WO2021174840A1 - 语音流程设置方法、装置、计算机设备和存储介质 - Google Patents

语音流程设置方法、装置、计算机设备和存储介质 Download PDF

Info

Publication number
WO2021174840A1
WO2021174840A1 PCT/CN2020/119378 CN2020119378W WO2021174840A1 WO 2021174840 A1 WO2021174840 A1 WO 2021174840A1 CN 2020119378 W CN2020119378 W CN 2020119378W WO 2021174840 A1 WO2021174840 A1 WO 2021174840A1
Authority
WO
WIPO (PCT)
Prior art keywords
question
classification
voice
user
consultation
Prior art date
Application number
PCT/CN2020/119378
Other languages
English (en)
French (fr)
Inventor
邹昆伦
Original Assignee
平安科技（深圳）有限公司
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 平安科技（深圳）有限公司 filed Critical 平安科技（深圳）有限公司
Publication of WO2021174840A1 publication Critical patent/WO2021174840A1/zh

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/60—Information retrieval; Database structures therefor; File system structures therefor of audio data
    • G06F16/65—Clustering; Classification
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • 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
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0281—Customer communication at a business location, e.g. providing product or service information, consulting

Definitions

  • This application relates to the field of computer network communication technology, and in particular to a method, device, computer equipment, and storage medium for setting a voice process.
  • the customer service system is abbreviated as the customer service system, which creates questions and answers to the questions according to the needs of users.
  • the user calls into the system through voice, and the system provides user options.
  • the user enters the questions according to the options to get the answers to the questions, because the customer service system is through a simple landline phone or mobile phone Operation, the options are all realized through the buttons of the phone.
  • the user selects and operates according to the classification level by level until the target is obtained. Questions and answers to questions.
  • the customer service system is also equipped with manual customer service, which can quickly understand the user’s questions and provide corresponding answers.
  • the manual customer service is limited. When multiple users consult, the overall manual customer service is busy and it takes a long time to access Manual customer service.
  • the inventor realizes that the existing customer service system has relatively rigid problem classification, and users often need to go through multi-level operations to get the target problem or cannot find the target problem at all.
  • the manual customer service is busy again, resulting in The efficiency of user problem solving is low.
  • a voice process setting method includes:
  • the user's consultation questions within the preset time period are input into the convergent customer service model as the test set data to obtain the final question classification result; wherein the convergent customer service model passes the user's consultation questions within the preset time period and the consultation
  • the standard classification of the problem is obtained through repeated training;
  • the number of consultations, average time-consuming and solution status of all consultation problems in each problem classification to set an initial level for the problem classification
  • a voice call-in method for a customer service system includes:
  • the question classification option is the last level question classification option of the voice flow
  • the answer to the target question is sent to the user terminal according to the question selection instruction.
  • a voice flow setting device includes:
  • a customer service model to obtain a final problem classification result wherein the convergent customer service model is obtained through repeated training of the user's consultation questions within the preset time period and the standard classification of the consultation questions;
  • the initial level setting module is used to set the initial level of the problem classification according to the number of consultations, the average time consumption and the solution status of all the consultation problems in each problem classification in the final problem classification result;
  • the voice process setting module is used to set the voice process according to the initial level of the question classification.
  • a voice call-in device for a customer service system comprising:
  • the voice flow setting module is configured to obtain the voice flow set by the voice flow setting method described in the foregoing embodiment, and the voice flow includes options for question classification;
  • a selection instruction receiving module configured to receive a question selection instruction sent by the user terminal according to the question classification options
  • the sending module is configured to send the next question classification option to the user terminal according to the question selection instruction when the question classification option is not the last level question classification option of the voice process; When the question classification option is the last level question classification option of the voice process, the answer to the target question is sent to the user terminal according to the question selection instruction.
  • a computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the steps of a voice flow setting method when the computer program is executed:
  • the user's consultation questions within the preset time period are input into the convergent customer service model as the test set data to obtain the final problem classification result; wherein the convergent customer service model passes the user's consultation questions within the preset time period and the consultation The standard classification of the problem is obtained through repeated training;
  • the number of consultations, average time-consuming and solution status of all consultation problems in each problem classification to set an initial level for the problem classification
  • a computer device includes a memory and a processor, and the memory stores a computer program, wherein the steps of implementing a voice call-in method for a customer service system when the processor executes the computer program:
  • the question classification option is the last level question classification option of the voice flow
  • the answer to the target question is sent to the user terminal according to the question selection instruction.
  • the user's consultation questions within the preset time period are input into the convergent customer service model as the test set data to obtain the final problem classification result; wherein the convergent customer service model passes the user's consultation questions within the preset time period and the consultation The standard classification of the problem is obtained through repeated training;
  • the number of consultations, average time-consuming and solution status of all consultation problems in each problem classification to set an initial level for the problem classification
  • a computer-readable storage medium having a computer program stored thereon, wherein the steps of implementing a method for voice call in a customer service system when the computer program is executed by a processor:
  • the question classification option is the last level question classification option of the voice flow
  • the answer to the target question is sent to the user terminal according to the question selection instruction.
  • the above-mentioned voice flow setting method, device, computer equipment and storage medium classify the user's consultation questions through the convergent customer service model, and combine the consultation times, average time consumption and resolution status of all consultation questions in the question classification to solve the problem
  • the initial level of classification is set, and then the voice flow about the classification of the question is set, which can dynamically adjust the classification of the question according to the needs of the user, so that the user can quickly enter the question to be consulted and obtain the answer to the target question.
  • Fig. 1 is an application scenario diagram of a voice process setting method in an embodiment
  • Fig. 2 is a schematic flowchart of a method for setting a voice process in an embodiment
  • FIG. 3 is a schematic flowchart of a voice call-in method for a customer service system in an embodiment
  • Figure 4 is a structural block diagram of a voice flow setting device in an embodiment
  • Figure 5 is a structural block diagram of a voice call-in device for a customer service system in an embodiment
  • Fig. 6 is an internal structure diagram of a computer device in an embodiment.
  • the voice process setting method provided in this application can be applied to the application environment as shown in FIG. 1.
  • the user terminal 102 communicates with the server 104 through a wireless communication network through a network.
  • the user provides consultation questions to the server 104 through the user terminal 102, and the server 104 inputs the user's consultation questions within a preset time period as test set data into the convergent customer service model to obtain the final question classification result; wherein, the convergent customer service model Obtained through repeated training of the user's consultation questions within the preset time period and the standard classification of the consultation questions; according to the final question classification result, the consultation times, average time consumption and resolution of all consultation questions in each question classification State to set the initial level of the question classification; according to the initial level of the question classification, set the voice flow.
  • the user terminal 102 can be, but is not limited to, various personal computers, notebook computers, smart phones, tablet computers, landlines, mobile phones, and portable wearable devices.
  • the server 104 can be an independent server or a server cluster composed of multiple servers. accomplish.
  • a method for setting a voice process is provided. Taking the method applied to the server in FIG. 1 as an example for description, the method includes the following steps:
  • step S110 the user's consultation questions within the preset time period are input into the convergent customer service model as test set data to obtain the final question classification result; wherein the convergent customer service model passes the user's consultation questions and the user's consultation questions within the preset time period.
  • the standard classification of the consultation questions is obtained through repeated training.
  • the preset interval can be a specified month, a specified half year, or a specified year.
  • the preset time period cannot be too short.
  • the consultation questions of users obtained in too short time are not representative, and the time cannot be too long. Long, too long time has caused some canceled user consultation problems to still exist in the customer service system, resulting in excessive use of system resources.
  • the voice call record of each question asked by the user will be saved in the customer service system for subsequent analysis and query.
  • the standard classification of the consultation questions is manually marked and classified within the preset time period.
  • the customer service model may be an artificial neural network model, and the artificial neural network model scores and categorizes the user's consultation questions according to the standard classification of the consultation questions, and questions with the same score level are regarded as one question classification.
  • record the consultation questions of each user's incoming call in the last six months use each call's consultation questions of all users in the last six months as sample data, and manually mark the standard classification of each consultation question, according to the said
  • the sample data and the standard classification train the artificial neural network model to obtain a convergent customer service model.
  • the user's incoming consultation questions include S1, S2, L1, and L2.
  • the final question classification results include all question categories, all consultation questions in each question category, the number of all consultation questions in each question category, the length of each consultation question in each question category, and each question category in each question category. Consult the resolution status of the problem.
  • the final problem classification results include: problem classification S and L, consultation questions S1 and S2 in problem classification S, and the number of consultation questions in problem classification S 3 (consulting question S1 was consulted twice, and consultation question S2 was consulted Questions L1 and L2 in the question category L, the number of consultation questions in the question category L 5 (the consultation question L1 was consulted twice, and the consultation question L2 was consulted three times), each time the question S1 was consulted Duration and resolution status, the duration and resolution status of each consultation question S2, the duration and resolution status of each consultation question L1, and the duration and resolution status of each consultation question L2.
  • the customer service system has stored the user's voice call records within a preset time period to obtain the user's consultation questions, the duration of the consultation questions, and the resolution status.
  • step S120 an initial level is set for the problem classification according to the number of consultations, the average time consumption, and the solution status of all the consultation problems in each problem classification in the final problem classification result.
  • each valid call of the user can be set as one consultation, and the number of consultations in step S120 is the total number of consultations for all consultation questions, and each consultation question may be consulted by multiple users.
  • the average time is equal to the total time of all consultation questions divided by the number of consultations.
  • the solution status includes the problem is completely solved, the problem is partially solved, and the problem is not solved.
  • the level refers to the option level when the user calls into the customer service system. For example, when the user first calls in, the user is provided with the first-level option level. After selecting one of the options according to the first-level option level, it enters the second-level option level. And so on, until it corresponds to the answer to the final target question.
  • the level of this question classification can be set to be higher.
  • the higher the level the higher the classification of the question, so that the user can quickly enter the question to be consulted and obtain the answer; or the question classification
  • the average time-consuming is low and the solution status is when the problem is completely solved, and the level of the problem classification is set high.
  • Step S130 Set a voice flow according to the initial level of the question classification.
  • the user calls into the system through a landline or mobile phone.
  • the customer service model Before calling into the system, the customer service model has established a voice process based on the existing user call data.
  • the voice process includes options for first-level problem classification, and second-level problem classification... ...N-level question classification options and the answer to the target question, the user selects in turn according to the question classification options provided by the voice process until the answer to the target question is obtained.
  • the user’s consultation questions are classified through the convergent customer service model, and the initial level of the question classification is set by combining the consultation times, average time consumption and resolution status of all consultation questions in the question classification, and then setting Regarding the voice flow of the question classification, the dynamic adjustment of the question classification can be combined with the needs of the user, so that the user can quickly enter the question to be consulted and obtain the answer to the target question.
  • the voice process setting method further includes: inputting the user's consultation questions in the preset time period as the training set data and the standard classification of the consultation questions into the initial customer model to obtain the initial question classification Result; Calculate the error between the initial question classification result and the standard classification of the consultation question; use the error as the loss value of the network model and backpropagate it back to the convolutional layer of the initial customer model to optimize the customer model With the parameters of each convolutional layer, a convergent customer service model is obtained.
  • the error between the initial question classification result and the standard classification of the consultation question is the probability that the initial question classification result is incorrect with respect to the standard classification of the consultation question.
  • the initial customer model is repeatedly trained through the user's consultation questions within a preset time period, and a customer model with more and more accurate classification can be obtained.
  • the voice process setting method further includes: obtaining updated user consultation questions, consultation time duration and solution status; according to the updated user consultation questions, consultation time duration and resolution status, obtaining The updated question classification of the user's consultation question, the level of the question classification is adjusted in real time; wherein, when the user calls in, the voice flow is provided according to the adjusted question classification level.
  • the background records the user's call-in record in real time.
  • the call-in record includes recording the user's consultation questions, the duration of the consultation questions, and the resolution status.
  • There are many consultation times for question classification and the level of this question classification can be set to be higher. The higher the level, the higher the classification of the question, so that users can quickly enter the question to be consulted and obtain the answer; or the average time consumption of the question classification is low and the resolution status is low
  • the problem classification level is set to be high.
  • users call in more and more times according to the voice process their consulting services are constantly updated, and the problem classification needs to be adjusted in real time. By adjusting the level of the problem classification in real time, the demand for service update can be met.
  • the step of adjusting the level of the problem classification in real time according to the updated user's consultation question, the duration of the consultation question, and the resolution status includes: inputting the updated user's consultation question into the convergent customer service model , Get the problem classification of the updated user’s consultation questions; count the number of consultations, average time-consuming and resolution status of all consultation questions in the problem classification of the updated user’s consultation questions; according to the consultation times of all consultation questions, The average time-consuming and solving status adjust the level of the problem classification in real time.
  • the level of each question classification can be adjusted in real time by re-calculating the number of consultations, average time consumption and resolution status of all consultation questions according to the question classification of the user's consultation questions, and real-time changes in the business can be achieved.
  • the level of question classification is adjusted in real time to improve the voice process.
  • the step of obtaining the updated user's consulting question, the duration of the consulting issue, and the resolution status includes: optimizing and Improve the parameters of the convergent customer service model.
  • the system After the system counts the updated user's consultation questions, the duration of the consultation questions, and the resolution status, and finds that the updated user's consultation questions are too long, it obtains the user's current incoming voice call record, and analyzes the reason for the update.
  • the inaccurate classification of the user’s consultation questions leads to too long consultation time, and the parameters of the convergent customer service model need to be improved, so that the updated user’s consultation questions are classified into the problem categories with more user choices; if an updated user is found If the solution status of the consultation problem is that the problem is partially solved and the problem is not solved, the user's incoming voice call record is obtained.
  • the analysis is because there is no problem classification for the updated user's consultation problem, and a convergent customer service model is required Improve the parameters or set the standard classification of new consultation questions, so that the updated user's consultation questions are classified into the new standard classification of consultation questions.
  • a voice call-in method for a customer service system is provided. Taking the method applied to the server in FIG. 1 as an example for description, the method includes the following steps:
  • Step S210 Acquire the voice flow set by the voice flow setting method described in the foregoing embodiment, where the voice flow includes options for question classification.
  • the voice process includes options for the first-level question classification, options for the second-level question classification...the options for the N-level question classification and the answer to the target question.
  • Step S220 Receive a question selection instruction sent by the user terminal according to the question classification options.
  • the user terminal includes communication devices such as mobile phones and landlines, and the question selection instruction is issued through the user terminal.
  • the voice process provides options for the first-level question classification, and the options for the first-level question classification are specific
  • the user presses 1 at the user terminal the user terminal will issue a question selection instruction to select the Mandarin service.
  • Step S230 When the question classification option is not the last level question classification option of the voice process, the next question classification option is sent to the user terminal according to the question selection instruction. Wherein, the next question classification option is obtained according to the question selection instruction. For example, the user sends a question selection instruction through the user terminal according to the first-level question classification option, and the system enters the corresponding second-level question classification option according to the question selection instruction. The option of the second-level question classification is the next question classification option of the first-level question classification option.
  • the voice process includes options for first-level question classification, options for second-level question classification... N-level question classification options and answers to target questions, and N-level question classification options are the last-level question classification of the voice process Options.
  • Step S240 When the question classification option is the last-level question classification option of the voice flow, the answer to the target question is sent to the user terminal according to the question selection instruction.
  • the answer to the target question is determined by the user's selection based on the question classification options of the last level of the voice process.
  • the user answers the multiple question categories sent by the system through the mobile phone or landline, that is, the button option corresponding to the question category.
  • the user selects the question category by selecting the corresponding button, and then enters the next level of multiple question categories , And then press the button to select the corresponding question category, until the target question is finally selected, and answer the answer to the target question given by the system.
  • the button option corresponding to the question category.
  • one of the question classification options in the voice process includes: “Press 1 for calling charges”, “Press 2 for traffic inquiries,” “Press 3 for broadband access.”
  • Send call charge information the question classification option is the last level question classification option, and the call charge information is the answer to the target question; if the user presses button 3, according to the user's choice, enter the next question classification option,
  • the options for the next question classification include: “Press 1 for broadband type understanding” and "Press 2 for broadband account password query”. At this time, if the user presses button 2 to send the broadband account password to the user, the option for the next question classification is the last one.
  • There are options for classifying questions and broadband account password is the answer to the target question.
  • the voice call-in method of the customer service system described above can shorten the time for the user to consult questions and improve the efficiency of the user's consultation through the voice flow set by the voice flow setting method.
  • the method includes: receiving an incoming call request from a user terminal through a user number; querying historical incoming data corresponding to the user number according to the user number; judging according to the historical incoming data Whether the user number is an abnormal user number; if the user number is an abnormal user number, add a blacklist identifier to the user number; if the user number is not an abnormal user number, send all information to the user terminal A voice message describing the options for question classification.
  • the user number can be a mobile phone number or a landline number, and the user calls into the customer service system through a mobile phone or a landline.
  • the historical call-in data is the call record and voice flow of the user's historical call-in to the customer service system.
  • the step of judging whether the user number is an abnormal user number according to the historical incoming call data is specifically: calculating the number of consultations on the user's abnormal problem (harassing call) and normal business consultation according to the historical incoming call data The ratio of the number of times; when the ratio of the number of consultations for the user’s abnormal problem (harassing call) to the number of normal business consultations exceeds the preset value, the user number is judged to be an abnormal user number; when the user’s abnormal problem (harassing call) When the ratio of the number of consultations to the number of normal service consultations does not exceed the preset value, it is determined that the user number is not an abnormal user number.
  • the incoming calls of harassing calls can be effectively reduced, and the probability of harassing calls entering manual customer service can be reduced, unnecessary waste of manual customer service resources can be avoided, and the probability of normal incoming calls to manual customer service can be increased. .
  • the method includes: receiving an instruction for selecting manual customer service sent by the user terminal; obtaining the average waiting time for accessing manual customer service at the current time according to the instruction for selecting manual customer service; The user terminal sends the voice information of the average waiting time for accessing the manual customer service or sends the voice information suggesting that the intelligent customer service is selected.
  • the voice information suggesting the selection of the intelligent customer service is sent to the user terminal.
  • the average waiting time for accessing manual customer service is fed back to the user, so that the user can continue to wait for access to the manual customer service or choose other ways to ask questions according to personal needs, which effectively saves the user's time, and the user
  • the average waiting time of the customer service is long, it is recommended that the user choose the intelligent customer service, which effectively improves the efficiency of the user's inquiry.
  • the method includes: receiving an incoming call request from a user terminal through a user number; according to the user number, inquiring whether there is uncompleted voice call information corresponding to the user number; Unfinished voice call information, obtain the voice process and disconnection node corresponding to the unfinished voice call information; send the disconnection node to the user terminal according to the voice process and disconnection node corresponding to the unfinished voice call information The voice information of the question classification option corresponding to the line node.
  • the unfinished voice call information means that the user's incoming call to the customer service system is abnormally disconnected, for example, communication interruption occurs due to poor signal of the user terminal, communication interruption occurs when the user terminal malfunctions, or the user terminal power is exhausted to send communication Interrupted.
  • the voice process and disconnection node corresponding to the unfinished voice call information refers to the voice process and the problem classification level entered by the voice process before the disconnection. For example, the user is consulting the call fee inquiry process before the disconnection, and the disconnection When you have entered the second-level problem classification option, the disconnected node is the second-level problem classification option.
  • the voice process is continued according to the disconnected node, and the voice process can be restored when the user's communication is unexpectedly interrupted.
  • the user does not need to start the voice process from the beginning, which effectively saves the user's time.
  • the method includes: receiving an incoming call request from the user terminal through the user number; obtaining the historical selection of the user number for the conventional process; and performing the conventional process on the current voice process according to the historical selection s Choice.
  • each user’s voice call record will be stored.
  • the user’s historical voice call record will be queried, and the user’s choice of the conventional process in the historical voice call record will be determined. If the selections are the same, then the historical selection of the user number for the conventional process can be obtained.
  • the voice is set according to the language type previously selected by the user. If the user previously called into the customer service system to choose Mandarin, then the call will automatically select Mandarin, and then enter the next level of question classification options .
  • the user can quickly enter the consultation question, reduce the button selection layer by layer in the voice process, and directly obtain the desired customer service service.
  • a voice flow setting device including: a question classification module 310, an initial level setting module 320, and a voice flow setting module 330, wherein:
  • the question classification module 310 is configured to input the user’s consultation questions in a preset time period as test set data into the convergent customer service model to obtain the final question classification result; wherein the convergent customer service model passes the user’s consultation in the preset time period
  • the consultation questions and the standard classification of the consultation questions are obtained through repeated training.
  • the initial level setting module 320 is configured to set the initial level of the problem classification according to the number of consultations, the average time consumption, and the solution status of all the consultation problems in each problem classification in the final problem classification result.
  • the voice flow setting module 330 is configured to set the voice flow according to the initial level of the question classification.
  • the voice process setting device further includes: an initial question classification module, configured to input the consultation questions of the user within the preset time period as training set data and the standard classification of the consultation questions into the initial
  • the customer model obtains the initial problem classification result
  • the error calculation module is used to calculate the error between the initial problem classification result and the standard classification of the consulting question
  • the optimization module is used to use the error as the loss value of the network model, and Backpropagation back to the convolutional layer of the initial customer model to optimize the parameters of each convolutional layer of the customer model to obtain a convergent customer service model.
  • the voice process setting device further includes: an update question acquisition module, which is used to obtain updated user consultation questions, the duration of the consultation question, and the resolution status; and the level adjustment module is used to obtain updated user questions Questions, the duration and resolution status of the consultation questions, obtain the updated question classification of the user's consultation questions, and adjust the level of the question classification in real time; wherein, when the user calls in, the level of the adjusted question classification is Provide voice flow.
  • an update question acquisition module which is used to obtain updated user consultation questions, the duration of the consultation question, and the resolution status
  • the level adjustment module is used to obtain updated user questions Questions, the duration and resolution status of the consultation questions, obtain the updated question classification of the user's consultation questions, and adjust the level of the question classification in real time; wherein, when the user calls in, the level of the adjusted question classification is Provide voice flow.
  • the level adjustment module includes: an update question classification unit for inputting the updated user's consultation question into the convergent customer service model to obtain the updated question classification of the user's consultation question Statistic unit, used to count the number of consultations, average time consumption, and solution status of all consultation questions in the updated question classification of user’s consultation questions; Level adjustment unit, used to count the consultation times and average consumption of all consultation questions Time and resolution status adjust the level of the problem classification in real time.
  • the voice process setting device further includes: a parameter optimization module for optimizing and improving the parameters of the convergent customer service model according to the updated user's consultation question, the duration of the consultation question, and the solution status .
  • a voice call-in device for a customer service system including: a voice flow setting module 410, a selection instruction receiving module 420, and a sending module 430, wherein:
  • the voice flow setting module 410 is configured to obtain the voice flow set by the voice flow setting method described in the foregoing embodiment, and the voice flow includes options for question classification.
  • the selection instruction receiving module 420 is configured to receive a question selection instruction sent by the user terminal according to the question classification options
  • the sending module 430 is configured to send the next question classification option to the user terminal according to the question selection instruction when the question classification option is not the last level question classification option of the voice process; When the question classification option is the last level question classification option of the voice process, the answer to the target question is sent to the user terminal according to the question selection instruction.
  • the voice call-in device of the customer service system further includes: a receiving module, configured to receive a call-in request from a user terminal through a user number; and a query module, configured to query the user number according to the user number Corresponding historical call-in data; abnormal user number judgment module, used to judge whether the user number is an abnormal user number according to the historical call-in data; blacklist identification adding module, used if the user number is an abnormal user number , A blacklist identifier is added to the user number; a voice information sending module is configured to send voice information of the question classification options to the user terminal if the user number is not an abnormal user number.
  • the voice call-in device of the customer service system further includes: an instruction receiving module, configured to receive an instruction to select manual customer service sent by the user terminal; an average waiting time acquiring module, configured according to the selection Manual customer service instructions to obtain the average waiting time for accessing manual customer service at the current time; the voice sending module is used to send the voice information of the average waiting time for accessing manual customer service to the user terminal or to send a voice suggesting selecting intelligent customer service information.
  • the voice call-in device of the customer service system further includes: a receiving module, which is used to receive a call-in request from a user terminal through a user number; and a query module, which is used to query whether the user number exists according to the user number.
  • Unfinished voice call information corresponding to the user number Unfinished voice call information corresponding to the user number
  • disconnected node acquisition module used to obtain the voice process and disconnected node corresponding to the unfinished voice call information if the unfinished voice call information exists
  • voice sending The module is used to send the voice information of the problem classification options corresponding to the disconnected node to the user terminal according to the voice flow corresponding to the unfinished voice call information and the disconnected node.
  • the various modules in the above-mentioned voice flow setting device and the voice call-in device of the customer service system can be implemented in whole or in part by software, hardware and a combination thereof.
  • the above-mentioned modules may be embedded in the form of hardware or independent of the processor in the computer equipment, or may be stored in the memory of the computer equipment in the form of software, so that the processor can call and execute the operations corresponding to the above-mentioned modules.
  • a computer device is provided.
  • the computer device may be a server, and its internal structure diagram may be as shown in FIG. 6.
  • the computer equipment includes a processor, a memory, a network interface, and a database connected through a system bus. Among them, the processor of the computer device is used to provide calculation and control capabilities.
  • the memory of the computer device includes a non-volatile storage medium and an internal memory.
  • the non-volatile storage medium stores an operating system, a computer program, and a database.
  • the internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium.
  • the database of the computer equipment is used to store xxx data.
  • the network interface of the computer device is used to communicate with an external terminal through a network connection.
  • the computer program is executed by the processor to realize a voice flow setting method and a voice call-in method of the customer service system.
  • FIG. 6 is only a block diagram of part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied.
  • the specific computer device may Including more or fewer parts than shown in the figure, or combining some parts, or having a different arrangement of parts.
  • a computer device including a memory and a processor, the memory stores a computer program, and the processor implements the following steps when the processor executes the computer program:
  • the user's consultation questions within the preset time period are input into the convergent customer service model as the test set data to obtain the final problem classification result; wherein the convergent customer service model passes the user's consultation questions within the preset time period and the consultation The standard classification of the problem is obtained through repeated training;
  • the number of consultations, average time-consuming and solution status of all consultation problems in each problem classification to set an initial level for the problem classification
  • a computer-readable storage medium is provided, and the above-mentioned storage medium may be a non-volatile storage medium or a volatile storage medium.
  • a computer program is stored thereon, and the following steps are implemented when the computer program is executed by the processor:
  • the user's consultation questions within the preset time period are input into the convergent customer service model as the test set data to obtain the final question classification result; wherein the convergent customer service model passes the user's consultation questions within the preset time period and the consultation
  • the standard classification of the problem is obtained through repeated training;
  • the number of consultations, average time-consuming and solution status of all consultation problems in each problem classification to set an initial level for the problem classification
  • Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory.
  • Volatile memory may include random access memory (RAM) or external cache memory.
  • RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous chain Channel (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Business, Economics & Management (AREA)
  • General Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Computational Linguistics (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Biomedical Technology (AREA)
  • Finance (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Biophysics (AREA)
  • Strategic Management (AREA)
  • Health & Medical Sciences (AREA)
  • Accounting & Taxation (AREA)
  • Development Economics (AREA)
  • Multimedia (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Game Theory and Decision Science (AREA)
  • Databases & Information Systems (AREA)
  • Economics (AREA)
  • Marketing (AREA)
  • General Business, Economics & Management (AREA)
  • Telephonic Communication Services (AREA)

Abstract

本申请涉及一种语音流程设置方法、装置、计算机设备和存储介质。所述方法包括：将预设时间段内用户的咨询问题作为测试集数据输入收敛的客服模型，得到最终问题分类结果；其中，所述收敛的客服模型通过所述预设时间段内用户的咨询问题和所述咨询问题的标准分类反复训练获得；根据所述最终问题分类结果中，每个问题分类中所有咨询问题的咨询次数、平均耗时和解决状态来对所述问题分类设置初始级别；根据所述问题分类的初始级别，设置语音流程。采用本方法能够节省用户操作并提高用户咨询问题时的咨询效率。

Description

语音流程设置方法、装置、计算机设备和存储介质
本申请要求于2020年03月03日提交中国专利局、申请号为202010140296.2，发明名称为“语音流程设置方法、装置、计算机设备和存储介质”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本申请涉及计算机网络通信技术领域，特别是涉及一种语音流程设置方法、装置、计算机设备和存储介质。
 背景技术
客户服务系统简称客服系统，根据用户的需求建立问题及问题的答案，用户通过语音呼入系统，系统提供用户选项，用户根据选项进入问题获得问题答案，由于客服系统是通过简单的座机电话或者手机进行操作，选项都是通过电话的按键来实现，在进行流程问题设置时，同一级别问题不宜太多，因此，需要对用户的问题进行分类，用户根据分类一级一级进行选择操作直至获得目标问题和问题的答案。当然，客服系统也配置人工客服，通过人工客服能够快速了解用户的问题，并提供相应的答案，但是人工客服有限，在多个用户咨询时，整体人工客服繁忙，需要等待较长时间才能接入人工客服。
发明人意识到，现有的客服系统，问题分类比较死板，用户往往需要通过多级操作才能获得目标问题或者根本找不到目标问题，在接入人工客服时，人工客服又处于繁忙状态，导致用户问题解决的效率低。
 技术问题
基于此，有必要针对上述技术问题，提供一种能够提高用户问题咨询效率的语音流程设置方法、装置、计算机设备和存储介质。
 技术解决方案
一种语音流程设置方法，所述方法包括：
将预设时间段内用户的咨询问题作为测试集数据输入收敛的客服模型，得到最终问题分类结果；其中，所述收敛的客服模型通过所述预设时间段内用户的咨询问题和所述咨询问题的标准分类反复训练获得；
根据所述最终问题分类结果中，每个问题分类中所有咨询问题的咨询次数、平均耗时和解决状态来对所述问题分类设置初始级别；
根据所述问题分类的初始级别，设置语音流程。
一种客服系统语音呼入方法，所述方法包括：
获取上述语音流程设置方法设置的语音流程，所述语音流程包括问题分类的选项；
接收所述用户终端根据所述问题分类的选项发送的问题选择指令；
在所述问题分类的选项不为所述语音流程的最后一级的问题分类的选项时，根据所述问题选择指令向所述用户终端发送下一问题分类的选项；
在所述问题分类的选项为所述语音流程的最后一级的问题分类的选项时，根据所述问题选择指令向所述用户终端发送目标问题的答案。
一种语音流程设置装置，所述装置包括：
客服模型，得到最终问题分类结果；其中，所述收敛的客服模型通过所述预设时间段内用户的咨询问题和所述咨询问题的标准分类反复训练获得；
初始级别设置模块，用于根据所述最终问题分类结果中，每个问题分类中所有咨询问题的咨询次数、平均耗时和解决状态来对所述问题分类设置初始级别；
语音流程设置模块，用于根据所述问题分类的初始级别，设置语音流程。
一种客服系统语音呼入装置，所述装置包括：
语音流程设置模块，用于获取上述实施例所述的语音流程设置方法设置的语音流程，所述语音流程包括问题分类的选项；
选择指令接收模块，用于接收所述用户终端根据所述问题分类的选项发送的问题选择指令；
发送模块，用于在所述问题分类的选项不为所述语音流程的最后一级的问题分类的选项时，根据所述问题选择指令向所述用户终端发送下一问题分类的选项；在所述问题分类的选项为所述语音流程的最后一级的问题分类的选项时，根据所述问题选择指令向所述用户终端发送目标问题的答案。
一种计算机设备，包括存储器和处理器，所述存储器存储有计算机程序，所述处理器执行所述计算机程序时实现一种语音流程设置方法的步骤：
将预设时间段内用户的咨询问题作为测试集数据输入收敛的客服模型，得到最终问题分类结果；其中，所述收敛的客服模型通过所述预设时间段内用户的咨询问题和所述咨询问题的标准分类反复训练获得；
根据所述最终问题分类结果中，每个问题分类中所有咨询问题的咨询次数、平均耗时和解决状态来对所述问题分类设置初始级别；
根据所述问题分类的初始级别，设置语音流程。
一种计算机设备，包括存储器和处理器，所述存储器存储有计算机程序，其中，所述处理器执行所述计算机程序时实现一种客服系统语音呼入方法的步骤：
获取上述任一项所述语音流程设置方法设置的语音流程，所述语音流程包括问题分类的选项；
接收用户终端根据所述问题分类的选项发送的问题选择指令；
在所述问题分类的选项不为所述语音流程的最后一级的问题分类的选项时，根据所述问题选择指令向所述用户终端发送下一问题分类的选项；
在所述问题分类的选项为所述语音流程的最后一级的问题分类的选项时，根据所述问题选择指令向所述用户终端发送目标问题的答案。
一种计算机可读存储介质，其上存储有计算机程序，所述计算机程序被处理器执行时实现一种语音流程设置方法的步骤：
将预设时间段内用户的咨询问题作为测试集数据输入收敛的客服模型，得到最终问题分类结果；其中，所述收敛的客服模型通过所述预设时间段内用户的咨询问题和所述咨询问题的标准分类反复训练获得；
根据所述最终问题分类结果中，每个问题分类中所有咨询问题的咨询次数、平均耗时和解决状态来对所述问题分类设置初始级别；
根据所述问题分类的初始级别，设置语音流程。
一种计算机可读存储介质，其上存储有计算机程序，其中，所述计算机程序被处理器执行时实现一种客服系统语音呼入方法的步骤：
获取上述任一项所述语音流程设置方法设置的语音流程，所述语音流程包括问题分类的选项；
接收用户终端根据所述问题分类的选项发送的问题选择指令；
在所述问题分类的选项不为所述语音流程的最后一级的问题分类的选项时，根据所述问题选择指令向所述用户终端发送下一问题分类的选项；
在所述问题分类的选项为所述语音流程的最后一级的问题分类的选项时，根据所述问题选择指令向所述用户终端发送目标问题的答案。
 有益效果
上述语音流程设置方法、装置、计算机设备和存储介质，通过收敛的客服模型对用户的咨询问题进行分类，并结合问题分类中所有咨询问题的咨询次数、平均耗时和解决状态来对所述问题分类设置初始级别，进而设置关于所述问题分类的语音流程，能够结合用户的需要对问题分类的动态调整，让用户能够快速进入要咨询的问题及获得目标问题的答案。
 附图说明
图1为一个实施例中语音流程设置方法的应用场景图；
图2为一个实施例中语音流程设置方法的流程示意图；
图3为一个实施例中客服系统语音呼入方法的流程示意图；
图4为一个实施例中语音流程设置装置的结构框图；
图5为一个实施例中客服系统语音呼入装置的结构框图；
图6为一个实施例中计算机设备的内部结构图。
 本发明的最佳实施方式
为了使本申请的目的、技术方案及优点更加清楚明白，以下结合附图及实施例，对本申请进行进一步详细说明。应当理解，此处描述的具体实施例仅仅用以解释本申请，并不用于限定本申请。
本申请提供的语音流程设置方法，可以应用于如图1所示的应用环境中。其中，用户终端102通过网络与服务器104通过无线通信网络进行通信。用户通过用户终端102向所述服务器104提供咨询问题，服务器104将预设时间段内用户的咨询问题作为测试集数据输入收敛的客服模型，得到最终问题分类结果；其中，所述收敛的客服模型通过所述预设时间段内用户的咨询问题和所述咨询问题的标准分类反复训练获得；根据所述最终问题分类结果中，每个问题分类中所有咨询问题的咨询次数、平均耗时和解决状态来对所述问题分类设置初始级别；根据所述问题分类的初始级别，设置语音流程。其中，用户终端102可以但不限于是各种个人计算机、笔记本电脑、智能手机、平板电脑、座机、手机和便携式可穿戴设备，服务器104可以用独立的服务器或者是多个服务器组成的服务器集群来实现。
在一个实施例中，如图2所示，提供了一种语音流程设置方法，以该方法应用于图1中的服务器为例进行说明，包括以下步骤：
步骤S110，将预设时间段内用户的咨询问题作为测试集数据输入收敛的客服模型，得到最终问题分类结果；其中，所述收敛的客服模型通过所述预设时间段内用户的咨询问题和所述咨询问题的标准分类反复训练获得。
其中，所述预设之间段可以为指定的一个月、指定半年或者指定的一年，预设时间段不能太短，太短时间获得的用户的咨询问题不具有代表性，时间也不能太长，太长的时间导致一些已经取消的用户的咨询问题还存在于客服系统中，导致过多占用系统的资源。用户每次咨询问题的语音通话记录都会保存在客服系统中，以供后续分析查询使用。
其中，所述咨询问题的标准分类，通过人工对预设时间段内的咨询问题进行标注分类。
其中，客服模型可为一种人工神经网络模型，人工神经网络模型对用户的咨询问题根据咨询问题的标准分类进行打分分类，同一分数级别的问题作为一个问题分类。具体的，记录最近半年内每个用户呼入电话的咨询问题，将最近半年内所有用户的每个呼入电话的咨询问题作为样本数据，并手动标记每个咨询问题的标准分类，根据所述样本数据和所述标准分类对人工神经网络模型进行训练，获得收敛的客服模型。例如，最近半年内，用户呼入的咨询问题包括S1、S2、L1、L2，手动标记咨询问题S1、S2属于问题分类S，咨询问题L1、L2属于问题分类L，将咨询问题S1、S2、L1、L2输入人工神经网络模型，获得S1属于问题分类L打分为40、属于问题分类S打分为60，而预设S1属于问题分类S打分为80（此分数可以根据需要进行设置），此时需要对人工神经网络模型参数进行调整，以提高S1属于问题分类S打分，当S1属于问题分类S打分符合要求（不小于80）时，停止对人工神经网络模型进行训练，同理，对咨询问题S2、L1、L2采用同样的方法对人工神经网络模型进行训练，最终获得收敛的客服模型。
其中，最终问题分类结果包括所有问题分类、每个问题分类中所有咨询问题、每个问题分类中所有咨询问题的次数、每个问题分类中每个咨询问题的时长和每个问题分类中每个咨询问题的解决状态。例如，最终问题分类结果包括:问题分类S、L，问题分类S中的咨询问题S1、S2，问题分类S中的咨询问题的次数3（咨询问题S1被咨询了两次，咨询问题S2被咨询了一次），问题分类L中的咨询问题L1、L2，问题分类L中的咨询问题的次数5（咨询问题L1被咨询了两次，咨询问题L2被咨询了三次），每一次咨询问题S1的时长和解决状态，每一次咨询问题S2的时长和解决状态，每一次咨询问题L1的时长和解决状态, 每一次咨询问题L2的时长和解决状态。
具体的，客服系统根据预设时间段内，已经存储用户语音通话记录，来获取用户的咨询问题、咨询问题的时长和解决状态。
步骤S120，根据所述最终问题分类结果中，每个问题分类中所有咨询问题的咨询次数、平均耗时和解决状态来对所述问题分类设置初始级别。
其中，可以设置用户的每一次有效呼入为一次咨询，步骤S120中咨询次数为所有咨询问题总的咨询次数，每个咨询问题都可能被多个用户咨询。平均耗时等于所有咨询问题的总耗时除以所述咨询次数。解决状态包括问题完全解决、问题部分解决、问题未解决。可选的，可以在用户咨询结束之前让用户反馈解决状态。
其中，级别是指用户呼入客服系统时的选项级别，如用户刚呼入时，提供给用户的是一级选项级别，在根据一级选项级别选择其中一个选项后，进入二级选项级别，以此类推，直到对应到最后目标问题的答案。
在其中一个实施例中，问题分类的咨询次数多，可以设置此问题分类的级别高，级别越高的问题分类越靠前，让用户能够快速进入要咨询的问题及获得答案；或者问题分类的平均耗时低和解决状态为问题完全解决时设置所述问题分类的级别高。
步骤S130，根据所述问题分类的初始级别，设置语音流程。
其中，用户通过座机或者手机呼入系统，在呼入系统之前，客服模型已经根据已有的用户呼入数据建立了语音流程，语音流程包括一级问题分类的选项、二级问题分类的选项……N级问题分类的选项和目标问题的答案，用户根据语音流程提供的问题分类的选项，依次选择，直至获得目标问题的答案。
上述语音流程设置方法中，通过收敛的客服模型对用户的咨询问题进行分类，并结合问题分类中所有咨询问题的咨询次数、平均耗时和解决状态来对所述问题分类设置初始级别，进而设置关于所述问题分类的语音流程，能够结合用户的需要对问题分类的动态调整，让用户能够快速进入要咨询的问题及获得目标问题的答案。
在其中一个实施例中，所述语音流程设置方法，还包括：将所述预设时间段内用户的咨询问题作为训练集数据和所述咨询问题的标准分类输入初始客户模型，得到初始问题分类结果；计算所述初始问题分类结果与所述咨询问题的标准分类的误差；将所述误差作为网络模型的损失值，并反向传播回所述初始客户模型的卷积层，以优化客户模型的各卷积层的参数，得到收敛的客服模型。
其中，所述初始问题分类结果与所述咨询问题的标准分类的误差，即所述初始问题分类结果相对于所述咨询问题的标准分类错误的概率。通过预设时间段内用户的咨询问题对所述初始客户模型反复训练，能够获得分类越来越准确的客户模型。
在其中一个实施例中，所述语音流程设置方法，还包括：获取更新的用户的咨询问题、咨询问题的时长和解决状态；根据更新的用户的咨询问题、咨询问题的时长和解决状态，获取所述更新的用户的咨询问题的问题分类，实时调整所述问题分类的级别；其中，在用户呼入时，根据调整后的问题分类的级别提供语音流程。
其中，用户通过语音流程呼入系统时，后台会实时记录用户的呼入记录。所述呼入记录包括记录用户的咨询问题、咨询问题的时长和解决状态。问题分类的咨询次数多，可以设置此问题分类的级别高，级别越高的问题分类越靠前，让用户能够快速进入要咨询的问题及获得答案；或者问题分类的平均耗时低和解决状态为问题完全解决时设置所述问题分类的级别高。随着用户根据所述语音流程呼入的次数越来越多，其咨询业务也在不断更新，需要对问题分类进行实时调整，通过实时调整所述问题分类的级别能够满足业务更新的需求。
在其中一个实施例中，所述根据更新的用户的咨询问题、咨询问题的时长和解决状态，实时调整问题分类的级别步骤包括：将所述更新的用户的咨询问题输入所述收敛的客服模型，得到所述更新的用户的咨询问题的问题分类；统计所述更新的用户的咨询问题的问题分类中，所有咨询问题的咨询次数、平均耗时和解决状态；根据所有咨询问题的咨询次数、平均耗时和解决状态对所述问题分类实时调整级别。
其中，通过更新的用户的咨询问题的问题分类，并根据所述问题分类重新计算所有咨询问题的咨询次数、平均耗时和解决状态，来实时调整每个问题分类的级别，能够实现在业务更改时，实时对问题分类的级别进行调整，从而改进语音流程。
在其中一个实施例中，在所述获取更新的用户的咨询问题、咨询问题的时长和解决状态步骤之后，包括：根据所述更新的用户的咨询问题、咨询问题的时长和解决状态，优化和改进收敛的客服模型的参数。
其中，系统在统计更新的用户的咨询问题、咨询问题的时长和解决状态后，发现更新的用户的咨询问题的时长过长，则获取用户本次呼入的语音通话记录，分析是因为更新的用户的咨询问题的分类不准确导致咨询的时长过长，则需要对收敛的客服模型的参数进行改进，使得更新的用户的咨询问题分到用户选择较多的问题分类中；如果发现更新的用户的咨询问题的解决状态是问题部分解决、问题未解决，则获取用户本次呼入的语音通话记录，分析是因为不存在此更新的用户的咨询问题的问题分类，则需要对收敛的客服模型的参数进行改进或者设置新的咨询问题的标准分类，使得更新的用户的咨询问题分到新的咨询问题的标准分类中。
在一个实施例中，如图3所示，提供了一种客服系统语音呼入方法，以该方法应用于图1中的服务器为例进行说明，包括以下步骤：
步骤S210，获取上述实施例所述的语音流程设置方法设置的语音流程，所述语音流程包括问题分类的选项。
其中，语音流程包括一级问题分类的选项、二级问题分类的选项……N级问题分类的选项和目标问题的答案。
步骤S220，接收所述用户终端根据所述问题分类的选项发送的问题选择指令。
其中，所述用户终端包括手机、座机等通讯设备，所述问题选择指令通过用户终端发出，例如，用户在呼入系统后，语音流程提供一级问题分类的选项，一级问题分类的选项具体为“普通话服务请按1，English请按2,智能客服请按3，转人工客服请按4”，用户在用户终端按1，则用户终端发出选择普通话服务的问题选择指令。
步骤S230，在所述问题分类的选项不为所述语音流程的最后一级的问题分类的选项时，根据所述问题选择指令向所述用户终端发送下一问题分类选项。其中，下一问题分类选项是根据所述问题选择指令来获取，例如，用户根据一级问题分类的选项通过用户终端发送问题选择指令，系统根据问题选择指令进入相应的二级问题分类的选项，所述二级问题分类的选项为一级问题分类的选项的下一问题分类选项。
例如，语音流程包括一级问题分类的选项、二级问题分类的选项……N级问题分类的选项和目标问题的答案，N级问题分类的选项为所述语音流程的最后一级的问题分类的选项。步骤S240，在所述问题分类的选项为所述语音流程的最后一级的问题分类的选项时，根据所述问题选择指令向所述用户终端发送目标问题的答案。
其中，目标问题的答案是由用户根据所述语音流程的最后一级的问题分类的选项做出的选择确定的。
具体的，在用户终端，用户通过手机或者座机接听系统发过来的多个问题分类，即问题分类对应的按键选项，用户通过选择相应的按键来选择问题分类，然后进入下一级多个问题分类，然后再通过按键来选择对应的问题分类，直到最后选择到目标问题，接听系统给出的目标问题的答案。当然，在进入最后一级时，发现不存在目标问题，则可以根据语音提示进入上一级，重新对问题分类进行选择，或者直接选择人工客服。例如，所述语音流程其中一个问题分类的选项包括：“查询话费请按1”“查询流量请按2”“办理宽带请按3”；用户按下按键1，则根据用户的选择，向用户发送话费信息，所述问题分类的选项为最后一级的问题分类的选项，话费信息即为目标问题的答案；如果用户按下按键3，则根据用户的选择，进入下一问题分类的选项，下一问题分类的选项包括：“宽带类型了解按1”“宽带账户密码查询按2”，此时如果用户按下按键2，向用户发送宽带账户密码，则下一问题分类的选项为最后一级的问题分类的选项，宽带账户密码为目标问题的答案。
上述一种客服系统语音呼入方法，通过所述的语音流程设置方法设置的语音流程，能够缩短用户咨询问题的时间，提高用户的咨询效率。
在其中一个实施例中，在步骤S210之后，包括：接收用户终端通过用户号码的呼入请求；根据所述用户号码查询所述用户号码对应的历史呼入数据；根据所述历史呼入数据判断所述用户号码是否为异常用户号码；如果所述用户号码为异常用户号码，则对所述用户号码添加黑名单标识；如果所述用户号码不为异常用户号码，则向所述用户终端发送所述问题分类的选项的语音信息。
其中，用户号码可以为手机号码，也可为座机号码，用户通过手机或者座机呼入客服系统。历史呼入数据为用户历史呼入所述客服系统的通话记录和语音流程。
其中，所述根据所述历史呼入数据判断所述用户号码是否为异常用户号码步骤具体为：根据所述历史呼入数据，计算所述用户异常问题（骚扰电话）的咨询次数与正常业务咨询次数之比；在所述用户异常问题（骚扰电话）的咨询次数与正常业务咨询次数之比超过预设值时，则判断所述用户号码为异常用户号码；在所述用户异常问题（骚扰电话）的咨询次数与正常业务咨询次数之比不超过预设值时，则判断所述用户号码不为异常用户号码。
通过本实施例所述方法，能够有效减少骚扰电话的呼入，并且能够降低骚扰电话进入人工客服的概率，能够避免不必要的人工客服资源浪费，并提高了正常呼入接入人工客服的概率。
在其中一个实施例中，在步骤S210之后，包括：接收所述用户终端发送的选择人工客服的指令；根据所述选择人工客服的指令，获取当前时间接入人工客服的平均等待时间；向所述用户终端发送所述接入人工客服的平均等待时间的语音信息或者发送建议选择智能客服的语音信息。
其中，在接入人工客服的平均等待时间的情况下，向所述用户终端发送建议选择智能客服的语音信息。
本实施例通过向用户反馈接入人工客服的平均等待时间，便于用户能够根据个人需要是否继续等待接入人工客服还是选择其它方式来咨询问题，有效的节约了用户的时间，并且在接入人工客服的平均等待时间长的情况下，建议用户选择智能客服，有效的提高了用户咨询问题的效率。
在其中一个实施例中，在步骤S210之后，包括：接收用户终端通过用户号码的呼入请求；根据所述用户号码查询是否存在所述用户号码对应的未完成的语音通话信息；如果存在所述未完成的语音通话信息，获取所述未完成的语音通话信息对应的语音流程和断线节点；根据所述未完成的语音通话信息对应的语音流程和断线节点，向用户终端发送所述断线节点对应的问题分类的选项的语音信息。
其中，所述未完成的语音通话信息是指用户呼入客服系统是发送非正常断线，例如，用户终端信号不好发生通信中断、用户终端出现故障发生通信中断或者用户终端电量耗尽发送通信中断。所述未完成的语音通话信息对应的语音流程和断线节点，是指断线前用户正在进行语音流程和语音流程进入的问题分类级别，例如，断线前用户正在咨询话费查询流程，断线时已经进入到二级问题分类的选项，则断线节点为二级问题分类的选项。
本实施例，根据断线节点继续语音流程，能够用户通讯意外中断的情况下，对语音流程进行恢复，用户不必从头开始语音流程，有效的节省了用户的时间。
在其中一个实施例中，在步骤S210之后，包括：接收用户终端通过用户号码的呼入请求；获取所述用户号码对常规流程的历史选择；根据所述历史选择对当前的语音流程进行常规流程的选择。
其中，每次用户的语音通话记录都会存储，在用户下次呼入时，会查询用户的历史语音通话记录，并判断所述历史语音通话记录中用户对常规流程的选择，如果用户对常规流程的选择都相同，则可以获取所述用户号码对常规流程的历史选择。
例如，用户呼入客服系统后，根据用户之前选择的语言种类来设置语音，如用户之前呼入客服系统都是选择普通话，则此次呼入自动选择普通话，然后进入下一级问题分类的选项。
本实施例中，通过对常规流程的自动选择，能够让用户快速进入咨询问题，减少语音流程一层一层的按键选择，直接获得想要的客服服务。
应该理解的是，虽然图2-3的流程图中的各个步骤按照箭头的指示依次显示，但是这些步骤并不是必然按照箭头指示的顺序依次执行。除非本文中有明确的说明，这些步骤的执行并没有严格的顺序限制，这些步骤可以以其它的顺序执行。而且，图2-3中的至少一部分步骤可以包括多个子步骤或者多个阶段，这些子步骤或者阶段并不必然是在同一时刻执行完成，而是可以在不同的时刻执行，这些子步骤或者阶段的执行顺序也不必然是依次进行，而是可以与其它步骤或者其它步骤的子步骤或者阶段的至少一部分轮流或者交替地执行。
在一个实施例中，如图4所示，提供了一种语音流程设置装置，包括：问题分类模块310、初始级别设置模块320和语音流程设置模块330，其中：
问题分类模块310，用于将预设时间段内用户的咨询问题作为测试集数据输入收敛的客服模型，得到最终问题分类结果；其中，所述收敛的客服模型通过所述预设时间段内用户的咨询问题和所述咨询问题的标准分类反复训练获得。
初始级别设置模块320，用于根据所述最终问题分类结果中，每个问题分类中所有咨询问题的咨询次数、平均耗时和解决状态来对所述问题分类设置初始级别。
语音流程设置模块330，用于根据所述问题分类的初始级别，设置语音流程。
在其中一个实施例中，所述语音流程设置装置，还包括：初始问题分类模块，用于将所述预设时间段内用户的咨询问题作为训练集数据和所述咨询问题的标准分类输入初始客户模型，得到初始问题分类结果；误差计算模块，用于计算所述初始问题分类结果与所述咨询问题的标准分类的误差；优化模块，用于将所述误差作为网络模型的损失值，并反向传播回所述初始客户模型的卷积层，以优化客户模型的各卷积层的参数，得到收敛的客服模型。
在其中一个实施例中，所述语音流程设置装置，还包括：更新问题获取模块，用于获取更新的用户的咨询问题、咨询问题的时长和解决状态；级别调整模块，用于根据更新的用户的咨询问题、咨询问题的时长和解决状态，获取所述更新的用户的咨询问题的问题分类，实时调整所述问题分类的级别；其中，在用户呼入时，根据调整后的问题分类的级别提供语音流程。
在其中一个实施例中，所述级别调整模块包括：更新问题分类单元，用于将所述更新的用户的咨询问题输入所述收敛的客服模型，得到所述更新的用户的咨询问题的问题分类；统计单元，用于统计所述更新的用户的咨询问题的问题分类中，所有咨询问题的咨询次数、平均耗时和解决状态；级别调整单元，用于根据所有咨询问题的咨询次数、平均耗时和解决状态对所述问题分类实时调整级别。
在其中一个实施例中，所述语音流程设置装置，还包括：参数优化模块，用于根据所述更新的用户的咨询问题、咨询问题的时长和解决状态，优化和改进收敛的客服模型的参数。
在一个实施例中，提供了一种客服系统语音呼入装置，包括：语音流程设置模块410、选择指令接收模块420和发送模块430，其中：
语音流程设置模块410，用于获取上述实施例所述的语音流程设置方法设置的语音流程，所述语音流程包括问题分类的选项。
选择指令接收模块420，用于接收所述用户终端根据所述问题分类的选项发送的问题选择指令；
发送模块430，用于在所述问题分类的选项不为所述语音流程的最后一级的问题分类的选项时，根据所述问题选择指令向所述用户终端发送下一问题分类的选项；在所述问题分类的选项为所述语音流程的最后一级的问题分类的选项时，根据所述问题选择指令向所述用户终端发送目标问题的答案。
在其中一个实施例中，所述客服系统语音呼入装置，还包括：接收模块，用于接收用户终端通过用户号码的呼入请求；查询模块，用于根据所述用户号码查询所述用户号码对应的历史呼入数据；异常用户号码判断模块，用于根据所述历史呼入数据判断所述用户号码是否为异常用户号码；黑名单标识添加模块，用于如果所述用户号码为异常用户号码，则对所述用户号码添加黑名单标识；语音信息发送模块，用于如果所述用户号码不为异常用户号码，则向所述用户终端发送所述问题分类的选项的语音信息。
在其中一个实施例中，所述客服系统语音呼入装置，还包括：指令接收模块，用于接收所述用户终端发送的选择人工客服的指令；平均等待时间获取模块，用于根据所述选择人工客服的指令，获取当前时间接入人工客服的平均等待时间；语音发送模块，用于向所述用户终端发送所述接入人工客服的平均等待时间的语音信息或者发送建议选择智能客服的语音信息。
在其中一个实施例中，所述客服系统语音呼入装置，还包括：接收模块，用于接收用户终端通过用户号码的呼入请求；查询模块，用于根据所述用户号码查询是否存在所述用户号码对应的未完成的语音通话信息；断线节点获取模块，用于如果存在所述未完成的语音通话信息，获取所述未完成的语音通话信息对应的语音流程和断线节点；语音发送模块，用于根据所述未完成的语音通话信息对应的语音流程和断线节点，向用户终端发送所述断线节点对应的问题分类的选项的语音信息。
关于语音流程设置装置、客服系统语音呼入装置的具体限定可以参见上文中对于语音流程设置方法、客服系统语音呼入方法的限定，在此不再赘述。上述语音流程设置装置、客服系统语音呼入装置中的各个模块可全部或部分通过软件、硬件及其组合来实现。上述各模块可以硬件形式内嵌于或独立于计算机设备中的处理器中，也可以以软件形式存储于计算机设备中的存储器中，以便于处理器调用执行以上各个模块对应的操作。
在一个实施例中，提供了一种计算机设备，该计算机设备可以是服务器，其内部结构图可以如图6所示。该计算机设备包括通过系统总线连接的处理器、存储器、网络接口和数据库。其中，该计算机设备的处理器用于提供计算和控制能力。该计算机设备的存储器包括非易失性存储介质、内存储器。该非易失性存储介质存储有操作系统、计算机程序和数据库。该内存储器为非易失性存储介质中的操作系统和计算机程序的运行提供环境。该计算机设备的数据库用于存储xxx数据。该计算机设备的网络接口用于与外部的终端通过网络连接通信。该计算机程序被处理器执行时以实现一种语音流程设置方法、客服系统语音呼入方法。
本领域技术人员可以理解，图6中示出的结构，仅仅是与本申请方案相关的部分结构的框图，并不构成对本申请方案所应用于其上的计算机设备的限定，具体的计算机设备可以包括比图中所示更多或更少的部件，或者组合某些部件，或者具有不同的部件布置。
在一个实施例中，提供了一种计算机设备，包括存储器和处理器，该存储器存储有计算机程序，该处理器执行计算机程序时实现以下步骤：
将预设时间段内用户的咨询问题作为测试集数据输入收敛的客服模型，得到最终问题分类结果；其中，所述收敛的客服模型通过所述预设时间段内用户的咨询问题和所述咨询问题的标准分类反复训练获得；
根据所述最终问题分类结果中，每个问题分类中所有咨询问题的咨询次数、平均耗时和解决状态来对所述问题分类设置初始级别；
根据所述问题分类的初始级别，设置语音流程。
在一个实施例中，提供了一种计算机可读存储介质，上述存储介质可以是非易失性存储介质，也可以是易失性存储介质。其上存储有计算机程序，计算机程序被处理器执行时实现以下步骤：
将预设时间段内用户的咨询问题作为测试集数据输入收敛的客服模型，得到最终问题分类结果；其中，所述收敛的客服模型通过所述预设时间段内用户的咨询问题和所述咨询问题的标准分类反复训练获得；
根据所述最终问题分类结果中，每个问题分类中所有咨询问题的咨询次数、平均耗时和解决状态来对所述问题分类设置初始级别；
根据所述问题分类的初始级别，设置语音流程。
本领域普通技术人员可以理解实现上述实施例方法中的全部或部分流程，是可以通过计算机程序来指令相关的硬件来完成，所述的计算机程序可存储于一非易失性计算机可读取存储介质中，该计算机程序在执行时，可包括如上述各方法的实施例的流程。其中，本申请所提供的各实施例中所使用的对存储器、存储、数据库或其它介质的任何引用，均可包括非易失性和/或易失性存储器。非易失性存储器可包括只读存储器（ROM）、可编程ROM（PROM）、电可编程ROM（EPROM）、电可擦除可编程ROM（EEPROM）或闪存。易失性存储器可包括随机存取存储器（RAM）或者外部高速缓冲存储器。作为说明而非局限，RAM以多种形式可得，诸如静态RAM（SRAM）、动态RAM（DRAM）、同步DRAM（SDRAM）、双数据率SDRAM（DDRSDRAM）、增强型SDRAM（ESDRAM）、同步链路（Synchlink） DRAM（SLDRAM）、存储器总线（Rambus）直接RAM（RDRAM）、直接存储器总线动态RAM（DRDRAM）、以及存储器总线动态RAM（RDRAM）等。
以上实施例的各技术特征可以进行任意的组合，为使描述简洁，未对上述实施例中的各个技术特征所有可能的组合都进行描述，然而，只要这些技术特征的组合不存在矛盾，都应当认为是本说明书记载的范围。
以上所述实施例仅表达了本申请的几种实施方式，其描述较为具体和详细，但并不能因此而理解为对发明专利范围的限制。应当指出的是，对于本领域的普通技术人员来说，在不脱离本申请构思的前提下，还可以做出若干变形和改进，这些都属于本申请的保护范围。因此，本申请专利的保护范围应以所附权利要求为准。

Claims (20)

  1. 一种语音流程设置方法，其中，所述方法包括：
    将预设时间段内用户的咨询问题作为测试集数据输入收敛的客服模型，得到最终问题分类结果；其中，所述收敛的客服模型通过所述预设时间段内用户的咨询问题和所述咨询问题的标准分类反复训练获得；
    根据所述最终问题分类结果中，每个问题分类中所有咨询问题的咨询次数、平均耗时和解决状态来对所述问题分类设置初始级别；
    根据所述问题分类的初始级别，设置语音流程。
  2. 根据权利要求1所述的方法，其中，还包括：
    将所述预设时间段内用户的咨询问题作为训练集数据和所述咨询问题的标准分类输入初始客户模型，得到初始问题分类结果；
    计算所述初始问题分类结果与所述咨询问题的标准分类的误差；
    将所述误差作为网络模型的损失值，并反向传播回所述初始客户模型的卷积层，以优化客户模型的各卷积层的参数，得到收敛的客服模型。
  3. 根据权利要求1所述的方法，其中，还包括：
    获取更新的用户的咨询问题、咨询问题的时长和解决状态；
    根据更新的用户的咨询问题、咨询问题的时长和解决状态，获取所述更新的用户的咨询问题的问题分类，实时调整所述问题分类的级别；其中，在用户呼入时，根据调整后的问题分类的级别提供语音流程。
  4. 一种客服系统语音呼入方法，其中，所述方法包括：
    获取权利要求1-3任一项所述语音流程设置方法设置的语音流程，所述语音流程包括问题分类的选项；
    接收用户终端根据所述问题分类的选项发送的问题选择指令；
    在所述问题分类的选项不为所述语音流程的最后一级的问题分类的选项时，根据所述问题选择指令向所述用户终端发送下一问题分类的选项；
    在所述问题分类的选项为所述语音流程的最后一级的问题分类的选项时，根据所述问题选择指令向所述用户终端发送目标问题的答案。
  5. 根据权利要求4所述的方法，其中，在所述获取权利要求1-3任一项所述语音流程设置方法设置的语音流程，所述语音流程包括问题分类的选项步骤之后，包括：
    接收用户终端通过用户号码的呼入请求；
    根据所述用户号码查询所述用户号码对应的历史呼入数据；
    根据所述历史呼入数据判断所述用户号码是否为异常用户号码；
    如果所述用户号码为异常用户号码，则对所述用户号码添加黑名单标识；
    如果所述用户号码不为异常用户号码，则向所述用户终端发送所述问题分类的选项的语音信息。
  6. 根据权利要求4所述的方法，其中，在所述获取权利要求1-5任一项所述语音流程设置方法设置的语音流程，所述语音流程包括问题分类的选项步骤之后，包括：
    接收用户终端通过用户号码的呼入请求；
    根据所述用户号码查询是否存在所述用户号码对应的未完成的语音通话信息；
    如果存在所述未完成的语音通话信息，获取所述未完成的语音通话信息对应的语音流程和断线节点；
    根据所述未完成的语音通话信息对应的语音流程和断线节点，向用户终端发送所述断线节点对应的问题分类的选项的语音信息。
  7. 一种语音流程设置装置，其中，所述装置包括：
    问题分类模块，用于将预设时间段内用户的咨询问题作为测试集数据输入收敛的客服模型，得到最终问题分类结果；其中，所述收敛的客服模型通过所述预设时间段内用户的咨询问题和所述咨询问题的标准分类反复训练获得；
    初始级别设置模块，用于根据所述最终问题分类结果中，每个问题分类中所有咨询问题的咨询次数、平均耗时和解决状态来对所述问题分类设置初始级别；
    语音流程设置模块，用于根据所述问题分类的初始级别，设置语音流程。
  8. 一种客服系统语音呼入装置，其中，所述装置包括：
    语音流程设置模块，用于获取权利要求1-3任一项所述的语音流程设置方法设置的语音流程，所述语音流程包括问题分类的选项；
    选择指令接收模块，用于接收所述用户终端根据所述问题分类的选项发送的问题选择指令；
    发送模块，用于在所述问题分类的选项不为所述语音流程的最后一级的问题分类的选项时，根据所述问题选择指令向所述用户终端发送下一问题分类的选项；在所述问题分类的选项为所述语音流程的最后一级的问题分类的选项时，根据所述问题选择指令向所述用户终端发送目标问题的答案。
  9. 一种计算机设备，包括存储器和处理器，所述存储器存储有计算机程序，其中，所述处理器执行所述计算机程序时实现一种语音流程设置方法的步骤：
    将预设时间段内用户的咨询问题作为测试集数据输入收敛的客服模型，得到最终问题分类结果；其中，所述收敛的客服模型通过所述预设时间段内用户的咨询问题和所述咨询问题的标准分类反复训练获得；
    根据所述最终问题分类结果中，每个问题分类中所有咨询问题的咨询次数、平均耗时和解决状态来对所述问题分类设置初始级别；
    根据所述问题分类的初始级别，设置语音流程。
  10. 根据权利要求9所述的计算机设备，其中，还包括：
    将所述预设时间段内用户的咨询问题作为训练集数据和所述咨询问题的标准分类输入初始客户模型，得到初始问题分类结果；
    计算所述初始问题分类结果与所述咨询问题的标准分类的误差；
    将所述误差作为网络模型的损失值，并反向传播回所述初始客户模型的卷积层，以优化客户模型的各卷积层的参数，得到收敛的客服模型。
  11. 根据权利要求9所述的计算机设备，其中，还包括：
    获取更新的用户的咨询问题、咨询问题的时长和解决状态；
    根据更新的用户的咨询问题、咨询问题的时长和解决状态，获取所述更新的用户的咨询问题的问题分类，实时调整所述问题分类的级别；其中，在用户呼入时，根据调整后的问题分类的级别提供语音流程。
  12. 一种计算机设备，包括存储器和处理器，所述存储器存储有计算机程序，其中，所述处理器执行所述计算机程序时实现一种客服系统语音呼入方法的步骤：
    获取权利要求1-3任一项所述语音流程设置方法设置的语音流程，所述语音流程包括问题分类的选项；
    接收用户终端根据所述问题分类的选项发送的问题选择指令；
    在所述问题分类的选项不为所述语音流程的最后一级的问题分类的选项时，根据所述问题选择指令向所述用户终端发送下一问题分类的选项；
    在所述问题分类的选项为所述语音流程的最后一级的问题分类的选项时，根据所述问题选择指令向所述用户终端发送目标问题的答案。
  13. 根据权利要求12所述的计算机设备，其中，在所述获取权利要求1-3任一项所述语音流程设置方法设置的语音流程，所述语音流程包括问题分类的选项步骤之后，包括：
    接收用户终端通过用户号码的呼入请求；
    根据所述用户号码查询所述用户号码对应的历史呼入数据；
    根据所述历史呼入数据判断所述用户号码是否为异常用户号码；
    如果所述用户号码为异常用户号码，则对所述用户号码添加黑名单标识；
    如果所述用户号码不为异常用户号码，则向所述用户终端发送所述问题分类的选项的语音信息。
  14. 根据权利要求12所述的计算机设备，其中，在所述获取权利要求1-5任一项所述语音流程设置方法设置的语音流程，所述语音流程包括问题分类的选项步骤之后，包括：
    接收用户终端通过用户号码的呼入请求；
    根据所述用户号码查询是否存在所述用户号码对应的未完成的语音通话信息；
    如果存在所述未完成的语音通话信息，获取所述未完成的语音通话信息对应的语音流程和断线节点；
    根据所述未完成的语音通话信息对应的语音流程和断线节点，向用户终端发送所述断线节点对应的问题分类的选项的语音信息。
  15. 一种计算机可读存储介质，其上存储有计算机程序，其中，所述计算机程序被处理器执行时实现一种语音流程设置方法的步骤：
    将预设时间段内用户的咨询问题作为测试集数据输入收敛的客服模型，得到最终问题分类结果；其中，所述收敛的客服模型通过所述预设时间段内用户的咨询问题和所述咨询问题的标准分类反复训练获得；
    根据所述最终问题分类结果中，每个问题分类中所有咨询问题的咨询次数、平均耗时和解决状态来对所述问题分类设置初始级别；
    根据所述问题分类的初始级别，设置语音流程。
  16. 根据权利要求15所述的计算机可读存储介质，其中，还包括：
    将所述预设时间段内用户的咨询问题作为训练集数据和所述咨询问题的标准分类输入初始客户模型，得到初始问题分类结果；
    计算所述初始问题分类结果与所述咨询问题的标准分类的误差；
    将所述误差作为网络模型的损失值，并反向传播回所述初始客户模型的卷积层，以优化客户模型的各卷积层的参数，得到收敛的客服模型。
  17. 根据权利要求15所述的计算机可读存储介质，其中，还包括：
    获取更新的用户的咨询问题、咨询问题的时长和解决状态；
    根据更新的用户的咨询问题、咨询问题的时长和解决状态，获取所述更新的用户的咨询问题的问题分类，实时调整所述问题分类的级别；其中，在用户呼入时，根据调整后的问题分类的级别提供语音流程。
  18. 一种计算机可读存储介质，其上存储有计算机程序，其中，所述计算机程序被处理器执行时实现一种客服系统语音呼入方法的步骤：
    获取权利要求1-3任一项所述语音流程设置方法设置的语音流程，所述语音流程包括问题分类的选项；
    接收用户终端根据所述问题分类的选项发送的问题选择指令；
    在所述问题分类的选项不为所述语音流程的最后一级的问题分类的选项时，根据所述问题选择指令向所述用户终端发送下一问题分类的选项；
    在所述问题分类的选项为所述语音流程的最后一级的问题分类的选项时，根据所述问题选择指令向所述用户终端发送目标问题的答案。
  19. 根据权利要求18所述的计算机可读存储介质，其中，在所述获取权利要求1-3任一项所述语音流程设置方法设置的语音流程，所述语音流程包括问题分类的选项步骤之后，包括：
    接收用户终端通过用户号码的呼入请求；
    根据所述用户号码查询所述用户号码对应的历史呼入数据；
    根据所述历史呼入数据判断所述用户号码是否为异常用户号码；
    如果所述用户号码为异常用户号码，则对所述用户号码添加黑名单标识；
    如果所述用户号码不为异常用户号码，则向所述用户终端发送所述问题分类的选项的语音信息。
  20. 根据权利要求18所述的计算机可读存储介质，其中，在所述获取权利要求1-5任一项所述语音流程设置方法设置的语音流程，所述语音流程包括问题分类的选项步骤之后，包括：
    接收用户终端通过用户号码的呼入请求；
    根据所述用户号码查询是否存在所述用户号码对应的未完成的语音通话信息；
    如果存在所述未完成的语音通话信息，获取所述未完成的语音通话信息对应的语音流程和断线节点；
    根据所述未完成的语音通话信息对应的语音流程和断线节点，向用户终端发送所述断线节点对应的问题分类的选项的语音信息。
PCT/CN2020/119378 2020-03-03 2020-09-30 语音流程设置方法、装置、计算机设备和存储介质 WO2021174840A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202010140296.2A CN111460212A (zh) 2020-03-03 2020-03-03 语音流程设置方法、装置、计算机设备和存储介质
CN202010140296.2 2020-03-03

Publications (1)

Publication Number Publication Date
WO2021174840A1 true WO2021174840A1 (zh) 2021-09-10

Family

ID=71684167

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/119378 WO2021174840A1 (zh) 2020-03-03 2020-09-30 语音流程设置方法、装置、计算机设备和存储介质

Country Status (2)

Country Link
CN (1) CN111460212A (zh)
WO (1) WO2021174840A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114374769A (zh) * 2021-12-01 2022-04-19 恒安嘉新（北京）科技股份公司 一种异常号码的获取方法、装置、服务器和存储介质
CN114710471A (zh) * 2022-03-21 2022-07-05 京东科技信息技术有限公司 基于网络的客服语音通信方法、装置、电子设备及介质

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111460212A (zh) * 2020-03-03 2020-07-28 平安科技（深圳）有限公司 语音流程设置方法、装置、计算机设备和存储介质

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20060210046A1 (en) * 2002-09-06 2006-09-21 Smith Benjamin V Real time customer service data manipulation to allow multiple services per trigger type
CN103533186A (zh) * 2013-09-23 2014-01-22 安徽科大讯飞信息科技股份有限公司 一种基于语音呼叫的业务流程实现方法及系统
JP2019152816A (ja) * 2018-03-06 2019-09-12 株式会社日立情報通信エンジニアリング 通話音声処理システム及び通話音声処理方法
CN110704618A (zh) * 2019-09-20 2020-01-17 阿里巴巴集团控股有限公司 确定对话数据对应的标准问题的方法及装置
CN111460212A (zh) * 2020-03-03 2020-07-28 平安科技（深圳）有限公司 语音流程设置方法、装置、计算机设备和存储介质

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20060210046A1 (en) * 2002-09-06 2006-09-21 Smith Benjamin V Real time customer service data manipulation to allow multiple services per trigger type
CN103533186A (zh) * 2013-09-23 2014-01-22 安徽科大讯飞信息科技股份有限公司 一种基于语音呼叫的业务流程实现方法及系统
JP2019152816A (ja) * 2018-03-06 2019-09-12 株式会社日立情報通信エンジニアリング 通話音声処理システム及び通話音声処理方法
CN110704618A (zh) * 2019-09-20 2020-01-17 阿里巴巴集团控股有限公司 确定对话数据对应的标准问题的方法及装置
CN111460212A (zh) * 2020-03-03 2020-07-28 平安科技（深圳）有限公司 语音流程设置方法、装置、计算机设备和存储介质

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114374769A (zh) * 2021-12-01 2022-04-19 恒安嘉新（北京）科技股份公司 一种异常号码的获取方法、装置、服务器和存储介质
CN114710471A (zh) * 2022-03-21 2022-07-05 京东科技信息技术有限公司 基于网络的客服语音通信方法、装置、电子设备及介质

Also Published As

Publication number Publication date
CN111460212A (zh) 2020-07-28

Similar Documents

Publication Publication Date Title
WO2021174840A1 (zh) 语音流程设置方法、装置、计算机设备和存储介质
CN110990545B (zh) 一种人工智能电话拓客营销管理系统、方法
AU2017404110B2 (en) Agent allocation method and device, server, and storage medium
US20060215824A1 (en) System and method for handling a voice prompted conversation
WO2011017961A1 (zh) 一种排队路由的方法及装置
US20080225872A1 (en) Dynamically defining queues and agents in a contact center
KR101975320B1 (ko) 빅데이터 및 인공 지능 기반 콜 센터 공유 서비스 제공 방법 및 시스템
CN111246025B (zh) 人机混排的客服方法及系统
US11451939B2 (en) Routing of toll-free numbers and text messages using a toll-free exchange
CN105898086A (zh) 一种预测外呼的方法
CN111586019A (zh) 身份认证方法、装置及服务设备
CN108076237A (zh) 一种电话客服数据处理方法及装置
EP1563674B1 (en) Predective dialling by monitoring progress of agent script
WO2012126213A1 (zh) 来电原因的判定方法及装置
CN111916071A (zh) 通话呼叫方法、装置、计算机设备和存储介质
WO2020125529A1 (zh) 互动式语音应答菜单的应用方法、系统、装置及存储介质
WO2019019778A1 (zh) 通话数据处理方法、装置、存储介质和计算机设备
KR20190100668A (ko) 고객 상담 서비스 시스템 및 방법
US20130208883A1 (en) Outbound effectiveness through management of idle agent pool
WO2016169364A1 (zh) 自助语音业务的自动处理方法、装置及移动终端
CN113726961A (zh) 外呼数量的确定方法、装置、外呼系统及存储介质
US20130084839A1 (en) System and method for delivering caller name information to mobile devices
CN116489275A (zh) 一种基于语音识别的ai智能应答系统及方法
CN102238261A (zh) 一种控制通话接收设备接听通话请求的控制方法及装置
CN114205467A (zh) 一种电网客服工单智能外呼系统及使用方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20923454

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20923454

Country of ref document: EP

Kind code of ref document: A1