Outbound calling method, device and electronic equipment

Abstract

The present application provides an outbound call method, device and electronic device, the outbound call method comprising: receiving an outbound call task issued by an upstream task system, the outbound call task including a target user number; obtaining user call behavior characteristics of the target user number, the user call behavior characteristics being obtained based on historical call data mining of the target user number; inputting the user call behavior characteristics into a pre-trained time period prediction model to estimate the dialing time period, and obtaining a target time period, the target time period indicating that the connection rate of calling the target user number during the target time period is the highest; initiating an outbound call to the target user number during the target time period. The embodiment of the present application can accurately estimate the target time period with the highest connection rate through the user call behavior characteristics, so initiating an outbound call to the target user number during the target time period can improve the connection rate of the outbound call.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to an outbound calling method, device and electronic device. Background Art

[0002] Outbound calls are calls that the outbound call system actively makes to users through technical means. Outbound calls can be used to complete certain tasks, thereby reducing costs and increasing task conversion rates. Outbound calls are widely used in a wide range of scenarios, such as marketing, return visits, questionnaires, and debt collection, etc. The connection rate is usually used to measure the quality of outbound calls. The connection rate refers to the ratio of the number of calls connected by users to the number of calls made by outbound calls. For an outbound call system, the higher the connection rate, the more users can be reached when the same number of calls are made, and the better the outbound call effect.

[0003] At present, after receiving an outbound call task, the outbound call system uses a random dialing method to make outbound calls. However, this method has the problem of low connection rate. Summary of the invention

[0004] Multiple aspects of the present application provide an outbound calling method, device and electronic device to solve the problem of low connection rate in the current outbound calling method.

[0005] In a first aspect, an embodiment of the present application provides an outbound call method, comprising: receiving an outbound call task issued by an upstream task system, the outbound call task including a target user number; obtaining user call behavior characteristics of the target user number, the user call behavior characteristics being obtained based on historical call data mining of the target user number, the user call behavior characteristics including: behavior characteristics of a user corresponding to the target user number in historical calls; inputting the user call behavior characteristics into a pre-trained time period prediction model to perform dialing time period prediction to obtain a target time period, the target time period indicating that the connection rate of calling the target user number during the target time period is the highest; and initiating an outbound call to the target user number within the target time period.

[0006] A second aspect of an embodiment of the present application provides an outbound calling device, including:

[0007] A receiving module, used to receive outbound call tasks sent by an upstream task system, where the outbound call tasks include target user numbers;

[0008] An acquisition module is used to acquire a user calling behavior feature of a target user number. The user calling behavior feature is obtained based on historical calling data mining of the target user number. The user calling behavior feature includes: a behavior feature of a user corresponding to the target user number in historical calls;

[0009] A processing module, used for inputting the user's calling behavior characteristics into a pre-trained time period prediction model to predict the dialing time period, and obtaining a target time period, wherein the target time period indicates that the call connection rate of the target user number is the highest during the target time period;

[0010] The outbound call module is used to initiate an outbound call to a target user number within a target time period.

[0011] A third aspect of an embodiment of the present application provides an electronic device, including: a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor implements the outbound call method of the first aspect when executing the computer program.

[0012] A fourth aspect of an embodiment of the present application provides a computer-readable storage medium, in which a computer program is stored. When the computer program runs on an electronic device, the electronic device executes the outbound call method of the first aspect.

[0013] The embodiment of the present application is applied in an outbound call scenario, and the outbound call method provided includes: receiving an outbound call task issued by an upstream task system, the outbound call task includes a target user number; obtaining user call behavior characteristics of the target user number, the user call behavior characteristics are obtained based on historical call data mining of the target user number; inputting the user call behavior characteristics into a pre-trained time period prediction model to estimate the dialing time period, and obtaining a target time period, the target time period indicates that the connection rate of calling the target user number during the target time period is the highest; initiating an outbound call to the target user number during the target time period. The embodiment of the present application can accurately estimate the target time period with the highest connection rate through the user call behavior characteristics, so initiating an outbound call to the target user number during the target time period can improve the connection rate of the outbound call. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0015] Figure 1 A flowchart of an outbound calling method provided for an exemplary embodiment of the present application;

[0016] Figure 2 A schematic diagram of an outbound calling method provided for an exemplary embodiment of the present application;

[0017] Figure 3 A flowchart of another outbound calling method provided for an exemplary embodiment of the present application;

[0018] Figure 4 A schematic diagram of another outbound calling method provided for an exemplary embodiment of the present application;

[0019] Figure 5 A flowchart of a time period prediction model training method provided for an exemplary embodiment of the present application;

[0020] Figure 6 A structural block diagram of an outbound calling device provided for an exemplary embodiment of the present application;

[0021] Figure 7 A schematic structural diagram of an electronic device provided for an exemplary embodiment of the present application. DETAILED DESCRIPTION

[0022] In order to make the purpose, technical solution and advantages of the present application clearer, the technical solution of the present application will be clearly and completely described below in combination with the specific embodiments of the present application and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application.

[0023] In view of the problem of low outbound call connection rate in random time periods in existing outbound call scenarios, the outbound call method provided in the embodiment of the present application includes: receiving an outbound call task, the outbound call task includes a target user number; obtaining user call behavior characteristics of the target user number, the user call behavior characteristics are obtained based on historical call data mining of the target user number; inputting the user call behavior characteristics into a pre-trained time period prediction model to estimate the dialing time period to obtain a target time period, the target time period indicates that the connection rate of calling the target user number during the target time period is the highest; initiating an outbound call to the target user number during the target time period. The embodiment of the present application can accurately estimate the target time period with the highest connection rate through the user call behavior characteristics, so initiating an outbound call to the target user number during the target time period can improve the connection rate of the outbound call.

[0024] In this embodiment, the outbound calling method can be an overall outbound calling method implemented with the help of a cloud computing system. In addition, the server that executes the outbound calling method can be a cloud server, so as to take advantage of the resources on the cloud to run various neural network models, such as the time period estimation model of the present application; relative to the cloud, the outbound calling method can also be applied to service-side devices such as conventional servers or server arrays, which are not limited here.

[0025] In addition, the outbound call method provided in the embodiment of the present application is applied to outbound call scenarios, where, for example, in a questionnaire scenario, the outbound call system calls the user to collect relevant information provided by the user. In a collection scenario, the outbound call system calls the user to prompt the user to complete the relevant payment as soon as possible. In these scenarios, only when the user connects to the outbound call can the completion of the relevant task be improved. Therefore, the improvement of the connection rate will further improve the completion of the task, and will also reduce the number of outbound calls of the outbound call system and improve the efficiency of the outbound call.

[0026] The technical solutions provided by various embodiments of the present application are described in detail below in conjunction with the accompanying drawings.

[0027] Figure 1 The following is a flowchart of a method for outbound calling provided by an exemplary embodiment of the present application. Figure 1 The outbound calling method shown specifically includes the following steps:

[0028] S101, receiving an outbound call task issued by an upstream task system.

[0029] Among them, the outbound call task includes the target user number. Specifically, refer to Figure 1 The execution subject of the present application may be a server 21, in which an outbound call system 211 is deployed. The outbound call system 211 receives an outbound call task issued by an upstream task system 22, wherein the upstream task system 22 may be deployed in the server 21 or in other servers. The outbound call task includes a target user number and may also include: specific content that requires an outbound call, such as a questionnaire survey, collection, etc.

[0030] S102: Acquire the user calling behavior characteristics of the target user number.

[0031] Reference Figure 2 A preset storage space 212 may be deployed in the server 21, and the preset storage space 212 may pre-store a correspondence between a user number and a user calling behavior feature. When the outbound calling system 211 obtains a target user number, the corresponding user calling behavior feature may be obtained in the preset storage space 212 according to the target user number.

[0032] Furthermore, the user call behavior characteristics are obtained by mining the historical call data based on the target user number. The user call behavior characteristics include: the behavior characteristics of the user corresponding to the target user number in the historical calls. If the outbound call system has made an outbound call to the target user number in the past, the historical call data of each outbound call to the target user number can be recorded. The historical call data of the target user number include: outbound call time, ringing duration, whether the call is connected, and the duration of the call. Mining the historical call data of the target user number for multiple times can mine the user call behavior characteristics, wherein the user call behavior characteristics refer to the behavior characteristics of the target user number when receiving an outbound call from the outbound call system, specifically, the user call behavior characteristics include: at least one of the number of connections, the number of hang-ups, the total connection rate (the number of connections to the number of outbound calls), the total connection duration, the average connection duration, and the connection rate in each time period of the user corresponding to the target user number in the history of the calls.

[0033] In the embodiment of the present application, the user calling behavior characteristics can represent some characteristics corresponding to the target user number, thereby improving the estimation accuracy of the target time period.

[0034] S103, inputting the user's calling behavior characteristics into a pre-trained time period prediction model to estimate the dialing time period to obtain a target time period.

[0035] The target time period indicates that the connection rate of calling the target user number during the target time period is the highest. Specifically, the time period prediction model is pre-trained. The time period prediction model can be pre-trained in the server 21, and then the trained time period prediction model is put on the outbound call system 211.

[0036] Furthermore, a day can be divided into multiple time periods, and the target time period is one of the multiple time periods. For example, every half hour, one hour or two hours of a day is a time period. Among them, in an embodiment of the present application, one hour is determined as a time period, and the target time period is one of the hours. This division method can effectively cut and cover the whole day, and each cut time period is not too short, because a too short time period will increase the difficulty of estimating the target time period, resulting in poor estimation results for the target time period, and a long time period will not effectively improve the connection rate of outbound calls during the estimated target time period.

[0037] In the embodiment of the present application, the estimated target time period can effectively improve the connection rate of outbound calls.

[0038] S104: Initiate an outbound call to a target user number within a target time period.

[0039] When the time enters the target time period, an outbound call is initiated to the target user number. Specifically, the target time period refers to a period of time in a day, and an outbound call can be initiated to the target user number after entering the target time period on the same day, or after entering the target time period on the next day.

[0040] For example, if an outbound call task is received at 12:00 on the same day, if the target time period is between 14:00 and 15:00, an outbound call can be initiated to the target user number between 14:00 and 15:00 on the same day. If the target time period is between 10:00 and 11:00, an outbound call can be initiated to the target user number on the next day.

[0041] Furthermore, an outbound call reminder may be given to the staff during the target time period, and the staff may use the outbound call system to initiate an outbound call to the target user number. Alternatively, the outbound call system may intelligently and automatically initiate an outbound call to the target user number.

[0042] The embodiment of the present application is applied in an outbound call scenario, and the outbound call method provided includes: receiving an outbound call task issued by an upstream task system, the outbound call task includes a target user number; obtaining user call behavior characteristics of the target user number, the user call behavior characteristics are obtained based on historical call data mining of the target user number; inputting the user call behavior characteristics into a pre-trained time period prediction model to estimate the dialing time period, and obtaining a target time period, the target time period indicates that the connection rate of calling the target user number during the target time period is the highest; initiating an outbound call to the target user number during the target time period. The embodiment of the present application can accurately estimate the target time period with the highest connection rate through the user call behavior characteristics, so initiating an outbound call to the target user number during the target time period can improve the connection rate of the outbound call.

[0043] Figure 3 A flowchart of another outbound calling method provided by an exemplary embodiment of the present application. Figure 3 As shown, the specific steps include:

[0044] S301, obtaining historical outbound call data corresponding to the outbound call system.

[0045] The outbound call system corresponds to multiple caller numbers, and each caller number is used to make outbound calls to a user number in a corresponding time period.

[0046] Specifically, the outbound call system deploys multiple caller numbers, such as 8XXXXXX, 8YYYYYY and 8ZZZZZZ. When making an outbound call, the outbound call system can select any caller number to make an outbound call to the user number. Each outbound call will generate corresponding historical outbound call data, including: outbound call task identifier, caller number, called number, user portrait features corresponding to the called number, outbound call time, ringing duration, whether the call is connected, and connection duration.

[0047] For example, one of the historical outbound call data includes: the outbound call task identifier is 0S01, the calling number is 8XXXXXX, the called number is 137XXXXXXX, the user profile features are female, worker, the outbound call time is 18:00, the ringing time is 15 seconds, the call is connected, and the connection time is 3 seconds. Another historical outbound call data includes: the outbound call task identifier is 0X02, the calling number is 8YYYYYY, the called number is 177XXXXXXX, the user profile features are female, student, the outbound call time is 15:00, the ringing time is 8 seconds, the call is connected, and the connection time is 45 seconds.

[0048] In the embodiment of the present application, all historical outbound call data of the outbound call system within a historical period of time can be obtained, wherein the period of time can be one month or three months, which is not limited.

[0049] S302, data mining is performed on historical outbound call data to obtain global characteristics of the time period, user calling behavior characteristics and user portrait characteristics of each user number, and calling behavior characteristics of each calling number.

[0050] Among them, the time period global features include: the call features of the outbound call system in each time period. Specifically, the time period global features include: the number of calls, total connection rate, total call duration, and average call duration of the outbound call system in each historical time period. Each time period can refer to every hour of a day, or every two hours. For example, in the past month, the outbound call system has 40 calls between 6:00 and 7:00, with a connection rate of 5%, a total call duration of 10 seconds, and an average call duration of 5 seconds. The number of calls between 8:00 and 9:00 is 500, with a connection rate of 30%, a total call duration of 50 minutes, and an average call duration of 20 seconds. The number of calls between 13:00 and 14:00 is 200, with a connection rate of 10%, a total call duration of 2600 seconds, and an average call duration of 13 seconds. The number of calls between 19:00 and 20:00 is 700, with a connection rate of 35%, a total call duration of 7840 seconds, and an average call duration of 32 seconds.

[0051] In addition, the user call behavior characteristics of the user number include the number of connections, the number of hang-ups, the total connection rate (the ratio of the number of connections to the number of outbound calls), the total connection time, the average connection time, and the connection rate in each time period corresponding to each user number. For example, the user number is 137XXXXXXX, the number of connections for this user number in the past month is 5 times, the number of hang-ups is 3 times, the total connection rate is 62.5%, the total connection time is 60 seconds, the average connection time is 12 seconds, the connection rate between 8:00 and 9:00 is 30%, the connection rate between 10:00 and 11:00 is 50%, the connection rate between 19:00 and 20:00 is 100%, and the connection rate at other times is 0. The user number is 177XXXXXXX. In the past month, the number of calls connected for this user number was 12 times, the number of calls disconnected was 18 times, the total connection rate was 40%, the total connection time was 48 seconds, the average connection time was 4, the connection rate between 8:00 and 9:00 was 20%, the connection rate between 10:00 and 11:00 was 30%, the connection rate between 13:00 and 14:00 was 80%, and the connection rate at other times was 0.

[0052] Furthermore, the user portrait features corresponding to each user number include user attribute information such as user age, gender, occupation, and education level. For example, the user age corresponding to user number 137XXXXXXX is 30 years old, gender is female, occupation is worker, and education level is college. The user age corresponding to user number 177XXXXXXX is 18 years old, gender is female, occupation is student, and education level is undergraduate.

[0053] In the embodiment of the present application, the calling behavior characteristics of each calling number refer to the behavior characteristics of each calling number in the historical calling process, wherein the calling behavior characteristics include: at least one of the number of calls, the number of connections, the number of connections in each period, the number of hang-ups in each period, and the connection rate in each period of the calling number in a historical period. For example, the total number of calls of the calling number 8XXXXXX is 20,000 times, the number of connections is 12,000 times, for example, the number of connections from 10:00 to 11:00 is 60 times, the number of hang-ups is 20 times, and the connection rate is 75%.

[0054] In the embodiment of the present application, by mining the historical outbound call data, the user calling behavior characteristics and user portrait characteristics obtained can better characterize the user characteristics, and the time period global characteristics obtained can better characterize the overall outbound call characteristics. The calling call behavior characteristics can better characterize the calling number characteristics.

[0055] In an embodiment of the present application, the mining of historical outbound call data can be performed in the server, and the global characteristics of the time period obtained by the mining, the user calling behavior characteristics and user portrait characteristics of each user number, and the calling calling behavior characteristics of each calling number are stored in a preset storage space. When the outbound call system needs to make an outbound call, these characteristic data can be called to determine the target time period.

[0056] S303: Receive the outbound call task sent by the upstream task system.

[0057] Among them, the outbound call task includes the target user number; the outbound call task also includes: multiple dialable time periods.

[0058] The specific implementation process of this step refers to S101 and will not be repeated here.

[0059] S304, when an outbound call task corresponding to a target user number is received for the first time, the global characteristics of the time period are input into a time period estimation model to estimate the dialing time period, and obtain a target time period.

[0060] The global features of a time period are obtained by mining the historical call data of the outbound call system in each time period.

[0061] In the embodiment of the present application, if the target user number is received for the first time, there is no user call behavior feature corresponding to the target user number, and only the time period global feature can be input into the time period estimation model to estimate the dialing time period to obtain the target time period.

[0062] In an optional embodiment, if the outbound call task includes a caller number and a dialable time period, if it is the first time to receive an outbound call task corresponding to the target user number, the caller call behavior characteristics corresponding to the caller number can also be determined in the preset storage space according to the caller number, and then the caller call behavior characteristics, the dialable time period and the time period global characteristics are input into the time period estimation model to estimate the dialing time period to obtain the target time period. Among them, the dialable time period is set in advance according to the task content. For example, if the task content is to recommend a certain product and requires an outbound call between 8:00 and 12:00, the target time period obtained is a time period between 8:00 and 12:00, such as 9:00 to 10:00.

[0063] In the embodiment of the present application, the target user number for the first outbound call can be accurately determined, and the target time period with the highest connection rate can be accurately determined, thereby improving the outbound call efficiency.

[0064] S305: acquiring a user calling behavior feature of the target user number without receiving an outbound calling task corresponding to the target user number for the first time.

[0065] The specific implementation process of this step refers to S102 and will not be repeated here.

[0066] S306, inputting the user's calling behavior characteristics into a pre-trained time period prediction model to estimate the dialing time period to obtain a target time period.

[0067] In an optional embodiment, the user's calling behavior characteristics are input into a pre-trained time period prediction model to predict the dialing time period, and the step of obtaining the target time period includes: inputting the user's calling behavior characteristics into the encoding layer of the time period prediction model for encoding processing to obtain a coding vector; inputting the coding vector into the estimation layer of the time period prediction model for estimation processing to obtain the target time period.

[0068] Among them, when one or more features among multiple dialable time periods, time period global features, user call behavior features, user portrait features and caller call behavior features are obtained, since the types of features are not the same, they cannot be directly input into the estimation layer for prediction. Therefore, it is necessary to perform feature preprocessing on different types of features. In an embodiment of the present application, these features include discrete features and continuous features. For discrete features, the encoding layer can use one-hot (a coding method) encoding method to map discrete feature values ​​to dictionary identifiers, and then perform feature transformation through DNN Embedding (a network embedding technology), so that the discrete dictionary identifiers are transformed into numerical features of a uniform interval (such as a 0-1 interval). At the same time, the dimension of the feature will usually be transformed from a one-dimensional feature to an N-dimensional feature. This can enhance the representation ability of the feature, which is conducive to the subsequent estimation layer to learn deeper feature differences. For continuous features, since the variation ranges of features of different dimensions are different, in order for the estimation layer to better fit the training data, the encoding layer normalizes the continuous features of different dimensions so that the features of different dimensions can be transformed into the same variation range (such as the 0-1 range). After feature transformation, a feature sequence that can be input into the estimation layer can be obtained.

[0069] In addition, during the training phase of the time period prediction model, since the outbound call scenario is a scenario with high concurrency and strict time requirements, the use of shallow CTR for offline smoothness of the time period prediction model can achieve fast training speed.

[0070] In an optional embodiment, the user's calling behavior characteristics are input into a pre-trained time period prediction model to estimate the dialing time period, and the step of obtaining the target time period includes: obtaining at least one of the caller's calling behavior characteristics, user portrait characteristics and time period global characteristics as target additional characteristics; inputting the user's calling behavior characteristics and the target additional characteristics into the time period prediction model to estimate the dialing time period to obtain the target time period.

[0071] Among them, the outbound call task also includes the target calling number, and the calling behavior characteristics of the calling number are obtained based on the historical call data mining of the target calling number. In this step, the calling behavior characteristics of the calling number include: at least one of the number of calls of the target calling number in the historical time, the number of connections, the number of connections in each time period, the number of hang-ups in each time period and the connection rate in each time period. The user portrait characteristics are obtained based on the historical call data mining of the target user number. In this step, the user portrait characteristics refer to the attribute information of the user to which the target user number belongs. Specifically, the user portrait characteristics include: at least one of the user age, gender, occupation and education level of the user corresponding to the target user number. The time period global characteristics are obtained based on the historical call data mining of the outbound call system in each time period. The time period global characteristics include: at least one of the number of calls of the outbound call system in each historical time period, the total connection rate, the total call duration and the average call duration.

[0072] In an optional embodiment, the outbound call task also includes: multiple dialable time periods, inputting the user's calling behavior characteristics into a pre-trained time period prediction model to estimate the dialing time period, and obtaining a target time period, including: inputting the user's calling behavior characteristics and multiple dialable time periods into the time period prediction model to estimate the dialing time period, and obtaining a target time period, the target time period being one of the multiple dialable time periods.

[0073] Among them, refer to Figure 4 , the outbound call task may include: at least one of: multiple dialable time periods, caller numbers and target user numbers. Each dialable time period may be a pre-divided time period, such as one hour. For example, the multiple dialable time periods include: four dialable time periods from 8:00 to 9:00, 9:00 to 10:00, 11:00 to 12:00, and 15:00 to 16:00. Then, the caller call behavior features may be obtained in the preset storage space according to the caller number, and the user call behavior features and user portrait features may be obtained in the preset storage space according to the target user number, and then the time period global features may be obtained, and the time period global features, user call behavior features, user portrait features, caller call behavior features and one or more of the dialable time periods may be input into the coding layer for encoding to obtain a coding vector, and the coding vector may be input into the estimation layer to obtain the target time period.

[0074] Furthermore, when multiple items are input to the coding layer, each item can be encoded to obtain a corresponding feature vector, and multiple feature vectors can be fused to obtain a coding vector. For example, a time period global feature and a user call behavior feature are input to the coding layer, and the coding layer encodes the time period global feature to obtain a first feature vector, and the coding layer encodes the user call behavior feature to obtain a second feature vector, and the first feature vector and the second feature vector are fused to obtain a coding vector.

[0075] In an optional embodiment, when an outbound call is made to a target user number for the first time, and there is no dialable time period and caller number in the outbound call task, the user portrait feature can be first obtained through the target user number. If the user portrait feature is not obtained, the time period global feature is input into the coding layer for encoding to obtain a coding vector. In addition, if it is the first time an outbound call is made to the target user number, and there is a dialable time period and caller number in the outbound call task, and the user portrait feature is obtained through the target user number, at least one of the time period global feature, multiple dialable time periods, user portrait feature, and caller call behavior feature can be input into the coding layer to obtain a coding vector.

[0076] Furthermore, if the outbound call task includes multiple dialable time periods, the target time period is one of the multiple dialable time periods.

[0077] In the embodiment of the present application, the time period prediction model has good Robustness and can output corresponding target time periods for one or more of the input multiple dialable time periods, time period global characteristics, user calling behavior characteristics, user portrait characteristics and calling calling behavior characteristics, so as to improve the connection rate of outbound calls.

[0078] S307, automatically initiate an outbound call to the target user number within the target time period.

[0079] In the embodiment of the present application, when entering the target time period, the outbound call system can intelligently and automatically use the target calling number to initiate an outbound call to the target user number.

