TW201818734A - Method and device for processing incoming call, and terminal - Google Patents

Abstract

Disclosed in the present application are a method and device for processing an incoming call, and a terminal. The method comprises: after receiving an incoming call from a user, obtaining at least one piece of feature information of the user according to the incoming call, wherein the feature information is information, associated with a menu item of an IVR voice menu, in historical information of the user; according to a pre-set probability prediction model, calculating a probability prediction value of the menu item associated with the obtained feature information; and according to the calculated probability prediction value, determining whether the incoming call is enabled to jump to a service link indicated by a corresponding menu item by skipping over menus at corresponding levels in the IVR voice menu. By implementing the present application, response operations on an IVR voice menu by a user can be effectively reduced and an incoming call from the user is quickly enabled to jump to a service link indicated by a corresponding menu item so as to reduce the customer effort score, and the operating cost of a call centre.

Description

Call processing method, device and terminal

The present invention relates to the field of computer technology, and in particular, to a method, a device, and a terminal for processing incoming calls.

The IVR (InteractiveVoiceResponse, interactive voice response) system is an important part of the voice value-added service system. Many enterprise call centers provide user services through the IVR system. The IVR voice function table in the IVR system serves as the call center portal. Generally, The IVR voice function table includes at least one level of function table. Each level of function table includes at least one function table item. By playing the function table item of each level function table, the calling customer can be prompted to respond to the function table item. Then, according to each response operation of the user, determine which menu item the user selects, and then jump to the corresponding link to implement navigation for the calling customer. For enterprises with call centers, with the growth and refinement of business, the functional requirements of business departments for IVR systems have increased dramatically. Correspondingly, in order to meet the needs of business departments, the calling customer is navigated to the corresponding business end. IVR voice functions The structure of the table is also getting larger and more complex. After the customer calls, the response operation is prompted by the IVR voice menu, and the complexity of the IVR voice menu increases the amount and duration of the customer's response to the IVR voice menu for the response operation, causing the customer to stay in the IVR system If the time is too long, it will increase the Customer Effort Score (CES) of the calling customer and the operation cost of the call center.

In view of this, the present invention provides a method, a device and a terminal for processing incoming calls, so as to solve the problems that the existing IVR system will increase the customer effort of the incoming callers and the operating cost of the call center. Specifically, the present invention is implemented by the following technical solutions: According to a first aspect of an embodiment of the present invention, an incoming call processing method is provided, including the following steps: 后 After receiving an incoming call from a user, obtain the user according to the incoming call At least one feature information, the feature information is information associated with the menu item of the IVR voice menu in the user's historical information; 计算 calculating the function table associated with the acquired feature information according to a preset probability prediction model Probability prediction value of the item; According to the calculated probability prediction value, determine whether to skip the function table of the corresponding level in the IVR voice function table and jump the incoming call to the business link indicated by the corresponding function table item. In one embodiment, the function table of the corresponding level is the first-level function table of the IVR voice function table, and the feature information includes at least one of the following information: The user's response to the function table item of the IVR voice function table Response operation information; the event category to which the event consulted by the user's incoming call belongs; historical information of the corresponding application in the terminal which initiated the incoming call. In one embodiment, after receiving a user's incoming call, the method further includes the following steps: judging whether the user's incoming call is received for the first time; if the user's incoming call is not received for the first time, obtaining according to the incoming call At least one characteristic information of the user; if it is the first time that the user's incoming call is received, push the IVR voice menu to the user; according to the user's response operation to the menu item of the IVR voice menu, The incoming call jumps to: the business link indicated by the menu item of the operation that the user responds to; records the user's response operation to the menu item of the IVR voice menu. In one embodiment, according to the calculated probability prediction value, determining whether to skip a function table of a corresponding level in the IVR voice function table, and redirecting the incoming call to a service link indicated by a corresponding function table item, Including: comparing the magnitude relationship between the calculated probability prediction value and the preset probability threshold; if the obtained probability prediction value is greater than the preset probability threshold value, skip the corresponding level function table in the IVR voice function table, and call the incoming call Jump to: the business link indicated by the function table item associated with the obtained feature information; if the obtained probability prediction value is not greater than a preset probability threshold, push the IVR voice function table to the user; The response operation of the menu item of the IVR voice menu jumps the incoming call to: the business link indicated by the menu item of the operation that the user responds to; records the user's response to the menu item of the IVR voice menu Response action. In one embodiment, the generating step of the preset probability prediction model includes: 选取 selecting historical information of some users as training samples from historical information of all users; from each training user included in the training sample The historical information of each, to obtain the amount of feature information associated with each menu item, and the operating probability of each training user to respond to each menu item, constituting each training user corresponding to each menu item Information quantity and operation probability; 以 Taking the information quantity and operation probability as independent variables and dependent variables of the preset function, and training each user based on the information quantity and operation probability of each menu item, The model function corresponding to the function. In one embodiment, the generating step of the preset probability prediction model further includes: 选取 selecting historical information of some users except verification samples from historical information of all users as a verification sample; from the verification In the historical information of each verified user included in the sample, the amount of information about the feature information associated with each menu item and the operation probability of each verified user's response operation to each menu item constitute each verification The amount of information and operation probability of the user corresponding to each menu item; The amount of information of each verification user corresponding to each menu item is input into the trained model function, and the operation probability of each verification user corresponding to each menu is calculated; By comparing the calculated operation probability and the obtained operation probability, the accuracy rate of the operation probability of each menu item corresponding to each verified user is obtained; If the accuracy rate is higher than the preset accuracy threshold, the number of users is higher than the preset coverage Threshold value, it is determined that the model function is the preset probability prediction Probabilistic prediction function. In one embodiment, the calculating a probability prediction value of a function table item corresponding to the acquired feature information according to a preset probability prediction model includes: 计算 calculating a function associated with each function table item based on the obtained feature information. The amount of information of the feature information; Enter the amount of information of the feature information associated with each menu item into the probability prediction function, and calculate the probability prediction value of the menu item. In one embodiment, the generating step of the preset probability prediction model further includes: 后 after a preset period of time, selecting historical information of all users different from the training sample, and updating historical information of some users.样本 training samples; from the historical information of each training user included in the updated training sample, obtain the amount of information about the feature information associated with each menu item, and each training user responds to each menu item The operation probability of each user constitutes the amount of information and operation probability of each menu item corresponding to each training user; 以 The amount of information and operation probability are the independent and dependent variables of the preset function, and each training user corresponds to each The amount of information and operation probability of each function table item are used to retrain a model function corresponding to the preset function. According to a second aspect of the embodiments of the present invention, an incoming call processing device is provided, including: a feature acquisition module, configured to obtain at least one feature information of a user according to the incoming call after receiving an incoming call from the user; The feature information is the information related to the menu item of the IVR voice menu in the historical information of the user; Probability prediction module is used to calculate the function table associated with the obtained feature information according to a preset probability prediction model. Probability prediction value of the item; Caller jump module, which is used to determine whether to skip the corresponding level of the function table in the IVR voice function table and jump the incoming call to the corresponding function table item according to the calculated probability prediction value. Indicated business links. In one embodiment, the function table of the corresponding level is a first-level function table of the IVR voice function table, and the feature information includes at least one of the following information: The user's menu item of the IVR voice function table操作 response operation information of the user; the event category to which the event consulted by the user's incoming call belongs; the historical information of the corresponding application in the terminal which initiated the incoming call. In one embodiment, the feature acquisition module includes: a call judging module, which is used to determine whether the user's incoming call is received for the first time after receiving an incoming call from the user; When the user's incoming call is received for the first time, at least one characteristic information of the user is acquired according to the incoming call; the device further includes: a first function table module, which is used for receiving the user's incoming call for the first time, Push the IVR voice menu to the user; a first jump module for redirecting the incoming call to: the operation responded by the user according to the user's response to the menu item of the IVR voice menu The business link indicated by the function table item of ；; a first recording module for recording the user's response operation to the function table item of the IVR voice function table. In one embodiment, the call forwarding module includes: a prediction value comparison module for comparing the magnitude relationship between the calculated probability prediction value and a preset probability threshold; a function table skip module for The probability prediction value is greater than a preset probability threshold, skips the corresponding level function table in the IVR voice function table, and jumps the incoming call to: the business link indicated by the function table item associated with the obtained feature information; The device further includes: a second function table module for pushing the IVR voice function table to the user when the obtained probability prediction value is not greater than a preset probability threshold; a second jump module for The response operation to the menu item of the IVR voice function table redirects the incoming call to: the business link indicated by the menu item of the operation that the user responded to; a second recording module for recording the user's A response operation of a menu item of the IVR voice menu. In one embodiment, the device further includes a model generation module, the model generation module includes: training sample selection module, for selecting historical information of some users as training samples from historical information of all users Training parameter acquisition module, for obtaining the information quantity of feature information associated with each menu item from the historical information of each training user contained in the training sample, and each training user for each function The operation probabilities of the response operations of the table items constitute the information quantity and operation probability of each training user corresponding to each menu item; model training module, which is used to set the independent variables and factors of the preset function with the information quantity and operation probability, respectively. Variables, and based on the amount of information and operation probability of each training user corresponding to each menu item, a model function corresponding to the preset function is trained. In one embodiment, the model generation further includes: a verification sample selection module, which is used to select historical information of some users as verification samples from the historical information of all users except the training sample; 获取 acquisition of sample parameters A module for obtaining the information quantity of feature information associated with each menu item from the historical information of each verified user included in the verification sample, and each verified user responding to each menu item The operation probability of operation constitutes the amount of information and operation probability of each verification user corresponding to each menu item; Model function verification module, which is used to input the amount of information corresponding to each menu item for each verification user into the trained model Function to calculate the operation probability of each function table for each authenticated user; Accuracy acquisition module, which is used to obtain the function table items of each authentication user by comparing the calculated operation probability and the obtained operation probability. Accuracy rate of operation probability; model function determination module for Exact number higher than a preset threshold value of the user, covering higher than a preset threshold value, determining the probability of said model function is a predetermined function of the predicted probability model prediction. In one embodiment, the probability prediction module includes: 计算 calculating the amount of information of the feature information associated with each menu item based on the obtained feature information; 数量 the amount of information of the feature information to be associated with each menu item Enter the probability prediction function to calculate the probability prediction value of the menu item. In one embodiment, the model generating module further includes: a training sample updating module for selecting, after a preset period of time, historical information of some users different from the training sample from historical information of all users As updated training samples; Update parameter acquisition module, used to obtain the amount of feature information associated with each menu item from the historical information of each trained user included in the updated training samples, and each The operating probability of training users to respond to each menu item constitutes the amount of information and operating probability of each training user corresponding to each menu item; model function update module, which is used for the amount of information and operating probability respectively The independent and dependent variables of the preset function are described, and a model function corresponding to the preset function is retrained based on the amount of information and operation probability of each training user corresponding to each menu item. According to a third aspect of the embodiments of the present invention, a terminal is provided, including: a processor; a memory for storing executable instructions of the processor; a processor configured to: receive a user's incoming call Then, at least one characteristic information of the user is obtained according to the incoming call, and the characteristic information is information related to the menu item of the IVR voice menu in the historical information of the user; 计算 calculated according to a preset probability prediction model The probability prediction value of the function table item associated with the obtained feature information; determining whether to skip the function table of the corresponding level in the IVR voice function table and jump the incoming call to the corresponding function according to the calculated probability prediction value; The business link indicated by the table item. Applying the embodiment of the present invention, after receiving a user's incoming call, at least one characteristic information of the user is acquired according to the incoming call; and then, a function table item associated with the acquired characteristic information is calculated according to a preset probability prediction model. Probability prediction value; finally, according to the calculated probability prediction value, it is determined whether to skip a function table of a corresponding level in the IVR voice function table and jump the incoming call to a service link indicated by a corresponding function table item. Therefore, based on the historical information of the caller, it is possible to predict the menu item that the caller will choose, and then jump to the business link indicated by the predicted menu item, which can effectively reduce the caller ’s response to the IVR voice menu. Quickly redirect the incoming call from the incoming caller to the business link indicated by the corresponding menu item to reduce customer effort and call center operating costs. It should be understood that the above general description and the following detailed description are merely exemplary and explanatory, and should not limit the present invention.

Exemplary embodiments will be described in detail here, Examples are shown in the drawings. When the following description relates to the drawings, Unless otherwise stated, The same digits in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. in contrast, They are only related to, Examples of devices and methods consistent with the present invention.  的 The terminology used in the present invention is for the purpose of describing particular embodiments only, It is not intended to limit the invention. The singular "a", "an", "The" and "the" are intended to include the plural forms as well, Unless the context clearly indicates otherwise. It should also be understood that The term "and / or" as used herein refers to and includes any or all possible combinations of one or more related listed projects.  I should understand that Although terminology may be used in the present invention, second, The third level is to describe various information, But this information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. E.g, Without departing from the scope of the present invention, The first information can also be called the second information. Similarly, The second information may also be referred to as the first information. Depending on the context, As used herein, the word "if" can be interpreted as "at" or "at ..." or "in response to a determination."  See Figure 1. Is an intention of a system structure for realizing incoming call processing in the embodiment of the present invention, The system may include at least one of the first incoming call terminal 111 and the second incoming call terminal 112, Call center system 120, And the first service terminal 131, The second service terminal 132 to the Nth service terminal 13N, N is an integer greater than or equal to two.  First call terminal 111, User A logs in to device of client A based on account A, Using Internet calling (Internet Protocol-based voice calling) This embodiment only uses a mobile phone as an example for description. In practice, the first incoming call terminal 111 may also be a tablet computer (Pad,  portable android device), Personal computer (PC,  Personal Computer). User A of this terminal is based on the call function of user A, Through the internet, Initiate an incoming call with account A to the call center system 120, The client A mentioned here can include contacts, Clients for social communication software such as DingDao.  Second call terminal 112, Can be a wired telephone terminal, Wireless telephone terminal, PHS, Call terminals such as smartphones, Use hand-held phone call. User B of the terminal is based on the call function of the terminal, Through the phone network provided by the mobile operator, Call the hotline (such as 95188, etc.), Initiate an incoming call to the call center system 120 carrying the phone number of the second incoming terminal 112, The phone number can be a common landline number, Phone number provided by the mobile operator, Mobile operator's short number, The short number provided by the trunking network service or the virtual temporary number provided by the virtual operator.  Call center system 120, Can be a server, Or a server cluster of multiple servers, Or a cloud computing service center based on cloud computing, Or build a virtual call center using intelligent network technology, Ability to store user's historical information, IVR voice menu, And the association relationship between the feature information in the history information and the menu item of the IVR voice menu. It is mentioned here that the historical information may include the response operation of the menu items of the IVR voice menu (or the selected business link, Selecting this business link indicates that the menu item indicating the business link has been responded); It can include client records installed in the terminal that originated the incoming call: User's historical operation information, E.g: User-inquired business information, Service requests sent by users; It may also include what is recorded by the server corresponding to the client: Historical operation information associated with the user, E.g: Business information pushed to that user, The service request received by the user.  所述 The IVR voice function table includes at least one level of function table, Each level of the menu includes at least one menu item, By playing the menu item of each level menu, Can prompt calling customers to respond to menu items, Then based on each response from the user, Determine which menu item the user chooses, Then jump to the corresponding business link, This business link may be a business service provided by any one of the first business terminal 131 to the Nth business terminal 13N corresponding to the function table item selected by the user. It can also be the menu service of the next-level menu corresponding to the menu item selected by the user. The specific function table hierarchy and function table items are determined by the type of hotline associated with the call center system 120 or the type of industry served.  In some scenarios, The call center system 120 is associated with the service hotline of Ant Financial, The stored IVR voice function table is shown in Figure 1. The first level of function table includes Alipay business, please press 1. Online business please press 2. For flower business, please press 3 and indicate the menu items of other business links. The second to M-level function tables are related to Alipay business, E-commerce business, Flower business and other business associations, M is a positive integer greater than or equal to 2, Can be set according to specific business scope, I won't repeat them here.  关联 The related relationship may include at least one of the following relationships: The correspondence between the user's response to a menu item and the menu item; Correspondence between the business link to which the incoming call jumps and the menu item indicating the business link; Business links corresponding to historical operation information in the client, Correspondence with function table items indicating the business link; The business link corresponding to the historical operation information recorded on the server. Correspondence to the menu items indicating the business link.   Practical applications, After the call center 120 receives an incoming call from the first incoming terminal 111 or the second incoming terminal 112, Obtaining at least one characteristic information of the user according to the incoming call; According to a preset probability prediction model, Calculating a probability prediction value of a function table item associated with the obtained feature information (the probability prediction value may be used to characterize a probability that a user responds to the function table item); Finally, according to the calculated probability prediction value, Determining whether to skip a function table at a corresponding level in the IVR voice function table, The incoming call is redirected to a service link indicated by a corresponding menu item. E.g, The characteristic information is the user's response to the menu item "Alipay business please press 1" in Figure 1 (triggering of button 1), The predicted value of the function table item "Alipay business please press 1" is 95%, Indicates that the probability of a user responding to the menu item "Alipay Business Please Press 1" is 95%, Determining to skip the first-level function table in the IVR voice function table, The incoming call is redirected to the service link indicated by the corresponding menu item "Alipay service please press 1."  First business terminal 131, The second service terminal 132 to the Nth service terminal 13N, After the user's incoming call jumps to the business link indicated by the corresponding menu item, Provide users with services corresponding to the business links indicated by the menu items, Services can be provided by Salesperson A, The operations from salesman B to salesman N are completed. E.g: The menu item is Alipay. Press 1. Online business please press 2. For flower business, please press 3 and indicate the menu items of other business links. First service terminal 131, The second service terminal 132 to the Nth service terminal 13N can provide the following services to the caller:  Receive Alipay user complaints, Explain that Alipay users query transaction information, Explain that online business users query deposit or loan information, Explain that Huayan users query credit lines and so on.  The embodiment of the present invention will be described in detail below with reference to FIG. 1.  See Figure 2. FIG. 2 is a flowchart of an embodiment of an incoming call processing method according to the present invention. The incoming call processing method may include the following steps 201-203:  201Step 201: After receiving an incoming call from a user, Obtaining at least one characteristic information of the user according to the incoming call, The characteristic information is historical information of the user. Information associated with menu items of the IVR voice menu.  的 Incoming call method according to the embodiment of the present invention, Can be applied to the call center system shown in Figure 1, After receiving an incoming call, Can use the user information of the user carried by the incoming call, Query the stored characteristic information corresponding to the user information; The user information of the user carried by the incoming call may also be used, Query the characteristic information of the user from the historical information of the user. The user information mentioned here may include, but is not limited to: user name, One or more phone numbers (when a smart terminal can provide multiple SIM card functions, Or when providing a virtual SIM card function, User information can include multiple phone numbers), fax number, Social communication accounts (e.g. Dingding account, Current account number, etc.) and so on. The call center system can record user information for each user, Feature information corresponding to each user information, Historical information for each user, And the relationship between the characteristic information of each user and each menu item.  