and storage medium | |
| CN108682420B (en) | Audio and video call dialect recognition method and terminal equipment | |
| CN111739516A (en) | Speech recognition system for intelligent customer service call | |
| CN111797632B (en) | Information processing method and device and electronic equipment | |
| CN104538043A (en) | Real-time emotion reminder for call | |
| MX2008016354A (en) | Detecting an answering machine using speech recognition. | |
| US20210027769A1 (en) | Voice alignment method and apparatus | |
| CN110060665A (en) | Word speed detection method and device, readable storage medium storing program for executing | |
| CN110704618B (en) | Method and device for determining standard problem corresponding to dialogue data | |
| US20210118464A1 (en) | Method and apparatus for emotion recognition from speech | |
| CN111091809A (en) | Regional accent recognition method and device based on depth feature fusion | |
| CN112802498A (en) | Voice detection method and device, computer equipment and storage medium | |
| CN113112992A (en) | Voice recognition method and device, storage medium and server | |
| CN110580899A (en) | Voice recognition method and device, storage medium and computing equipment | |
| CN109325155A (en) | A kind of novel dialogue state storage method and system | |
| US20130297311A1 (en) | Information processing apparatus, information processing method and information processing program | |
| CN114579751A (en) | Emotion analysis method and device, electronic equipment and storage medium | |
| US20110197206A1 (en) | System, Method And Program Product For Analyses Based On Agent-Customer Interactions And Concurrent System Activity By Agents | |
| CN109378007B (en) | Method for realizing gender recognition based on intelligent voice conversation |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |