CN107886233B - Service quality evaluation method and system for customer service - Google Patents

Service quality evaluation method and system for customer service Download PDF

Info

Publication number
CN107886233B
CN107886233B CN201711077542.9A CN201711077542A CN107886233B CN 107886233 B CN107886233 B CN 107886233B CN 201711077542 A CN201711077542 A CN 201711077542A CN 107886233 B CN107886233 B CN 107886233B
Authority
CN
China
Prior art keywords
service
dialogues
dialogue
process nodes
standard
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.)
Active
Application number
CN201711077542.9A
Other languages
Chinese (zh)
Other versions
CN107886233A (en
Inventor
李坤
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
GCI Science and Technology Co Ltd
Original Assignee
GCI Science and Technology Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by GCI Science and Technology Co Ltd filed Critical GCI Science and Technology Co Ltd
Priority to CN201711077542.9A priority Critical patent/CN107886233B/en
Publication of CN107886233A publication Critical patent/CN107886233A/en
Application granted granted Critical
Publication of CN107886233B publication Critical patent/CN107886233B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
    • G06Q10/06395—Quality analysis or management
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
    • G06F16/33—Querying
    • G06F16/3331—Query processing
    • G06F16/334—Query execution
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
    • G06Q10/06398—Performance of employee with respect to a job function
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/01—Customer relationship services
    • G06Q30/015—Providing customer assistance, e.g. assisting a customer within a business location or via helpdesk
    • G06Q30/016—After-sales

Landscapes

  • Business, Economics & Management (AREA)
  • Engineering & Computer Science (AREA)
  • Human Resources & Organizations (AREA)
  • Theoretical Computer Science (AREA)
  • Strategic Management (AREA)
  • Economics (AREA)
  • Development Economics (AREA)
  • General Physics & Mathematics (AREA)
  • Educational Administration (AREA)
  • Physics & Mathematics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • General Business, Economics & Management (AREA)
  • Marketing (AREA)
  • Game Theory and Decision Science (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • Computational Linguistics (AREA)
  • General Engineering & Computer Science (AREA)
  • Accounting & Taxation (AREA)
  • Finance (AREA)
  • Machine Translation (AREA)

Abstract

The invention discloses a method for evaluating the service quality of customer service, which comprises the following steps: extracting all service dialogues of the customer service in the dialog to be analyzed; acquiring all process nodes of all service dialogues according to all process nodes of a standard service process and all service dialogues; calculating the flow matching degree of all flow nodes of all service dialogues and a standard service flow; and evaluating the service quality of the customer service according to the flow matching degree. The method for evaluating the service quality of the customer service improves the accuracy and efficiency of evaluating the service quality of the customer service, and simultaneously provides a system for evaluating the service quality of the customer service.

Description

Service quality evaluation method and system for customer service
Technical Field
The invention relates to the technical field of data processing, in particular to a method and a system for evaluating the service quality of customer service.
Background
In order to ensure that the enterprise provides a high service level for the customer, the service quality of the customer service generally needs to be evaluated. At present, a service quality evaluation method of customer service adopts a manual mode, and monitors the service flow of customer service personnel and gives certain evaluation by following the recording of the customer service personnel in the working process.
In the process of implementing the invention, the inventor finds that the existing service flow quality inspection method has the following defects:
the existing service flow quality inspection method depends on manpower, has low inspection efficiency, and is easy to have the conditions of missed listening, wrong listening and the like in the recording following process, thereby causing the quality inspection result to lose the accuracy.
Disclosure of Invention
The invention provides a method and a system for evaluating the service quality of customer service, which improve the accuracy of the method for evaluating the service quality of customer service and reduce the implementation difficulty.
One aspect of the present invention provides a method for evaluating service quality of customer service, where the method includes:
extracting all service dialogues of the customer service in the dialog to be analyzed;
acquiring all process nodes of all service dialogues according to all process nodes of a standard service process and all service dialogues;
calculating the flow matching degree of all flow nodes of all service dialogues and a standard service flow;
and evaluating the service quality of the customer service according to the flow matching degree.
In an optional implementation manner, the acquiring all process nodes of all service dialogs according to all process nodes of a standard service process and all service dialogs includes:
extracting all keywords of each sentence of service dialogue from all the service dialogues respectively;
matching all process nodes of the standard service process with service dialogues corresponding to the process nodes in all service dialogues according to the process nodes and all keywords of all the service dialogues respectively;
and acquiring all the process nodes of all the service dialogs according to the service dialogs corresponding to the process nodes in all the service dialogs.
In an optional implementation manner, the matching, for all process nodes of the standard service process, service dialogs corresponding to the process nodes in all service dialogs according to the process nodes and all keywords of the all service dialogs respectively includes:
inputting all the keywords of the process nodes and all the service dialogs to a conceptual diagram tool configured according to a process corpus in advance for all the process nodes of the standard service process respectively so as to match the service dialogs corresponding to the process nodes in all the service dialogs; the process corpus comprises all process nodes of the standard service process and corpora corresponding to all the process nodes of the standard service process.
In an optional implementation manner, the calculating a flow matching degree of all flow nodes of all service dialogues with a standard service flow includes:
acquiring first preset weights corresponding to all process nodes of all service dialogues one to one; wherein, the sum of the first preset weights of all process nodes of the standard service process is equal to 1;
all the preset operation behaviors of the corresponding process nodes are respectively obtained for all the process nodes of all the service dialogues;
respectively acquiring customer service operation behaviors of corresponding process nodes for all process nodes of all service dialogues;
respectively acquiring corresponding operation scores of the process nodes according to all preset operation behaviors of all process nodes of all service dialogues and all acquired customer service operation behaviors;
respectively calculating products of the corresponding operation scores and the corresponding first preset weights for all the process nodes of all the service dialogues to obtain matching scores of the process nodes of all the service dialogues;
and adding the matching scores of the flow nodes of all the service pairs to obtain the flow matching degree.
In an optional implementation manner, the obtaining, according to all preset operation behaviors of all process nodes of all service dialogues and all obtained customer service operation behaviors, operation scores of corresponding process nodes respectively includes:
respectively acquiring corresponding second preset weights for all preset operation behaviors of all process nodes of all service dialogues; wherein the sum of the second preset weights of all different preset operation behaviors corresponding to the standard service flow is equal to 1;
respectively acquiring preset operation behaviors which are the same as the customer service operation behaviors for all the customer service operation behaviors of all the process nodes of all the service dialogues;
and adding second preset weights of preset operation behaviors which are the same as the customer service operation behaviors to all the process nodes of all the service dialogues respectively to obtain operation scores of the process nodes respectively.
The invention also provides a service quality evaluation system of customer service, which comprises:
the extraction module is used for extracting all service dialogues of the customer service in the dialog to be analyzed;
the acquisition module is used for acquiring all the process nodes of all the service dialogues according to all the process nodes of a standard service process and all the service dialogues;
the calculation module is used for calculating the flow matching degree of all the flow nodes of all the service dialogues and the standard service flow;
and the evaluation module is used for evaluating the service quality of the customer service according to the flow matching degree.
In an optional implementation, the obtaining module includes:
a keyword extraction unit, configured to extract all keywords of each sentence of service dialogue for all the service dialogues respectively;
the keyword matching unit is used for matching all the process nodes of the standard service process with the service dialogues corresponding to the process nodes in all the service dialogues according to the process nodes and all the keywords of all the service dialogues respectively;
and the process node acquisition unit is used for acquiring all the process nodes of all the service dialogs according to the service dialogs corresponding to the process nodes in all the service dialogs.
In an alternative embodiment, the keyword matching unit includes:
the service dialogue matching unit is used for inputting all the process nodes of the standard service process and all the keywords of all the service dialogues into a conceptual diagram tool configured according to a process corpus in advance respectively so as to match the service dialogues corresponding to the process nodes in all the service dialogues; the process corpus comprises all process nodes of the standard service process and corpora corresponding to all the process nodes of the standard service process.
In an alternative embodiment, the calculation module comprises:
a first obtaining unit, configured to obtain first preset weights in one-to-one correspondence with all process nodes of all service dialogues; wherein, the sum of the first preset weights of all process nodes of the standard service process is equal to 1;
a second obtaining unit, configured to obtain, for all the process nodes of all the service dialogues, all the preset operation behaviors of the corresponding process node respectively;
a third obtaining unit, configured to obtain, for all the process nodes in all the service dialogues, customer service operation behaviors of corresponding process nodes respectively;
a fourth obtaining unit, configured to obtain operation scores of corresponding process nodes according to all preset operation behaviors of all process nodes of all service dialogues and all obtained customer service operation behaviors;
the first calculation unit is used for calculating products of the corresponding operation scores and the corresponding first preset weights for all the process nodes of all the service dialogues respectively so as to obtain the matching scores of the process nodes of all the service dialogues;
and the first adding unit is used for adding the matching scores of the flow nodes of all the service dialogues to obtain the flow matching degree.
In an optional implementation manner, corresponding second preset weights are respectively obtained for all preset operation behaviors of all process nodes of all service dialogues; wherein the sum of the second preset weights of all different preset operation behaviors corresponding to the standard service flow is equal to 1;
a second obtaining subunit, configured to obtain, for all customer service operation behaviors of all process nodes of all service dialogues, preset operation behaviors that are the same as the customer service operation behaviors respectively;
and the first adding subunit is configured to add, to all the process nodes of all the service dialogues, second preset weights of preset operation behaviors that are the same as the customer service operation behaviors, so as to obtain operation scores of the process nodes respectively.
Compared with the prior art, the invention has the following outstanding beneficial effects: according to the method and the system for evaluating the service quality of the customer service, all service dialogues of the customer service are extracted from the dialog to be analyzed, so that all the dialogues of the customer service and the customer are prevented from being acquired by the process nodes, the data processing amount is reduced, and the efficiency of evaluating the service quality of the customer service is improved; all the process nodes of all the service dialogues are obtained according to all the process nodes of the standard service process and all the service dialogues, so that the accuracy of process matching is improved, the accuracy of service quality evaluation of customer service is improved, the subjective influence caused by a manual mode is avoided, and meanwhile, the quality inspection efficiency is improved; the quality of the service process of the customer service is evaluated through the process matching degree, the intellectualization of the service quality evaluation of the customer service is realized, and compared with a method for evaluating only through the number of process nodes, the objectivity of quality inspection is improved.
Drawings
FIG. 1 is a schematic flow chart diagram illustrating an embodiment of a method for evaluating quality of service of customer service provided by the present invention;
fig. 2 is a schematic structural diagram of an embodiment of a service quality evaluation system for customer service provided by the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Referring to fig. 1, it is a schematic flow chart of a first embodiment of a method for evaluating quality of service of customer service provided by the present invention, where the method includes:
s101, extracting all service dialogues of customer service in a dialog to be analyzed;
s102, acquiring all process nodes of all service dialogues according to all process nodes of a standard service process and all service dialogues;
s103, calculating the process matching degree of all process nodes of all service dialogues and a standard service process;
and S104, evaluating the service quality of the customer service according to the process matching degree. It should be noted that all process nodes of the standard service process cover process nodes corresponding to all service dialogues of the customer service.
In an optional embodiment, the method further comprises: before all service dialogues of the customer service are extracted from the dialog to be analyzed, acquiring a dialog voice between the customer service and the client; and converting the dialogue voice into text to obtain the dialogue to be analyzed.
For example, the dialog to be analyzed between an after-sales service and a customer of a certain air conditioner is:
customer service: you are nice and happy to serve you
Customer: that air conditioner of my home can not be opened
Customer service: is the air conditioner can not be started
Customer: jone (a Chinese character)
Customer service: do you provide your order number at a glance
Customer: 189383983
Customer service: you will, according to the situation that you reflect, handle the appointment of after-sale detection for you
Customer service: asking you to be mr. in river, grand people and river
Customer: to pair
Customer service: good, address is confirmed with you: XX city XX county XX district XX road XX building
Customer: to pair
If the standard service flow includes: greeting, confirming after-sale problem, checking order number of goods, confirming details of goods and after-sale maintenance appointment.
The mapping relation between all service dialogues of the customer service and the process nodes is as follows:
you are nice and happy to serve you
Does not start the air conditioner (flow node: confirm after-sale problem)
Just provide your order number (flow node: check commodity order number)
You can do the after-sales service reservation (flow node: after-sales service reservation) according to the condition reflected by you
Asking you is Mr. of the river (flow node: maintenance appointment after sale)
Good, address is confirmed with you: XX city XX county XX district XX road XX number building (flow node: after-sale maintenance reservation)
It should be noted that the standard service flow, the mapping relationship between the service dialog and the flow node, and the standard dialog are only examples, and the mapping relationship between the service dialog and the flow node is subject to the actual application.
All service dialogs of the customer service are extracted from the dialog to be analyzed, so that the process node acquisition of all the dialogs of the customer service and the customer is avoided, the data processing amount is reduced, and the service quality evaluation efficiency of the customer service is improved; all the process nodes of all the service dialogues are obtained according to all the process nodes of the standard service process and all the service dialogues, so that the accuracy of process matching is improved, the accuracy of service quality evaluation of customer service is improved, the subjective influence caused by a manual mode is avoided, and meanwhile, the quality inspection efficiency is improved; the quality of the service process of the customer service is evaluated through the process matching degree, the intellectualization of the service quality evaluation of the customer service is realized, and compared with a method for evaluating only through the number of process nodes, the objectivity of quality inspection is improved.
In an alternative embodiment, the extracting all service dialogs of the customer service in the dialog to be analyzed includes:
acquiring all discriminative words of the dialogue to be analyzed according to the dialogue to be analyzed and the discriminative word library; the distinguishing word library comprises distinguishing words which are obtained in advance and the number of which is a first set number;
acquiring labels of dialog roles corresponding to dialogues of the dialogues to be analyzed according to all discriminative words of the dialogues to be analyzed and a dialog role discrimination model established in advance according to a dialog corpus; the dialogue corpus comprises a plurality of standard dialogs and labels of dialogue roles corresponding to the standard dialogs;
and extracting all dialogs with the labels of the conversation roles in the conversation to be analyzed as customer service dialogs, and taking the dialogs as all service dialogs of the customer service.
It should be noted that, in practical application, the dialog corpus is a dialog corpus in the field to which the dialog to be analyzed belongs; the standard dialogs of the dialog corpus are the dialogs of all dialogs contained in the dialog corpus; the labels of the dialog roles corresponding to the standard dialogs of each sentence stored in the dialog corpus should include labels of all the dialog roles of the dialogs of the dialog to be analyzed; for example, if the dialog role of the dialog to be analyzed includes customer service and customer, the dialog corpus stores the label of the dialog role corresponding to the standard dialog as the label of the customer service or the customer.
The method comprises the steps that the discriminative words of the dialogue to be analyzed are obtained through a discriminative word library, so that the complexity of judging the discriminative words in the dialogue to be analyzed is reduced, and the processing efficiency is improved; by combining the dialogue corpus, the dialogue roles are distinguished according to the dialogue to be analyzed, the accuracy of dialogue role feature extraction is improved, a more comprehensive dialogue corpus is provided to obtain a dialogue role distinguishing model of the dialogue corpus, so that the labels of the dialogue roles are identified more accurately, and the accuracy of dialogue role distinguishing is improved.
In an optional implementation manner, the obtaining all the discriminative words of the dialog to be analyzed according to the dialog to be analyzed and the discriminative word library includes:
preprocessing the dialogue to be analyzed to obtain all words of the dialogue to be analyzed;
and acquiring the distinguished words of the dialogue to be analyzed according to all the words of the dialogue to be analyzed and all the distinguished words of the distinguished word library.
In an optional embodiment, the preprocessing the dialog of the dialog to be analyzed to obtain all the words of the dialog to be analyzed includes:
and performing word segmentation and odd-different word replacement on the dialogue to be analyzed to obtain all words of the dialogue to be analyzed.
In an optional implementation manner, the obtaining all the discriminative words of the dialog to be analyzed according to the dialog to be analyzed and the discriminative word library includes:
segmenting the dialogue of the dialogue to be analyzed to obtain all words of the dialogue to be analyzed;
and matching all the words of the dialogue to be analyzed with all the discriminative words of the discriminative word library to obtain the discriminative words of the dialogue to be analyzed.
The dialogue system comprises a dialogue analysis module, a dialogue analysis module and a dialogue analysis module, wherein the dialogue analysis module is used for analyzing dialogue components, the dialogue analysis module is used for analyzing dialogue components, the dialogue components are used for analyzing dialogue components, and the dialogue components are used for analyzing dialogue components.
In an optional implementation manner, the obtaining, according to all the discriminative words and the dialog corpus of the dialog to be analyzed, the label of the dialog role corresponding to the dialog of the dialog to be analyzed includes:
respectively acquiring word frequency of each discriminative word of the dialogue to be analyzed in the dialogue to be analyzed as a first group of parameters;
respectively acquiring the number of standard dialogs with corresponding differential words in a dialog corpus for each differential word of the dialogs to be analyzed;
for each discriminative word of the dialogue to be analyzed, respectively obtaining a second group of parameters according to the total number of standard dialogues of the dialogue corpus and the number of standard dialogues with corresponding discriminative words in the dialogue corpus; generating a feature vector of dialogue to be analyzed according to the first group of parameters and the second group of parameters;
and inputting the feature vector of the dialogue to be analyzed into a dialogue role discrimination model of the dialogue corpus so as to identify the label of the dialogue role corresponding to the dialogue to be analyzed.
In an alternative embodiment, the obtaining, for each of the discriminative words of the dialogs to be analyzed, a second set of parameters according to the total number of standard dialogs of the dialog corpus and the number of standard dialogs having corresponding discriminative words in the dialog corpus includes:
adding 1 to each discriminative word of the dialogue to be analyzed, wherein the number of standard dialogs with corresponding discriminative words in the dialogue corpus is used as a first effective denominator;
and respectively calculating the ratio of the total number of the standard dialogs of the dialog corpus to the first effective denominator as a second group of parameters for each discriminative word of the dialogs to be analyzed.
It should be noted that the word frequency refers to the number of times of occurrence of a word; the word frequency of all the discriminative words of the dialog to be analyzed in the dialog to be analyzed, i.e., the number of times all the discriminative words of the dialog to be analyzed appear in the dialog to be analyzed. Adding 1 to the number of standard dialogs in the corpus of dialogs having corresponding discriminative words such that the denominator of the second set of parameters is not 0.
In an alternative embodiment, the generating feature vectors of the dialog to be analyzed according to the first set of parameters and the second set of parameters includes:
carrying out logarithmic transformation on the second group of parameters to obtain a second group of parameters after logarithmic transformation;
and generating a feature vector of dialogue to be analyzed according to the first group of parameters and the logarithmically transformed second group of parameters.
The feature vectors are obtained through the word frequency and the ratio of the discriminative words in the dialogue to be analyzed, and then the labels of the dialogue roles are obtained through the dialogue role discrimination model established based on the dialogue corpus, so that the importance degree of the discriminative words in the dialogue to be analyzed and the importance degree of the discriminative words in the dialogue corpus are fully considered, and the accuracy of dialogue role discrimination is improved.
In an optional implementation manner, the obtaining, according to all the discriminative words of the dialog to be analyzed and a dialog role discrimination model established in advance according to a dialog corpus, a label of a dialog role corresponding to the dialog to be analyzed further includes:
before the first group of parameters are respectively obtained, calculating the quantity difference between the first set quantity and the quantity of the distinguished words of the dialogue to be analyzed;
generating feature vectors of dialogs of the dialog to be analyzed according to the first set of parameters and the second set of parameters includes:
generating filling word frequency with the number of the difference value; the filling word frequency is 0;
generating a filling ratio with the number being the number difference; the filling ratio is 0;
generating a first characteristic parameter with a first set quantity according to the filling word frequency and the first group of parameters;
generating second characteristic parameters with the first set quantity according to the filling ratio and the second set of parameter generation parameters;
and generating a feature vector of dialogue to be analyzed according to the first feature parameter and the second feature parameter.
For example, if the dialog to be analyzed is "hello, ask you for a recipient", and the words with distinction are "you" and "do", the word frequency of "you" in the dialog to be analyzed is 2, and the word frequency of "do" in the dialog to be analyzed is 1; if the total number of standard dialogs of all dialog characters in the dialog corpus is 100, it is assumed that the number of standard dialogs having the word "you" in the dialog corpus is 60, and the number of standard dialogs having the word "you" in the dialog corpus is 60The number of standard dialogs for the word "do" is 58, then the ratio of the total number of standard dialogs in the corpus of dialogs to the number of standard dialogs with the word "you" in the corpus of dialogs is
Figure GDA0002612001570000101
The ratio of the total number of standard dialogs in the corpus of dialogs to the number of standard dialogs having the word "do" in the corpus of dialogs is
Figure GDA0002612001570000111
Assuming that the first set number is 3, the first feature parameter is (2, 1, 0), the second feature parameter is (0.6, 0.58, 0), and the feature vector of the dialog to be analyzed for the dialog white is (2, 1, 0; 0.6, 0.58, 0).
The processing difficulty is reduced by generating filling word frequency and filling ratio and conveniently passing through the feature vectors with the determined parameter number; by generating the filling word frequency and the filling ratio of the null value, the substitution calculation of the parameters is reduced, the calculation speed is increased, and the processing efficiency is improved.
In an optional embodiment, the method further comprises:
responding to an instruction for training the conversation role distinguishing model, and respectively acquiring word frequencies of corresponding distinguishing words in each sentence of the standard spoken text of the conversation corpus as a third group of parameters for each distinguishing word of the distinguishing word library;
respectively acquiring the number of standard dialogues with corresponding differential words in the dialogue corpus for each differential word in the differential word library;
for each differentiated word in the differentiated word library, acquiring a fourth group of parameters according to the total number of standard dialogues in the dialogue corpus and the number of standard dialogues with the corresponding differentiated word in the dialogue corpus;
generating a feature vector of each sentence of the standard dialogue according to the third group of parameters and the fourth group of parameters;
and training the conversation role distinguishing model according to the feature vectors of the standard dialogs of each sentence and the labels of the conversation roles corresponding to the standard dialogs of each sentence based on a naive Bayes algorithm.
Wherein naive Bayes (A), (B)
Figure GDA0002612001570000112
Bayes) algorithm is a classification method based on bayesian theorem and independent assumptions of feature conditions.
In an alternative embodiment, the obtaining, for each distinct word in the distinct word library, a fourth set of parameters according to the total number of standard dialogs in the dialog corpus and the number of standard dialogs with corresponding distinct word in the dialog corpus includes:
adding 1 to each discriminative word in the discriminative word library respectively for the number of standard dialogues with the corresponding discriminative word in the dialogue corpus to serve as a second effective denominator;
and respectively calculating the ratio of the total number of standard dialogues of the dialogue corpus to the second effective denominator for each differential word of the differential word library to serve as a fourth group of parameters.
The conversation role distinguishing model is trained based on the naive Bayes algorithm, so that the contribution of each distinguishing word to the distinguishing of the conversation role can be accurately weighed, and the distinguishing accuracy of the conversation role is improved; adding 1 to the number of standard dialogs in the corpus of dialogs having corresponding discriminative terms so that the denominator of the second set of parameters is not 0.
In an optional embodiment, the method further comprises:
responding to an instruction for acquiring the differentiated word bank, and preprocessing all standard dialogues of the dialogue corpus to obtain all words of the dialogue corpus;
for each word in the dialogue corpus, respectively acquiring the word frequency of the corresponding word in each sentence of the standard dialogue in the dialogue corpus as a fifth group of parameters;
respectively acquiring the number of standard dialogues with corresponding words in the dialogue corpus for each word in the dialogue corpus;
for each word of the dialogue corpus, acquiring a sixth group of parameters according to the total number of standard dialogues of the dialogue corpus and the number of standard dialogues with corresponding words in the dialogue corpus;
generating a feature vector of each sentence of the standard dialogue according to the fifth group of parameters and the sixth group of parameters;
and selecting the words with the first set number from all the words in the dialogue corpus according to the feature vectors of the standard dialogues of each sentence and the labels of the dialogue roles corresponding to the standard dialogues of each sentence based on an information gain method so as to obtain the differential word library.
In an alternative embodiment, for each word of the dialog corpus, obtaining a sixth set of parameters according to the total number of standard dialogues of the dialog corpus and the number of standard dialogues with the corresponding word in the dialog corpus respectively includes:
adding 1 to each word in the dialogue corpus to serve as a third effective denominator, wherein the number of standard dialogues with the corresponding word in the dialogue corpus is 1;
and respectively calculating the ratio of the total number of standard dialogues of the dialogue corpus to the third effective denominator for each word of the dialogue corpus to serve as a sixth group of parameters.
The method selects the discriminative words through an information gain method, overcomes the subjective randomness and one-sidedness of manual screening, and accordingly enables the relevance between the screened discriminative words and the conversation roles to be larger, and further improves the accuracy of conversation role differentiation; adding 1 to the number of standard dialogs in the corpus of dialogs having corresponding words such that the denominator of the second set of parameters is not 0.
The present invention further provides another embodiment of a method for evaluating service quality of customer service, where the method includes steps S101 to S104 of the method for evaluating service quality of customer service of the above embodiment, and further defines that, the obtaining all process nodes of all service dialogues according to all process nodes of a standard service flow and all service dialogues includes:
extracting all keywords of each sentence of service dialogue from all the service dialogues respectively;
matching all process nodes of the standard service process with service dialogues corresponding to the process nodes in all service dialogues according to the process nodes and all keywords of all the service dialogues respectively;
and acquiring all the process nodes of all the service dialogs according to the service dialogs corresponding to the process nodes in all the service dialogs.
Namely, the efficiency of service quality evaluation is further improved by matching the flow nodes quickly through the keyword matching.
In an optional implementation manner, the matching, for all process nodes of the standard service process, service dialogs corresponding to the process nodes in all service dialogs according to the process nodes and all keywords of the all service dialogs respectively includes:
matching all keywords of the service dialogs with the sentence pattern templates corresponding to all the process nodes respectively through wildcards or regular expressions for all the service dialogs to obtain all the sentence pattern templates matched with the service dialogs;
and acquiring the service dialogues corresponding to the flow nodes in all the service dialogues according to all the flow nodes corresponding to the sentence pattern templates matched with the service dialogues.
For example, if the service dialog is: "do you need to modify the password", the keywords extracted from the service dialog are "modify", "password" and "do", then the sentence pattern template of each flow node is matched through a wildcard character "modify", "password", if the matched sentence pattern template is "do you want to modify the password", the flow node corresponding to the sentence pattern template is the flow node corresponding to the service dialog, if there is no matched sentence pattern template, there is no flow node corresponding to the service dialog.
Namely, the rapid matching is carried out through the wildcard or the regular expression, so that the complexity of calculation is reduced, and the efficiency of service quality evaluation is improved.
In an optional implementation manner, the matching, for all process nodes of the standard service process, service dialogs corresponding to the process nodes in all service dialogs according to the process nodes and all keywords of the all service dialogs respectively includes:
inputting all the keywords of the process nodes and all the service dialogs to a conceptual diagram tool configured according to a process corpus in advance for all the process nodes of the standard service process respectively so as to match the service dialogs corresponding to the process nodes in all the service dialogs; the process corpus comprises all process nodes of the standard service process and corpora corresponding to all the process nodes of the standard service process.
It should be noted that a conceptual diagram is a graphical representation of the concept and its relationship of a certain topic, and a conceptual diagram is a tool used to organize and characterize knowledge. It usually puts the related concepts of a certain subject in a circle or a box, and then connects the related concepts with propositions by connecting lines, and the connecting lines mark the meaning relationship between the two concepts.
Namely, the relation between the keywords and the process nodes is applied to the service dialogs through a concept graph tool, so that the service dialogs corresponding to the process nodes in all the service dialogs are matched more efficiently and more accurately, and the objectivity and the efficiency of service quality evaluation are further improved.
In an optional implementation manner, the calculating a flow matching degree of all flow nodes of all service dialogues with a standard service flow includes:
acquiring first preset weights corresponding to all process nodes of all service dialogues one to one; wherein, the sum of the first preset weights of all process nodes of the standard service process is equal to 1;
all the preset operation behaviors of the corresponding process nodes are respectively obtained for all the process nodes of all the service dialogues;
respectively acquiring customer service operation behaviors of corresponding process nodes for all process nodes of all service dialogues;
respectively acquiring corresponding operation scores of the process nodes according to all preset operation behaviors of all process nodes of all service dialogues and all acquired customer service operation behaviors;
respectively calculating products of the corresponding operation scores and the corresponding first preset weights for all the process nodes of all the service dialogues to obtain matching scores of the process nodes of all the service dialogues;
and adding the matching scores of the flow nodes of all the service pairs to obtain the flow matching degree.
The product of the first preset weight corresponding to each process node and the corresponding operation score is calculated, so that different first preset weights are set for different process nodes, the importance degree of different process nodes in the service process is fully considered, the data granularity is increased for calculating the process matching degree, and the objectivity of the evaluation result is further improved.
In an optional implementation manner, the obtaining, according to all preset operation behaviors of all process nodes of all service dialogues and all obtained customer service operation behaviors, operation scores of corresponding process nodes respectively includes:
respectively acquiring corresponding second preset weights for all preset operation behaviors of all process nodes of all service dialogues; wherein the sum of the second preset weights of all different preset operation behaviors corresponding to the standard service flow is equal to 1;
respectively acquiring preset operation behaviors which are the same as the customer service operation behaviors for all the customer service operation behaviors of all the process nodes of all the service dialogues;
and adding second preset weights of preset operation behaviors which are the same as the customer service operation behaviors to all the process nodes of all the service dialogues respectively to obtain operation scores of the process nodes respectively.
It should be noted that the customer service operation behavior is a conversation behavior and a business operation corresponding to customer service under the process node, for example:
customer service, do you want to modify account passwords?
Customer: is.
Customer service: please wait a bit.
The above-mentioned conversation behavior of the client and the service operation of the customer service for password modification thereafter are the customer service operation behavior.
For example, after all process nodes of the service dialogue of the customer service are obtained, the jth process node is set to have M preset operation behaviors, and the weight corresponding to the ith preset operation behavior is XiAnd satisfies the conditions of
Figure GDA0002612001570000161
If the ith preset operation behavior is the same as the customer service operation behavior, the P is orderediEqual to 0 or 1, the service score of the customer service of the jth process node is Sj：
Figure GDA0002612001570000162
Setting N all process nodes of all service dialogues, and setting the weight value corresponding to the jth process node as YjAnd satisfies the conditions of
Figure GDA0002612001570000163
The matching degree of the service conversation and the service process of the customer service is D:
Figure GDA0002612001570000164
the corresponding second preset weight is obtained for the preset operation behavior, so that different second preset weights are set for different preset operation behaviors, the importance degree of the different preset operation behaviors to the process node is fully considered, the data granularity is increased for the calculation of the process matching degree, and the objectivity of the evaluation result is improved.
Referring to fig. 2, it is a schematic structural diagram of an embodiment of a service quality evaluation system for customer service provided by the present invention, the system includes:
an extraction module 201, configured to extract all service dialogues of the customer service in the dialog to be analyzed;
an obtaining module 202, configured to obtain all process nodes of all service dialogues according to all process nodes of a standard service process and all service dialogues;
a calculating module 203, configured to calculate a process matching degree between all process nodes of all service dialogues and a standard service process;
and the evaluation module 204 is used for evaluating the service quality of the customer service according to the process matching degree.
It should be noted that all process nodes of the standard service process cover process nodes corresponding to all service dialogues of the customer service.
In an alternative embodiment, the system further comprises: the voice acquisition module is used for acquiring the conversation voice between the customer service and the client before all service dialogues of the customer service are extracted from the conversation to be analyzed; and the voice conversion module is used for converting the dialogue voice into a text to obtain the dialogue to be analyzed.
For example, the dialog to be analyzed between an after-sales service and a customer of a certain air conditioner is:
customer service: you are nice and happy to serve you
Customer: that air conditioner of my home can not be opened
Customer service: is the air conditioner can not be started
Customer: jone (a Chinese character)
Customer service: do you provide your order number at a glance
Customer: 189383983
Customer service: you will, according to the situation that you reflect, handle the appointment of after-sale detection for you
Customer service: asking you to be mr. in river, grand people and river
Customer: to pair
Customer service: good, address is confirmed with you: XX city XX county XX district XX road XX building
Customer: to pair
If the standard service flow includes: greeting, confirming after-sale problem, checking order number of goods, confirming details of goods and after-sale maintenance appointment.
The mapping relation between all service dialogues of the customer service and the process nodes is as follows:
you are nice and happy to serve you
Does not start the air conditioner (flow node: confirm after-sale problem)
Just provide your order number (flow node: check commodity order number)
You can do the after-sales service reservation (flow node: after-sales service reservation) according to the condition reflected by you
Asking you is Mr. of the river (flow node: maintenance appointment after sale)
Good, address is confirmed with you: XX city XX county XX district XX road XX number building (flow node: after-sale maintenance reservation)
It should be noted that the standard service flow, the mapping relationship between the service dialog and the flow node, and the standard dialog are only examples, and the mapping relationship between the service dialog and the flow node is subject to the actual application.
All service dialogs of the customer service are extracted from the dialog to be analyzed, so that the process node acquisition of all the dialogs of the customer service and the customer is avoided, the data processing amount is reduced, and the service quality evaluation efficiency of the customer service is improved; all the process nodes of all the service dialogues are obtained according to all the process nodes of the standard service process and all the service dialogues, so that the accuracy of process matching is improved, the accuracy of service quality evaluation of customer service is improved, the subjective influence caused by a manual mode is avoided, and meanwhile, the quality inspection efficiency is improved; the quality of the service process of the customer service is evaluated through the process matching degree, the intellectualization of the service quality evaluation of the customer service is realized, and compared with a method for evaluating only through the number of process nodes, the objectivity of quality inspection is improved.
In an alternative embodiment, the extraction module comprises:
the distinguishing word acquisition unit is used for acquiring all distinguishing words of the dialogue to be analyzed according to the dialogue to be analyzed and the distinguishing word library; the distinguishing word library comprises distinguishing words which are obtained in advance and the number of which is a first set number;
the label acquisition unit is used for acquiring labels of the dialog roles corresponding to the dialogues to be analyzed according to all the discriminative words and the dialog corpus of the dialogues to be analyzed; the dialogue corpus comprises a plurality of standard dialogs and labels of dialogue roles corresponding to the standard dialogs;
and the customer service dialog extraction unit is used for extracting all dialogs with the labels of the conversation roles in the conversation to be analyzed as customer service dialogs, and the extracted dialogs are used as all service dialogs of the customer service.
It should be noted that, in practical application, the dialog corpus is a dialog corpus in the field to which the dialog to be analyzed belongs; the standard dialogs of the dialog corpus are the dialogs of all dialogs contained in the dialog corpus; the labels of the dialog roles corresponding to the standard dialogs of each sentence stored in the dialog corpus should include labels of all the dialog roles of the dialogs of the dialog to be analyzed; for example, if the dialog role of the dialog to be analyzed includes customer service and customer, the dialog corpus stores the label of the dialog role corresponding to the standard dialog as the label of the customer service or the customer.
The method comprises the steps that the discriminative words of the dialogue to be analyzed are obtained through a discriminative word library, so that the complexity of judging the discriminative words in the dialogue to be analyzed is reduced, and the processing efficiency is improved; by combining the dialogue corpus, the dialogue roles are distinguished according to the dialogue to be analyzed, the accuracy of dialogue role feature extraction is improved, a more comprehensive dialogue corpus is provided to obtain a dialogue role distinguishing model of the dialogue corpus, so that the labels of the dialogue roles are identified more accurately, and the accuracy of dialogue role distinguishing is improved.
In an optional embodiment, the discriminative word obtaining unit includes:
the first preprocessing unit is used for preprocessing the dialogue to be analyzed to obtain all the words of the dialogue to be analyzed;
and the distinguishing word acquiring subunit is used for acquiring the distinguishing words of the dialogue to be analyzed according to all the words of the dialogue to be analyzed and all the distinguishing words of the distinguishing word library.
In an alternative embodiment, the pre-processing unit comprises:
and the preprocessing subunit is used for segmenting the dialogue to be analyzed and replacing strange words to obtain all the words of the dialogue to be analyzed.
In an optional embodiment, the discriminative word obtaining unit includes:
the word segmentation unit is used for segmenting the dialogue of the dialogue to be analyzed to obtain all words of the dialogue to be analyzed;
and the word matching unit is used for matching all the words of the dialogue to be analyzed with all the discriminative words of the discriminative word library so as to obtain the discriminative words of the dialogue to be analyzed.
The dialogue system comprises a dialogue analysis module, a dialogue analysis module and a dialogue analysis module, wherein the dialogue analysis module is used for analyzing dialogue components, the dialogue analysis module is used for analyzing dialogue components, the dialogue components are used for analyzing dialogue components, and the dialogue components are used for analyzing dialogue components.
In an alternative embodiment, the tag obtaining unit includes:
the first group of parameter acquisition units are used for respectively acquiring word frequencies of all discriminative words of the dialogue to be analyzed in the dialogue to be analyzed as a first group of parameters;
a first quantity obtaining unit, configured to obtain, for each discriminative term of the dialogs of the dialog to be analyzed, a quantity of standard dialogs having the corresponding discriminative term in a dialog corpus, respectively;
a second group of parameter obtaining unit, configured to obtain, for each distinct word of the dialogs of the dialog to be analyzed, a second group of parameters according to a total number of standard dialogs of the dialog corpus and a number of standard dialogs having corresponding distinct words in the dialog corpus, respectively;
the first feature vector generating unit is used for generating a feature vector of dialogue to be analyzed according to the first group of parameters and the second group of parameters;
and the label identification unit is used for inputting the feature vector of the dialogue to be analyzed into the dialogue role distinguishing model of the dialogue corpus so as to identify the label of the dialogue role corresponding to the dialogue to be analyzed.
In an alternative embodiment, the second set of parameter obtaining units includes:
a first valid denominator obtaining subunit, configured to add 1 to the standard number of the distinguished words in the dialogue corpus respectively for each distinguished word of the dialogue to be analyzed, where the standard number of the distinguished words is used as a first valid denominator;
and the second group of parameter acquisition subunit is used for respectively calculating the ratio of the total number of the standard dialogs of the dialog corpus to the first effective denominator for each discriminative word of the dialogs to be analyzed, and taking the ratio as a second group of parameters.
It should be noted that the word frequency refers to the number of times of occurrence of a word; the word frequency of all the discriminative words of the dialog to be analyzed in the dialog to be analyzed, i.e., the number of times all the discriminative words of the dialog to be analyzed appear in the dialog to be analyzed. Adding 1 to the number of standard dialogs in the corpus of dialogs having corresponding discriminative words such that the denominator of the second set of parameters is not 0.
In an alternative embodiment, the generating feature vectors of the dialog to be analyzed according to the first set of parameters and the second set of parameters includes:
carrying out logarithmic transformation on the second group of parameters to obtain a second group of parameters after logarithmic transformation;
and generating a feature vector of dialogue to be analyzed according to the first group of parameters and the logarithmically transformed second group of parameters.
The feature vectors are obtained through the word frequency and the ratio of the discriminative words in the dialogue to be analyzed, and then the labels of the dialogue roles are obtained through the dialogue role discrimination model established based on the dialogue corpus, so that the importance degree of the discriminative words in the dialogue to be analyzed and the importance degree of the discriminative words in the dialogue corpus are fully considered, and the accuracy of dialogue role discrimination is improved.
In an optional embodiment, the tag obtaining unit further includes:
a quantity difference calculation unit, configured to calculate a quantity difference between the first set quantity and the quantity of the distinctive words of the dialog to be analyzed before respectively obtaining word frequencies of the distinctive words of the dialog to be analyzed in the dialog of the dialog to be analyzed;
the first feature vector generation unit includes:
the first generating unit is used for generating filling word frequency with the number being the number difference; the filling word frequency is 0;
the second generating unit is used for generating a filling ratio with the number being the number difference; the filling ratio is 0;
a third generating unit, configured to generate the first characteristic parameters with the first set quantity according to the filling word frequency and the first group of parameter generation parameters;
the fourth generating unit is used for generating second characteristic parameters with the first set quantity according to the filling ratio and the second group of parameter generation parameters;
and the fifth generating unit is used for generating the feature vector of the dialogue to be analyzed according to the first feature parameter and the second feature parameter.
For example, if the dialog to be analyzed is "hello, ask you for a recipient", and the words with distinction are "you" and "do", the word frequency of "you" in the dialog to be analyzed is 2, and the word frequency of "do" in the dialog to be analyzed is 1; if the total number of standard dialogs of all dialog characters in the dialog corpus is 100, assuming that the number of standard dialogs with the word "you" in the dialog corpus is 60 and the number of standard dialogs with the word "do" is 58, the ratio of the total number of standard dialogs of the dialog corpus to the number of standard dialogs with the word "you" in the dialog corpus is
Figure GDA0002612001570000211
The ratio of the total number of standard dialogs in the corpus of dialogs to the number of standard dialogs having the word "do" in the corpus of dialogs is
Figure GDA0002612001570000212
Assuming that the first set number is 3, the first feature parameter is (2, 1, 0), the second feature parameter is (0.6, 0.58, 0), and the feature vector of the dialog to be analyzed for the dialog white is (2, 1, 0; 0.6, 0.58, 0).
The processing difficulty is reduced by generating filling word frequency and filling ratio and conveniently passing through the feature vectors with the determined parameter number; by generating the filling word frequency and the filling ratio of the null value, the substitution calculation of the parameters is reduced, the calculation speed is increased, and the processing efficiency is improved.
In an optional implementation, the extraction module further includes:
a third group parameter obtaining unit, configured to, in response to an instruction for training the dialogue role discrimination model, obtain, for each discriminative term in the discriminative term library, a term frequency of the corresponding discriminative term in each standard dialog in the dialogue corpus as a third group parameter;
a second number obtaining unit, configured to obtain, for each distinct word in the distinct word library, a number of standard dialogues having the corresponding distinct word in the dialog corpus;
a fourth group parameter obtaining unit, configured to obtain, for each distinct word in the distinct word library, a fourth group parameter according to a total number of standard dialogues in the dialog corpus and a number of standard dialogues having the corresponding distinct word in the dialog corpus, respectively;
a second feature vector generating unit, configured to generate feature vectors of the standard dialogues for each sentence according to the third set of parameters and the fourth set of parameters;
and the model establishing unit is used for training the conversation role distinguishing model according to the feature vectors of the standard dialogs of the sentences and the labels of the conversation roles corresponding to the standard dialogs of the sentences based on a naive Bayesian algorithm.
In an alternative embodiment, the fourth parameter obtaining unit includes:
a second valid denominator acquiring subunit, configured to add 1 to the number of standard dialogues with corresponding discriminative terms in the dialog corpus, respectively, for each discriminative term in the discriminative term library, to serve as a second valid denominator;
and the fourth group parameter acquiring subunit is used for calculating the ratio of the total number of the standard dialogues of the dialogue corpus to the second effective denominator of each differential word of the differential word library respectively as a fourth group parameter.
The conversation role distinguishing model is trained based on the naive Bayes algorithm, so that the contribution of each distinguishing word to the distinguishing of the conversation role can be accurately weighed, and the distinguishing accuracy of the conversation role is improved; adding 1 to the number of standard dialogs in the corpus of dialogs having corresponding discriminative terms so that the denominator of the second set of parameters is not 0.
In an optional implementation, the extraction module further includes:
the second preprocessing unit is used for responding to an instruction for acquiring the discriminative word bank and preprocessing all standard dialogues of the dialogue corpus to obtain all words of the dialogue corpus;
a fifth group of parameter obtaining unit, configured to obtain, for each word in the dialog corpus, a word frequency of the corresponding word in each standard dialog in the dialog corpus as a fifth group of parameters;
a third quantity obtaining unit, configured to obtain, for each word of the dialog corpus, a quantity of standard dialogues having the corresponding word in the dialog corpus;
a sixth group of parameter obtaining unit, configured to obtain, for each word of the dialog corpus, a sixth group of parameters according to a total number of standard dialogues of the dialog corpus and a number of standard dialogues having corresponding words in the dialog corpus, respectively;
a third feature vector generation unit, configured to generate feature vectors of the standard dialogues for each sentence according to the fifth set of parameters and the sixth set of parameters;
and the word selecting unit is used for selecting the words with the first set number from all the words in the dialogue corpus according to the feature vectors of the standard dialogues of each sentence and the labels of the dialogue roles corresponding to the standard dialogues of each sentence based on an information gain method so as to obtain the discriminative word library.
In an alternative embodiment, the sixth set of parameter obtaining units comprises:
a third effective denominator obtaining subunit, configured to add 1 to the standard dialogues of the words in the dialogue corpus respectively to serve as a third effective denominator;
and the sixth group of parameter acquisition subunits are used for respectively calculating the ratio of the total number of standard dialogues of the dialogue corpus to the third effective denominator for each word of the dialogue corpus, and the ratio is used as the sixth group of parameters.
The method selects the discriminative words through an information gain method, overcomes the subjective randomness and one-sidedness of manual screening, and accordingly enables the relevance between the screened discriminative words and the conversation roles to be larger, and further improves the accuracy of conversation role differentiation; adding 1 to the number of standard dialogs in the corpus of dialogs having corresponding words such that the denominator of the second set of parameters is not 0.
The invention also provides service quality evaluation of customer service, the system comprises the extraction module 201, the acquisition module 202, the calculation module 203 and the scoring module 204 of the service quality evaluation system of customer service of the embodiment, and the acquisition module further comprises:
a keyword extraction unit, configured to extract all keywords of each sentence of service dialogue for all the service dialogues respectively;
the keyword matching unit is used for matching all the process nodes of the standard service process with the service dialogues corresponding to the process nodes in all the service dialogues according to the process nodes and all the keywords of all the service dialogues respectively;
and the process node acquisition unit is used for acquiring all the process nodes of all the service dialogs according to the service dialogs corresponding to the process nodes in all the service dialogs.
Namely, the efficiency of service quality evaluation is further improved by matching the flow nodes quickly through the keyword matching.
In an alternative embodiment, the keyword extraction unit includes: and the third extraction subunit is used for respectively extracting all keywords of each sentence of service dialogue from all the service dialogues based on a natural semantic understanding algorithm.
Namely, all keywords of each sentence of service dialogue are respectively extracted from all service dialogues through a natural semantic understanding algorithm, and the accuracy of keyword extraction is greatly improved.
In an alternative embodiment, the keyword matching unit includes:
a sentence pattern template matching unit, configured to match, for all service dialogs, all keywords of the service dialogs with sentence pattern templates corresponding to each flow node through wildcards or regular expressions, respectively, so as to obtain all sentence pattern templates matched with the service dialogs;
and the fourth obtaining unit is used for obtaining the service dialogue corresponding to the flow node in all the service dialogs according to all the flow nodes corresponding to the sentence pattern template matched with the service dialogue.
For example, if the service dialog is: "do you need to modify the password", the keywords extracted from the service dialog are "modify", "password" and "do", then the sentence pattern template of each flow node is matched through a wildcard character "modify", "password", if the matched sentence pattern template is "do you want to modify the password", the flow node corresponding to the sentence pattern template is the flow node corresponding to the service dialog, if there is no matched sentence pattern template, there is no flow node corresponding to the service dialog.
Namely, the rapid matching is carried out through the wildcard or the regular expression, so that the complexity of calculation is reduced, and the efficiency of service quality evaluation is improved.
In an alternative embodiment, the keyword matching unit includes:
the service dialogue matching unit is used for inputting all the process nodes of the standard service process and all the keywords of all the service dialogues into a conceptual diagram tool configured according to a process corpus in advance respectively so as to match the service dialogues corresponding to the process nodes in all the service dialogues; the process corpus comprises all process nodes of the standard service process and corpora corresponding to all the process nodes of the standard service process.
It should be noted that a conceptual diagram is a graphical representation of the concept and its relationship of a certain topic, and a conceptual diagram is a tool used to organize and characterize knowledge. It usually puts the related concepts of a certain subject in a circle or a box, and then connects the related concepts with propositions by connecting lines, and the connecting lines mark the meaning relationship between the two concepts.
Namely, the relation between the keywords and the process nodes is applied to the service dialogs through a concept graph tool, so that the service dialogs corresponding to the process nodes in all the service dialogs are matched more efficiently and more accurately, and the objectivity and the efficiency of service quality evaluation are further improved.
In an alternative embodiment, the calculation module comprises:
a first obtaining unit, configured to obtain first preset weights in one-to-one correspondence with all process nodes of all service dialogues; wherein, the sum of the first preset weights of all process nodes of the standard service process is equal to 1;
a second obtaining unit, configured to obtain, for all the process nodes of all the service dialogues, all the preset operation behaviors of the corresponding process node respectively;
a third obtaining unit, configured to obtain, for all the process nodes in all the service dialogues, customer service operation behaviors of corresponding process nodes respectively;
a fourth obtaining unit, configured to obtain operation scores of corresponding process nodes according to all preset operation behaviors of all process nodes of all service dialogues and all obtained customer service operation behaviors;
the first calculation unit is used for calculating products of the corresponding operation scores and the corresponding first preset weights for all the process nodes of all the service dialogues respectively so as to obtain the matching scores of the process nodes of all the service dialogues;
and the first adding unit is used for adding the matching scores of the flow nodes of all the service dialogues to obtain the flow matching degree.
The product of the first preset weight corresponding to each process node and the corresponding operation score is calculated, so that different first preset weights are set for different process nodes, the importance degree of different process nodes in the service process is fully considered, the data granularity is increased for calculating the process matching degree, and the objectivity of the evaluation result is further improved.
In an optional implementation, the fourth obtaining unit includes:
the first obtaining subunit is configured to obtain corresponding second preset weights for all preset operation behaviors of all process nodes of all service dialogues respectively; wherein the sum of the second preset weights of all different preset operation behaviors corresponding to the standard service flow is equal to 1;
a second obtaining subunit, configured to obtain, for all customer service operation behaviors of all process nodes of all service dialogues, preset operation behaviors that are the same as the customer service operation behaviors respectively;
and the first adding subunit is configured to add, to all the process nodes of all the service dialogues, second preset weights of preset operation behaviors that are the same as the operation behaviors of the customer service, so as to obtain operation scores of the process nodes respectively.
It should be noted that the customer service operation behavior is a conversation behavior and a business operation corresponding to customer service under the process node, for example:
customer service, do you want to modify account passwords?
Customer: is.
Customer service: please wait a bit.
The above-mentioned conversation behavior of the client and the service operation of the customer service for password modification thereafter are the customer service operation behavior.
For example, after all process nodes of the service dialogue of the customer service are obtained, the jth process node is set to have M preset operation behaviors, and the weight corresponding to the ith preset operation behavior is XiAnd satisfies the conditions of
Figure GDA0002612001570000261
If the ith preset operation behavior is the same as the customer service operation behavior, the P is orderediEqual to 0 or 1, the service score of the customer service of the jth process node is Sj：
Figure GDA0002612001570000262
Setting N all process nodes of all service dialogues, and setting the weight value corresponding to the jth process node as YjAnd satisfies the conditions of
Figure GDA0002612001570000271
The matching degree of the service conversation and the service process of the customer service is D:
Figure GDA0002612001570000272
the corresponding second preset weight is obtained for the preset operation behavior, so that different second preset weights are set for different preset operation behaviors, the importance degree of the different preset operation behaviors to the process node is fully considered, the data granularity is increased for the calculation of the process matching degree, and the objectivity of the evaluation result is improved.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by a computer program, which can be stored in a computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. The storage medium may be a magnetic disk, an optical disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), or the like.
While the foregoing is directed to the preferred embodiment of the present invention, it will be understood by those skilled in the art that various changes and modifications may be made without departing from the spirit and scope of the invention.

Claims (10)

1. A method for evaluating the service quality of customer service is characterized by comprising the following steps:
extracting all service dialogues of the customer service in the dialog to be analyzed; acquiring all discriminative words of the dialogue to be analyzed according to the dialogue to be analyzed and a discriminative word library, wherein the discriminative word library comprises the discriminative words which are acquired in advance and the number of the discriminative words is a first set number; acquiring labels of dialog roles corresponding to dialogues of the dialogues to be analyzed according to all discriminative words of the dialogues to be analyzed and a dialog role discrimination model established in advance according to a dialog corpus, wherein the dialog corpus comprises multiple standard dialogues and labels of the dialog roles corresponding to the standard dialogues; extracting all dialogues with the labels of the conversation roles in the conversation to be analyzed as customer service dialogues, and taking the dialogues as all service dialogues of the customer service;
the word frequency of each discriminative word of the dialogue to be analyzed in the dialogue to be analyzed is respectively obtained and used as a first group of parameters; respectively acquiring the number of standard dialogs with corresponding differential words in a dialog corpus for each differential word of the dialogs to be analyzed; for each discriminative word of the dialogue to be analyzed, respectively obtaining a second group of parameters according to the total number of standard dialogues of the dialogue corpus and the number of standard dialogues with corresponding discriminative words in the dialogue corpus; generating a feature vector of dialogue to be analyzed according to the first group of parameters and the second group of parameters; inputting the feature vector of the dialogue to be analyzed into a dialogue role discrimination model of the dialogue corpus so as to identify a label of a dialogue role corresponding to the dialogue of the dialogue to be analyzed;
acquiring all process nodes of all service dialogues according to all process nodes of a standard service process and all service dialogues;
calculating the flow matching degree of all flow nodes of all service dialogues and a standard service flow;
and evaluating the service quality of the customer service according to the flow matching degree.
2. The method of claim 1, wherein the obtaining all process nodes of all service dialogs according to all process nodes of a standard service process and all service dialogs comprises:
extracting all keywords of each sentence of service dialogue from all the service dialogues respectively;
matching all process nodes of the standard service process with service dialogues corresponding to the process nodes in all service dialogues according to the process nodes and all keywords of all the service dialogues respectively;
and acquiring all the process nodes of all the service dialogs according to the service dialogs corresponding to the process nodes in all the service dialogs.
3. The method of claim 2, wherein the step of matching, for all process nodes of the standard service process, the service dialogue corresponding to the process node in all service dialogues according to the process node and all keywords of all service dialogues, comprises:
inputting all the keywords of the process nodes and all the service dialogs to a conceptual diagram tool configured according to a process corpus in advance for all the process nodes of the standard service process respectively so as to match the service dialogs corresponding to the process nodes in all the service dialogs; the process corpus comprises all process nodes of the standard service process and corpora corresponding to all the process nodes of the standard service process.
4. The method of claim 1, wherein the calculating the process matching degree of all process nodes of all service dialogues with the standard service process comprises:
acquiring first preset weights corresponding to all process nodes of all service dialogues one to one; wherein, the sum of the first preset weights of all process nodes of the standard service process is equal to 1;
all the preset operation behaviors of the corresponding process nodes are respectively obtained for all the process nodes of all the service dialogues;
respectively acquiring customer service operation behaviors of corresponding process nodes for all process nodes of all service dialogues;
respectively acquiring corresponding operation scores of the process nodes according to all preset operation behaviors of all process nodes of all service dialogues and all acquired customer service operation behaviors;
respectively calculating products of the corresponding operation scores and the corresponding first preset weights for all the process nodes of all the service dialogues to obtain matching scores of the process nodes of all the service dialogues;
and adding the matching scores of the flow nodes of all the service pairs to obtain the flow matching degree.
5. The method for evaluating the service quality of customer service according to claim 4, wherein the step of obtaining the operation scores of the corresponding process nodes according to all the preset operation behaviors of all the process nodes of all the service dialogues and all the obtained customer service operation behaviors comprises:
respectively acquiring corresponding second preset weights for all preset operation behaviors of all process nodes of all service dialogues; wherein the sum of the second preset weights of all different preset operation behaviors corresponding to the standard service flow is equal to 1;
respectively acquiring preset operation behaviors which are the same as the customer service operation behaviors for all the customer service operation behaviors of all the process nodes of all the service dialogues;
and adding second preset weights of preset operation behaviors which are the same as the customer service operation behaviors to all the process nodes of all the service dialogues respectively to obtain operation scores of the process nodes respectively.
6. A system for evaluating the quality of service of a customer service, the system comprising:
the extraction module is used for extracting all service dialogues of the customer service in the dialog to be analyzed; acquiring all discriminative words of the dialogue to be analyzed according to the dialogue to be analyzed and a discriminative word library, wherein the discriminative word library comprises the discriminative words which are acquired in advance and the number of the discriminative words is a first set number; acquiring labels of dialog roles corresponding to dialogues of the dialogues to be analyzed according to all discriminative words of the dialogues to be analyzed and a dialog role discrimination model established in advance according to a dialog corpus, wherein the dialog corpus comprises multiple standard dialogues and labels of the dialog roles corresponding to the standard dialogues; extracting all dialogues with the labels of the conversation roles in the conversation to be analyzed as customer service dialogues, and taking the dialogues as all service dialogues of the customer service;
the word frequency of each discriminative word of the dialogue to be analyzed in the dialogue to be analyzed is respectively obtained and used as a first group of parameters; respectively acquiring the number of standard dialogs with corresponding differential words in a dialog corpus for each differential word of the dialogs to be analyzed; for each discriminative word of the dialogue to be analyzed, respectively obtaining a second group of parameters according to the total number of standard dialogues of the dialogue corpus and the number of standard dialogues with corresponding discriminative words in the dialogue corpus; generating a feature vector of dialogue to be analyzed according to the first group of parameters and the second group of parameters; inputting the feature vector of the dialogue to be analyzed into a dialogue role discrimination model of the dialogue corpus so as to identify a label of a dialogue role corresponding to the dialogue of the dialogue to be analyzed;
the acquisition module is used for acquiring all the process nodes of all the service dialogues according to all the process nodes of a standard service process and all the service dialogues;
the calculation module is used for calculating the flow matching degree of all the flow nodes of all the service dialogues and the standard service flow;
and the evaluation module is used for evaluating the service quality of the customer service according to the flow matching degree.
7. The system of claim 6, wherein the acquisition module comprises:
a keyword extraction unit, configured to extract all keywords of each sentence of service dialogue for all the service dialogues respectively;
the keyword matching unit is used for matching all the process nodes of the standard service process with the service dialogues corresponding to the process nodes in all the service dialogues according to the process nodes and all the keywords of all the service dialogues respectively;
and the process node acquisition unit is used for acquiring all the process nodes of all the service dialogs according to the service dialogs corresponding to the process nodes in all the service dialogs.
8. The system of claim 7, wherein the keyword matching unit comprises:
the service dialogue matching unit is used for inputting all the process nodes of the standard service process and all the keywords of all the service dialogues into a conceptual diagram tool configured according to a process corpus in advance respectively so as to match the service dialogues corresponding to the process nodes in all the service dialogues; the process corpus comprises all process nodes of the standard service process and corpora corresponding to all the process nodes of the standard service process.
9. The system of claim 6, wherein the computing module comprises:
a first obtaining unit, configured to obtain first preset weights in one-to-one correspondence with all process nodes of all service dialogues; wherein, the sum of the first preset weights of all process nodes of the standard service process is equal to 1;
a second obtaining unit, configured to obtain, for all the process nodes of all the service dialogues, all the preset operation behaviors of the corresponding process node respectively;
a third obtaining unit, configured to obtain, for all the process nodes in all the service dialogues, customer service operation behaviors of corresponding process nodes respectively;
a fourth obtaining unit, configured to obtain operation scores of corresponding process nodes according to all preset operation behaviors of all process nodes of all service dialogues and all obtained customer service operation behaviors;
the first calculation unit is used for calculating products of the corresponding operation scores and the corresponding first preset weights for all the process nodes of all the service dialogues respectively so as to obtain the matching scores of the process nodes of all the service dialogues;
and the first adding unit is used for adding the matching scores of the flow nodes of all the service dialogues to obtain the flow matching degree.
10. The system for evaluating the quality of service of a customer service according to claim 9, wherein the fourth obtaining unit comprises:
the first obtaining subunit is configured to obtain corresponding second preset weights for all preset operation behaviors of all process nodes of all service dialogues respectively; wherein the sum of the second preset weights of all different preset operation behaviors corresponding to the standard service flow is equal to 1;
a second obtaining subunit, configured to obtain, for all customer service operation behaviors of all process nodes of all service dialogues, preset operation behaviors that are the same as the customer service operation behaviors respectively;
and the first adding subunit is configured to add, to all the process nodes of all the service dialogues, second preset weights of preset operation behaviors that are the same as the customer service operation behaviors, so as to obtain operation scores of the process nodes respectively.
CN201711077542.9A 2017-11-03 2017-11-03 Service quality evaluation method and system for customer service Active CN107886233B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201711077542.9A CN107886233B (en) 2017-11-03 2017-11-03 Service quality evaluation method and system for customer service