To determine user characteristics, Provided that the user is a regular user, Called to the current call center before this call, For users who call the current call center for the first time, Unable to determine its characteristic information. In order to distinguish incoming calls from new and existing users, Upon receiving an incoming call, The incoming call processing method in this embodiment may include the following operations: Determine whether the user's incoming call is received for the first time; If this is not the first time an incoming call has been received for that user, Obtaining at least one characteristic information of the user according to the incoming call; If this is the first time an incoming call has been received for that user, Push the IVR voice function table to the user; According to the user's response operation to the menu item of the IVR voice menu, Redirect the incoming call to: The business link indicated by the menu item that the user responded to; Record the user's response operation to the menu item of the IVR voice menu. The response operations mentioned here can include the user's touch on the corresponding key, And user-selected menu selection voice and so on.  实施 The embodiment of the present invention, By distinguishing incoming calls from new and existing users, You can call for incoming calls of new and old users, Quickly and immediately adapt to the call processing of different users, On the premise of reducing customer effort and call center operating costs, Meet the needs of different users.  For old users, Its characteristic information can be recorded by recording its historical call information, Query historical information in the terminal (such as the first incoming terminal 111 or the second incoming terminal 112 in FIG. 1) that originated its incoming call, Or query the historical information acquisition in the terminal receiving the incoming call (such as the terminal corresponding to the call center system 120 in FIG. 1) For example, at least one of the following: User response operation information to menu items of IVR voice menu; The event category to which the user consults for the event consulted; Historical information of the corresponding application in the terminal that initiated the incoming call.   among them, The user's response to the menu items of the IVR voice menu can include the user's touch of the corresponding key, And user-selected menu selection voice and so on.  The event category to which the event consulted by the user's incoming call belongs, It can be determined according to the preset event classification structure. E.g: For the IVR voice menu shown in Figure 1, For example, to classify the questions asked by users of Ant Financial's incoming calls, Classify the questions that are consulted into unpaid questions, Password issues, etc. Questions about payment failures in users' consultations in the Huali business belong to the category of payment problems. Questions about user locked Alipay payment passwords belong to the category of password problems.  的 historical information of the corresponding application in the terminal that initiated the incoming call, It can include client records installed in the terminal that originated the incoming call: User's historical operation information, Such as business information that users query, Service requests sent by users, etc. The corresponding application may be an application associated with the hotline of the call.  Step 202: According to a preset probability prediction model, Calculate the probability prediction value of the menu item associated with the obtained feature information.  中 In the embodiment of the present invention, The preset probability prediction model is used to predict: When the user calls this time, Probability of selecting a menu item associated with the acquired feature information. corresponding, The probability prediction value may be used to represent a probability that a user responds to the menu item. If the obtained feature information is an item, Only the probability of the menu item associated with that feature information is predicted, If the obtained feature information is at least two, Then the probability of the menu item associated with each feature information is predicted separately. Individual feature information can be associated with only one menu item. You can also associate a set of menu items, This set of menu items consists of the various levels of the IVR voice menu, Used to indicate a certain business link, The navigation path for the business link.   Practical applications, In order to predict the probability of a menu item associated with the feature information selected by the calling user, The function table items of the IVR voice function table may be classified in advance, Divided into at least two types of menus, The information quantity of the feature information associated with each type of menu item is then an independent variable. Taking the probability of each type of menu item as the dependent variable, Calculated by the probability calculation formula: Probability of the menu item associated with each feature information obtained. The variable parameters of each dependent variable can be set according to the actual application. It can also be obtained by training the user's historical information stored in the call center system.  数目 the number of classifications corresponding to the function table item classification of the IVR voice function table, The probability prediction model may be a two-class probability prediction model or a multi-class probability prediction model. The function table can be classified according to actual needs. Function tables that indicate different levels of the same business segment can be grouped into the same category, Make the entire IVR voice menu into at least two categories, Or just categorize menus at the same level, Divide the same level of function tables into at least two categories. Take the IVR menu shown in Figure 1 as an example. The menu item “Alipay business please press 1” in the first level menu can be divided into one category, The remaining menu items in the first level menu are divided into another category; corresponding, The probability of the menu item "Alipay business please press 1" is a dependent variable. The total probability of all remaining menu items is one dependent variable, The information quantity of the feature information associated with the menu item "Alipay Business Please Press 1" is an independent variable. The sum of the information quantity of the characteristic information associated with all the remaining menu items is another independent variable, Construct a binary classification probability prediction model.  In one example, In order to determine the specific value of the variable parameter of the dependent variable, The preset probability prediction model can be determined and generated by the following operations:  From the historical information of all users, Select historical information of some users as training samples.  From the historical information of each trained user contained in the training sample, Gets the amount of feature information associated with each menu item. And the operation probability of each trained user responding to each menu item, The amount of information and operation probability of each training user corresponding to each menu item.  以 Take the amount of information and the probability of operation as the independent and dependent variables of the preset function, respectively. Based on the amount of information and operation probability of each training user corresponding to each menu item, A model function corresponding to the preset function is trained.   among them, If the historical information of the user only includes the response operation to the menu item of the IVR voice menu when the user calls. When obtaining the information quantity of feature information associated with a menu item, From the historical information of each trained user, Obtain the total number of operations that train users to respond to a menu item as the amount of information. When obtaining the operation probability of each training user to respond to a menu item: From the historical information of each trained user, First get the total number of calls for all incoming calls, Then obtain the total number of operations for training users to respond to a menu item, The ratio of the total number of operations to the total number of calls is finally obtained as the operation probability.  When training a model function corresponding to the preset function, Can be obtained by data fitting. The preset function can be based on the amount of information and operation probability of each menu item corresponding to each trained user, The line shape of the distribution map in the coordinate space is determined, E.g: The line shape of the distribution chart is a straight line, The preset function is a linear function containing at least two independent variables.  This example only uses the historical information of some users as training samples. Data sources that can be used for validation and continuous training of preset probability models can be reserved, While ensuring the accuracy of the current model, After model effects decay over time, Update models in time. In some scenarios, The training sample can include historical information for 90% of users.  In another example, In order to verify the accuracy of the trained model, You can verify the model by:  选取 Select from the historical information of all users except the training samples, The historical information of some users is used as a verification sample.  From the historical information of each verified user contained in the verification sample, Gets the amount of feature information associated with each menu item. And the probability of each verification user responding to each menu item, The amount of information and operation probability corresponding to each menu item for each authenticated user.  Enter the amount of information of each verified user corresponding to each menu item into the trained model function, Calculate the operation probability of each authenticated user corresponding to each function table.  By comparing the calculated operation probability with the obtained operation probability, The accuracy rate of the operation probability corresponding to each menu item of each verified user is obtained.  If the accuracy rate is higher than the number of users with a preset accuracy threshold, Above a preset coverage threshold, Then, it is determined that the model function is a probability prediction function of the preset probability prediction model.   among them, If the historical information of the user only includes the response operation to the menu item of the IVR voice menu when the user calls. When obtaining the information quantity of feature information associated with a menu item, From the history of each authenticated user, Obtain the total number of operations for verifying the user's response to a menu item as the information quantity. When obtaining the operation probability of each authenticated user responding to a menu item: From the history of each authenticated user, First get the total number of calls for all incoming calls, And then obtain the total number of operations for verifying that the user responds to a menu item, The ratio of the total number of operations to the total number of calls is finally obtained as the operation probability.  In this example, When comparing the calculated operation probability with the obtained operation probability, You can compare the difference or ratio of the two, The difference is 0 or the ratio is 1, Is accurate, Otherwise it is inaccurate. Accuracy per user, It can be the ratio of the number of calls with the calculated operation probability to the total number of calls. The preset accurate threshold can be set according to actual needs. In some scenarios, The verification sample can include historical information for 10% of users, The preset accuracy threshold can be 95%, The preset coverage threshold may be 18%.  In another example, To ensure the accuracy of the probability model, You can instantly update the preset probability model by:  After a preset period, Selecting from the historical information of all users different from the training sample, The historical information of some users is used as the updated training sample.  From the historical information of each trained user contained in the updated training sample, Gets the amount of feature information associated with each menu item. And the operation probability of each trained user responding to each menu item, The amount of information and operation probability of each training user corresponding to each menu item.  以 Taking the amount of information and the operation probability as the independent variable and the dependent variable of the preset function, respectively, Based on the amount of information and operation probability of each training user corresponding to each menu item, A model function corresponding to the preset function is retrained.  In this example, The historical information of some users can be selected from the verification samples as updated training samples; Or select the historical information of some users from the verification sample to add to the original training sample. Compose updated training samples.  Corresponding to the training process of the above model, In an optional implementation, The following operations can be used to predict the model based on a preset probability, Calculate the probability prediction value of the menu item corresponding to the obtained feature information:  Based on the acquired feature information, Calculates the amount of information about the feature information associated with each menu item.  Enter the information quantity of feature information associated with each menu item into the probability prediction function, Calculate the predicted probability value for this menu item.  This implementation, It can quickly and accurately calculate the probability prediction value of the menu item associated with each feature information obtained.  Step 203: Based on the calculated probability predictions, Determining whether to skip a function table at a corresponding level in the IVR voice function table, The incoming call is redirected to a service link indicated by a corresponding menu item.  中 In the embodiment of the present invention, You can compare the calculated probability prediction value with a preset probability threshold. Determining whether to skip a function table at a corresponding level in the IVR voice function table, The incoming call is redirected to a service link indicated by a corresponding menu item. The business link may include a problem identification link, Pipeline identification, Consulting and answering sessions, etc.  In some scenarios, The characteristic information may be historical information of the user, Information associated with a menu item in the first-level menu in the IVR voice menu, The function table of the corresponding level may be the first-level function table.  In one example, The predicted value according to the calculated probability, Determining whether to skip a function table at a corresponding level in the IVR voice function table, Redirecting the incoming call to the business link indicated by the corresponding menu item, include:  Compare the relationship between the calculated probability prediction value and the preset probability threshold.  If the obtained probability prediction value is greater than a preset probability threshold, Skip the function table of the corresponding level in the IVR voice function table, Redirect the incoming call to: The business link indicated by the menu item associated with the obtained feature information.  If the obtained probability prediction value is not greater than a preset probability threshold, Push the IVR voice function table to the user.  According to the user's response operation to the menu item of the IVR voice menu, Redirect the incoming call to: The business link indicated by the menu item that the user responded to.  Record the user's response to the menu item of the IVR voice menu.  In this example, By recording a user's response operation to a menu item of the IVR voice menu, Get historical information for first time callers. So that when the user calls again Based on its historical information, Predict the menu item it chooses. The preset probability threshold may be set according to actual needs, For example, the preset probability threshold is set to 95%.  When skipping the menu of the corresponding level in the IVR voice menu, Determine the level at which a function table item whose calculated predicted probability value is greater than the probability threshold is skipped, And the hierarchy menu that played before the menu item, Then, the incoming call is jumped to the business link indicated by the menu item. If the calculated probability prediction value is greater than a menu item with a preset probability threshold, Is the last level menu in the IVR voice menu project, Skip the entire IVR voice menu.  In other examples, The predicted values can be calculated by comparison. Determining whether to skip a function table at a corresponding level in the IVR voice function table, The incoming call is redirected to a service link indicated by a corresponding menu item. E.g: Skip the hierarchy of the menu item with the highest forecast value, And a hierarchy of menus that play before the menu item. Then, the incoming call is jumped to the business link indicated by the menu item.  其他 In other embodiments of the invention, When a user calls, Can record the number of user incoming calls and the user's response to menu items each time they call the profile, Then set the preset number of times when the user calls, After 5 times, Before you can call based on the user ’s incoming call, Get the characteristic information of the user, According to a preset probability prediction model, Calculate the probability prediction value of the menu item associated with the obtained feature information.  According to the above embodiment, After receiving an incoming call from a user, the present invention Obtaining at least one characteristic information of the user according to the incoming call; According to a preset probability prediction model, Calculate the probability prediction value of the menu item associated with the obtained feature information; Finally, according to the calculated probability prediction value, Determining whether to skip a function table at a corresponding level in the IVR voice function table, The incoming call is redirected to a service link indicated by a corresponding menu item. therefore, Based on the historical information of the calling user, Predict the menu items that users will choose, Then jump to the business link indicated by the predicted menu item, Can effectively reduce the response operation of the caller to the IVR voice menu, Quickly jump the incoming call of the incoming user to the business link indicated by the corresponding menu item, To reduce customer effort and call center operating costs.  的 The incoming call processing method of the embodiment of the present invention can be applied to e-commerce, bank, express delivery, aviation, In the hotline services of various industries such as telecommunications, Based on historical information from callers in these industries, Predict the menu items that users will choose, Then jump to the business link indicated by the predicted menu item, Can effectively reduce the response operations of callers in these industries to the IVR voice menu, Quickly jump the incoming call of the incoming user to the business link indicated by the corresponding menu item, In order to reduce customer effort and call center operating costs, Improve user experience for these industries.  Corresponds to the foregoing embodiment of the incoming call processing method, The invention also provides an embodiment of the incoming call processing device.  来电 The embodiment of the call processing device of the present invention can be applied to a terminal. Device embodiments can be implemented by software, It can also be implemented by hardware or a combination of software and hardware. Taking software implementation as an example, As a logical device, It is formed by reading the corresponding computer program instructions in the non-volatile memory into the memory through the processor of the terminal where it is located. At the hardware level, As shown in Figure 3, Is a hardware structure diagram of a terminal where a call processing device of the present invention is located, In addition to the processor 310 shown in FIG. 3, Network interface 320, Memory 330, And non-volatile memory 340, The terminal where the device is located in the embodiment is usually based on the actual function of the terminal. Can also include other hardware, I will not repeat them here.  The above processor 310 is configured as: After receiving an incoming call from a user, Obtaining at least one characteristic information of the user according to the incoming call, The characteristic information is historical information of the user. Information associated with menu items of the IVR voice menu; According to a preset probability prediction model, Calculate the probability prediction value of the menu item associated with the obtained feature information; Based on the calculated probability predictions, Determining whether to skip a function table at a corresponding level in the IVR voice function table, The incoming call is redirected to a service link indicated by a corresponding menu item.  See Figure 4, 4 is a block diagram of an embodiment of a call processing device according to the present invention. The device may include: Feature acquisition module 410, Probability prediction module 420 and call forwarding module 430.   among them, Feature acquisition module 410, Used to receive an incoming call from a user, Obtaining at least one characteristic information of the user according to the incoming call, The characteristic information is historical information of the user. Information associated with menu items of the IVR voice menu.  Probability prediction module 420, For predicting models based on preset probabilities, Calculate the probability prediction value of the menu item associated with the obtained feature information.  Call forwarding module 430, Used to predict values based on calculated probabilities, Determining whether to skip a function table at a corresponding level in the IVR voice function table, The incoming call is redirected to a service link indicated by a corresponding menu item.  In an optional implementation, The feature acquisition module 410 may include (not specifically shown in FIG. 4):  Call judgment module, Used to receive an incoming call from a user, Determines whether the user's incoming call is received for the first time.  Information determination module, Used when an incoming call from the user is not received for the first time, Acquiring at least one characteristic information of the user according to the incoming call.   further, The device for processing incoming calls in the embodiment of the present invention may further include (not specifically shown in FIG. 4):  First menu module, Used when the user's first incoming call is received, Push the IVR voice function table to the user.  First jump module, Configured to operate according to the user ’s response to a menu item of the IVR voice menu, Redirect the incoming call to: The business link indicated by the menu item that the user responded to.  First record module, It is used to record the user's response operation to the menu item of the IVR voice menu.  In another optional implementation, The call forwarding module 430 may include (not specifically shown in FIG. 4):  Predicted value comparison module, It is used to compare the magnitude relationship between the calculated probability prediction value and a preset probability threshold.  Menu skip module, Used to obtain the predicted probability value greater than a preset probability threshold, Skip the function table of the corresponding level in the IVR voice function table, Redirect the incoming call to: The business link indicated by the menu item associated with the obtained feature information.   further, The device for processing incoming calls in the embodiment of the present invention may further include (not specifically shown in FIG. 4):  Second menu module, When the obtained probability prediction value is not greater than a preset probability threshold, Push the IVR voice function table to the user.  Second Jump Module, Configured to operate according to the user ’s response to a menu item of the IVR voice menu, Redirect the incoming call to: The business link indicated by the menu item that the user responded to.  Second recording module, It is used to record the user's response operation to the menu item of the IVR voice menu.  In another optional implementation, The call processing device according to the embodiment of the present invention may further include (not specifically shown in FIG. 4) a model generation module, The model generation module may include:  Training sample selection module, Used from historical information for all users, Select historical information of some users as training samples.  Training parameter acquisition module, And is configured to extract historical information of each trained user included in the training sample, Gets the amount of feature information associated with each menu item. And the operation probability of each trained user responding to each menu item, The amount of information and operation probability of each training user corresponding to each menu item.  Model training module, Independent variables and dependent variables that use the amount of information and the probability of operation as preset functions, respectively. Based on the amount of information and operation probability of each training user corresponding to each menu item, A model function corresponding to the preset function is trained.  In another optional implementation, The model generation module may further include (not specifically shown in FIG. 4):  Verify the sample selection module, For selecting from the historical information of all users except the training samples, The historical information of some users is used as a verification sample.  Sample parameter acquisition module, Used to extract historical information from each verified user contained in the verification sample, Gets the amount of feature information associated with each menu item. And the probability of each verification user responding to each menu item, The amount of information and operation probability corresponding to each menu item for each authenticated user.  Model function verification module, The amount of information for each authenticated user corresponding to each menu item is input into the trained model function Calculate the operation probability of each authenticated user corresponding to each function table.  Accuracy acquisition module, Used to compare the calculated operation probability with the obtained operation probability, The accuracy rate of the operation probability corresponding to each menu item of each verified user is obtained.  Model function determination module, Used for the number of users whose accuracy is above a preset accuracy threshold, Above a preset coverage threshold, Then, it is determined that the model function is a probability prediction function of the preset probability prediction model.  In another optional implementation, The probability prediction module 420 may include (not specifically shown in FIG. 4):  Based on the acquired feature information, Calculates the amount of information about the feature information associated with each menu item.  Enter the information quantity of feature information associated with each menu item into the probability prediction function, Calculate the predicted probability value for this menu item.  In another optional implementation, The model generation module may further include (not specifically shown in FIG. 4):  Training sample update module, Used after a preset period of time, Selecting from the historical information of all users different from the training sample, The historical information of some users is used as the updated training sample.  Update parameter acquisition module, Used to retrieve historical information about each trained user contained in the updated training sample, Gets the amount of feature information associated with each menu item. And the operation probability of each trained user responding to each menu item, The amount of information and operation probability of each training user corresponding to each menu item.  Model function update module, Used to set the independent variable and the dependent variable of the preset function respectively with the amount of information and the operation probability, Based on the amount of information and operation probability of each training user corresponding to each menu item, A model function corresponding to the preset function is retrained.  的 For the implementation process of the functions and functions of each module in the above device, see the implementation process of the corresponding steps in the above method for details. I will not repeat them here.  For device embodiments, Since it basically corresponds to the method embodiment, Therefore, for related parts, refer to the description of the method embodiments. The device embodiments described above are only schematic, The modules described as separate components may or may not be physically separated, The component displayed as a module may or may not be a physical module. Can be located in one place, Or it can be distributed to multiple network modules. Some or all of the modules can be selected according to actual needs to achieve the objective of the solution of the present invention.  普通 Under the condition that ordinary technicians in the field do not pay progressive labor, That can be understood and implemented. Those skilled in the art, after considering the specification and practicing the invention disclosed herein, will Other embodiments of the invention will readily come to mind. The invention is intended to cover any variations of the invention, Use or adaptation, These variants, Uses or adaptive changes follow the general principles of the present invention and include common knowledge or common technical means in the technical field not disclosed in the present invention. The description and examples are to be regarded as merely exemplary, The true scope and spirit of the present invention is indicated by the scope of patent application.  I should understand that The invention is not limited to the precise structure already described above and shown in the drawings, And various modifications and changes can be made without departing from its scope. The scope of the present invention is limited only by the scope of the accompanying patent applications.

111‧‧‧The first incoming call terminal

112‧‧‧Second call terminal

120‧‧‧Call Center System

131‧‧‧The first business terminal

132‧‧‧Second business terminal

13N‧‧‧th Nth service terminal

201 ~ 203‧‧‧step

310‧‧‧ processor

320‧‧‧ web interface

330‧‧‧Memory

340‧‧‧Non-volatile memory

410‧‧‧Feature Acquisition Module

420‧‧‧Probability prediction module

430‧‧‧Call Jump Module

The drawings herein are incorporated in and constitute a part of this specification, show embodiments consistent with the present invention, and together with the description serve to explain the principles of the present invention. Fig. 1 is a schematic diagram of a system structure for implementing incoming call processing according to an embodiment of the present invention; Fig. 2 is a flowchart of an embodiment of an incoming call processing method of the present invention; Fig. 3 is a hardware structural diagram of a terminal where an incoming call processing device of the present invention is located; It is a block diagram of an embodiment of the incoming call processing device of the present invention.

Claims (17)

An incoming call processing method includes the following steps: After receiving an incoming call from a user, obtain at least one characteristic information of the user according to the incoming call, the characteristic information is a function of the historical information of the user and the function of the IVR voice menu Table item related information; Calculate the probability prediction value of the function table item associated with the acquired feature information according to a preset probability prediction model; Determine whether to skip the IVR voice function table according to the calculated probability prediction value The function table of the corresponding level jumps the incoming call to the service link indicated by the corresponding function table item. The method according to item 1 of the scope of patent application, wherein the function table of the corresponding level is a first-level function table of the IVR voice function table, and the characteristic information includes at least one of the following information: Response operation information of the menu items of the voice menu; an event category to which an event consulted by a user's incoming call belongs; 的 historical information of a corresponding application in a terminal initiating the incoming call. The method according to item 1 of the scope of patent application, wherein after receiving a user's incoming call, the method further includes the following steps: judging whether the user's incoming call is received for the first time; if the user's incoming call is not received for the first time Call, obtain at least one characteristic information of the user according to the incoming call; push the IVR voice menu to the user if it is the first time that the user's incoming call is received; according to the user's function of the IVR voice menu The response operation of the table item redirects the incoming call to: the business link indicated by the menu item of the operation that the user responds to; records the response operation of the user to the menu item of the IVR voice menu. The method according to item 1 of the scope of patent application, wherein, according to the calculated probability prediction value, determining whether to skip a function table of a corresponding level in the IVR voice function table and jump the incoming call to a corresponding function The business links indicated by the table items include: Compare the calculated relationship between the predicted probability value and the preset probability threshold; If the obtained predicted probability value is greater than the preset probability threshold, skip the corresponding level in the IVR voice function table The function table redirects the incoming call to: the business link indicated by the function table item associated with the obtained feature information; if the obtained probability prediction value is not greater than a preset probability threshold, push the IVR voice function to the user ； According to the user's response operation to the menu item of the IVR voice menu, the incoming call is redirected to: the business link indicated by the menu item of the operation to which the user responds; record the user's response to the IVR Response actions for menu items of the voice menu. The method according to any one of claims 1 to 4, wherein the generating step of the preset probability prediction model includes: 选取 selecting historical information of some users as training from historical information of all users Samples; 获取 obtaining the information quantity of feature information associated with each menu item from the historical information of each training user contained in the training sample, and the operation of each training user performing a response operation on each menu item Probability, which constitutes the amount of information and operation probability of each menu item corresponding to each trained user; 以 Take the amount of information and operation probability as the independent and dependent variables of the preset function, and based on each training user corresponding to each menu item The amount of information and operation probability are used to train a model function corresponding to the preset function. The method according to item 5 of the scope of patent application, wherein the generating step of the preset probability prediction model further includes: 选取 selecting historical information of all users except the training sample from historical information of all users to do Is a verification sample; 获取 obtain the information quantity of feature information associated with each menu item from the historical information of each verification user contained in the verification sample, and each verification user responds to each menu item Operation probability, which constitutes the amount of information and operation probability of each menu item for each authenticated user; Enter the amount of information of each menu item for each authenticated user into the trained model function, and calculate the correspondence for each authenticated user Operation probability of each function table; By comparing the calculated operation probability and the obtained operation probability, obtain the accuracy rate of the operation probability of each verified user corresponding to each function table item; If the accuracy rate is higher than a preset accuracy threshold If the number is higher than a preset coverage threshold, it is determined that the model function is The probability of default probability prediction model described prediction function. The method according to item 6 of the scope of patent application, wherein calculating the probability prediction value of the function table item corresponding to the obtained feature information according to a preset probability prediction model includes: 计算 calculating based on the obtained feature information The amount of feature information information associated with each menu item; Enter the amount of feature information information associated with each menu item into the probability prediction function to calculate a probability prediction value for the menu item. The method according to item 5 of the scope of patent application, wherein the generating step of the preset probability prediction model further comprises: 后 after a preset period of time, selecting from the historical information of all users different from the training sample, The historical information of some users is used as the updated training sample; From the historical information of each training user included in the updated training sample, obtain the amount of feature information associated with each menu item, and each training user's The operation probability of each menu item's response operation constitutes the amount of information and operation probability of each training user corresponding to each menu item; 以 the amount of information and operation probability are the independent and dependent variables of the preset function, Based on the amount of information and operation probability of each training user corresponding to each menu item, a model function corresponding to the preset function is retrained. An incoming call processing device includes: (1) a feature acquisition module for acquiring at least one characteristic information of the user according to the incoming call after receiving the incoming call of the user, wherein the characteristic information is historical information of the user, and Information related to the menu items of the IVR voice menu; Probability prediction module, which is used to calculate the probability prediction value of the menu item associated with the obtained feature information according to a preset probability prediction model; Call forwarding module, which uses Based on the calculated probability prediction value, it is determined whether to skip a function table of a corresponding level in the IVR voice function table and jump the incoming call to a service link indicated by a corresponding function table item. The device according to item 9 of the scope of patent application, wherein the function table of the corresponding level is a first-level function table of the IVR voice function table, and the characteristic information includes at least one of the following information: Response operation information of the menu items of the voice menu; an event category to which an event consulted by a user's incoming call belongs; 的 historical information of a corresponding application in a terminal initiating the incoming call. The device according to item 9 of the scope of patent application, wherein the feature acquisition module includes: a call judgment module, which is used to determine whether the user's incoming call is received for the first time after receiving the user's incoming call; information A determining module for obtaining at least one characteristic information of the user according to the incoming call when the incoming call of the user is not received for the first time; the device further includes: a first function table module for receiving When the user's incoming call is called, push the IVR voice menu to the user; a first jump module for calling the incoming call according to the user's response to the menu item of the IVR voice menu Jump to: the business link indicated by the menu item of the operation responded by the user; a first recording module for recording the user's response operation to the menu item of the IVR voice menu. The device according to item 9 of the scope of patent application, wherein the call forwarding module includes: a prediction value comparison module for comparing the magnitude relationship between the calculated probability prediction value and a preset probability threshold; a function table skip A module for skipping the function table of the corresponding level in the IVR voice function table when the obtained probability prediction value is greater than a preset probability threshold, and jumping the incoming call to: a function table associated with the acquired feature information The business link indicated by the project; The device further includes: a second function table module for pushing the IVR voice function table to the user when the obtained probability prediction value is not greater than a preset probability threshold; the second jump A module for redirecting the incoming call to: the business link indicated by the menu item of the operation that the user responds to according to the user's response operation to the menu item of the IVR voice menu; the second recording mode A group for recording a response operation of the user to a menu item of the IVR voice menu. The device according to any one of claims 9 to 12, in which the device further includes a model generation module, and the model generation module includes: a training sample selection module for In the historical information, select the historical information of some users as training samples; ； Training parameter acquisition module is used to obtain the features associated with each menu item from the historical information of each training user contained in the training sample The amount of information and the operation probability of each training user's response to each menu item constitute the amount of information and operation probability of each training user corresponding to each menu item; Model training modules, which are used to The information quantity and operation probability are independent variables and dependent variables of a preset function, and a model function corresponding to the preset function is trained based on the information quantity and operation probability of each training user corresponding to each menu item. The device according to item 13 of the scope of patent application, wherein the model generation further comprises: a verification sample selection module for selecting historical information of some users except the training sample from the historical information of all users As a verification sample; a sample parameter acquisition module for obtaining the information quantity of feature information associated with each menu item from the historical information of each verification user included in the verification sample, and each verification user The operation probability of responding to each menu item constitutes the amount of information and operation probability of each verification user corresponding to each menu item; A model function verification module is used to map each verification user to each menu item Enter the trained model function with the amount of information, and calculate the operation probability of each verification user corresponding to each function table; Accuracy acquisition module, which is used to obtain each verification by comparing the calculated operation probability and the obtained operation probability. The accuracy rate of the user's operation probability corresponding to each menu item; 确定 The model function is determined A module configured to determine that the model function is a probability prediction function of the preset probability prediction model if the number of users whose accuracy rate is higher than a preset accuracy threshold is higher than a preset coverage threshold. The device according to item 14 of the scope of patent application, wherein the probability prediction module includes: 计算 based on the obtained feature information, calculating the amount of feature information information associated with each menu item; The information quantity of the feature-related feature information is input into the probability prediction function, and the probability prediction value of the item of the function table is calculated. The device according to item 13 of the scope of patent application, wherein the model generation module further comprises: a training sample update module for selecting, after a preset period of time, different information from the training information of all users Samples, historical information of some users as updated training samples; Update parameter acquisition module, used to obtain the features associated with each menu item from the historical information of each training user included in the updated training sample The amount of information and the operation probability of each training user to respond to each menu item constitute the amount of information and operation probability of each training user corresponding to each menu item; Model function update module for separately Taking the amount of information and the operation probability as the independent variable and the dependent variable of the preset function, and retraining the model corresponding to the preset function based on the amount of information and the operation probability of each training user corresponding to each menu item function. A terminal comprising: a processor; a memory for storing instructions executable by the processor; a processor configured to: after receiving an incoming call from a user, obtain at least the user's at least according to the incoming call. A feature information, which is the information associated with the menu items of the IVR voice menu in the historical information of the user; 计算 calculating the function table items associated with the obtained feature information according to a preset probability prediction model. Probability prediction value; Determine whether to skip the function table of the corresponding level in the IVR voice function table and jump the incoming call to the business link indicated by the corresponding function table item according to the calculated probability prediction value.