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201711077542.9A CN107886233B (en) 2017-11-03 2017-11-03 Service quality evaluation method and system for customer service

Publications (2)

Publication Number Publication Date
CN107886233A CN107886233A (en) 2018-04-06
CN107886233B true CN107886233B (en) 2020-11-13

Family

ID=61778778

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201711077542.9A Active CN107886233B (en) 2017-11-03 2017-11-03 Service quality evaluation method and system for customer service

Country Status (1)

Country Link
CN (1) CN107886233B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110633912A (en) * 2019-09-20 2019-12-31 苏州思必驰信息科技有限公司 Method and system for monitoring service quality of service personnel
CN111883115B (en) * 2020-06-17 2022-01-28 马上消费金融股份有限公司 Voice flow quality inspection method and device

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101005531A (en) * 2007-01-30 2007-07-25 华为技术有限公司 Quality detecting method, quality detecting device and quality detecting system
CA2781042A1 (en) * 2009-11-16 2011-05-19 Joanne Curry Method and apparatus for modeling a client service journey
CN102945628A (en) * 2011-12-23 2013-02-27 江西省电力公司信息通信中心 Accident consequence information acquiring method, device and training system
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

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101005531A (en) * 2007-01-30 2007-07-25 华为技术有限公司 Quality detecting method, quality detecting device and quality detecting system
CA2781042A1 (en) * 2009-11-16 2011-05-19 Joanne Curry Method and apparatus for modeling a client service journey
CN102945628A (en) * 2011-12-23 2013-02-27 江西省电力公司信息通信中心 Accident consequence information acquiring method, device and training system
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

Also Published As

Publication number Publication date
CN107886233A (en) 2018-04-06

Similar Documents

Publication Publication Date Title
CN107886231B (en) Service quality evaluation method and system for customer service
CN110096570B (en) Intention identification method and device applied to intelligent customer service robot
CN107833059B (en) Service quality evaluation method and system for customer service
TWI723528B (en) Computer-executed event risk assessment method and device, computer-readable storage medium and computing equipment
CN107766560B (en) Method and system for evaluating customer service flow
CN107885723B (en) Conversation role distinguishing method and system
CN111709630A (en) Voice quality inspection method, device, equipment and storage medium
WO2022095353A1 (en) Speech recognition result evaluation method, apparatus and device, and storage medium
CN112163424A (en) Data labeling method, device, equipment and medium
CN112732871A (en) Multi-label classification method for acquiring client intention label by robot
CN109582788A (en) Comment spam training, recognition methods, device, equipment and readable storage medium storing program for executing
CN113626573B (en) Sales session objection and response extraction method and system
CN112671985A (en) Agent quality inspection method, device, equipment and storage medium based on deep learning
CN111462752A (en) Client intention identification method based on attention mechanism, feature embedding and BI-L STM
CN112069312A (en) Text classification method based on entity recognition and electronic device
CN111177402A (en) Evaluation method and device based on word segmentation processing, computer equipment and storage medium
CN107818173B (en) Vector space model-based Chinese false comment filtering method
CN107886233B (en) Service quality evaluation method and system for customer service
CN110347833B (en) Classification method for multi-round conversations
CN107480126B (en) Intelligent identification method for engineering material category
CN116975910A (en) Method, device, equipment and medium for determining security level of data table
CN107886232B (en) Service quality evaluation method and system for customer service
CN116644148A (en) Keyword recognition method and device, electronic equipment and storage medium
CN116070620A (en) Information processing method and system based on big data
CN115994531A (en) Multi-dimensional text comprehensive identification method

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