[0080] The embodiment of the present application is applied in an outbound call scenario, and the outbound call method provided includes: receiving an outbound call task issued by an upstream task system, the outbound call task includes a target user number; obtaining user call behavior characteristics of the target user number, the user call behavior characteristics are obtained based on historical call data mining of the target user number; inputting the user call behavior characteristics into a pre-trained time period prediction model to estimate the dialing time period, and obtaining a target time period, the target time period indicates that the connection rate of calling the target user number during the target time period is the highest; initiating an outbound call to the target user number during the target time period. The embodiment of the present application can accurately estimate the target time period with the highest connection rate through the user call behavior characteristics, so initiating an outbound call to the target user number during the target time period can improve the connection rate of the outbound call.

[0081] Figure 5 A flowchart of the steps of a training process of a time period prediction model provided by an exemplary embodiment of the present application. Figure 3 As shown, the specific steps include:

[0082] S501, obtaining multiple training samples.

[0083] The training samples include: a user number sample, a calling number sample, a calling period and a calling result in a historical call, and the calling result indicates whether the call is connected during the calling period.

[0084] In an embodiment of the present application, before S504, the historical outbound call data corresponding to the outbound call system is obtained, and the outbound call system corresponds to multiple caller numbers, each caller number is used to make outbound calls to the user number in the corresponding time period; data mining is performed on the historical outbound call data to obtain the global characteristics of the time period, the user calling behavior characteristics and user portrait characteristics of each user number, and the calling behavior characteristics of each calling number. The global characteristics of the time period include: the calling characteristics of the outbound call system in each time period. Please refer to S301 and S302 for details, which will not be repeated here.

[0085] In addition, the time period prediction model is trained multiple times, and each training requires a training sample. Among them, a training sample is the relevant data in a historical call. For example, the user number sample of a historical call is 176XXXXXXxx, the caller number sample is 8XXXXXX, the call time period is 10:00 to 11:00, and the call result is connected.

[0086] S502, obtaining a user history call feature sample and / or a user portrait feature sample corresponding to the user number sample.

[0087] In the embodiments of this application, refer to Figure 4 , in the preset storage space, according to the historical call feature samples and / or user portrait feature samples corresponding to the user number sample, wherein the historical call feature samples are the historical call features corresponding to the user number sample, and the user portrait features are the user portrait features corresponding to the user number sample.

[0088] S503, obtaining a caller calling behavior feature sample corresponding to the caller number sample.

[0089] Reference Figure 4 In the preset storage, a corresponding calling behavior feature sample is obtained according to the calling number sample, wherein the calling behavior feature sample is the calling behavior feature of the calling number sample.

[0090] S504, inputting at least one of the user's historical call feature samples, the user portrait feature samples, the caller's call behavior feature samples and the time period global feature into a time period estimation model to estimate the dialing time period to obtain an estimated time period.

[0091] Among them, multiple dialable time periods can also be randomly set according to each training sample, and the multiple dialable time periods and at least one of the user's historical call feature samples, user portrait feature samples, caller call behavior feature samples and time period global features can be input into the time period estimation model to estimate the dialing time period and obtain the estimated time period.

[0092] In an embodiment of the present application, the time period prediction model is trained multiple times with different data (multiple dialable time periods, and the multiple dialable time periods and user historical call feature samples, user portrait feature samples, caller call behavior feature samples, and time period global features), which can improve the robustness of the time period prediction model.

[0093] S505, adjusting the time period estimation model according to the estimated time period and the call result to obtain a trained time period estimation model.

[0094] Among them, the cross entropy loss function can be used to determine the loss value of the estimated time period and the call result, and the time period estimation model can be adjusted according to the loss value and the loss value threshold, wherein the call result refers to whether the call period is dialed. If the estimated time period is 10:00 to 11:00, and the call result is not connected from 10:00 to 11:00, the loss value is greater than the loss value threshold, and the time period estimation model is adjusted according to the loss value. If the call result is connected from 10:00 to 11:00, the loss value is less than the loss value threshold, and the next training sample is used to continue training the time period estimation model. If the estimated time period is 10:00 to 11:00, and the call result is connected from 9:00 to 10:00, the loss value is greater than the loss value threshold, and the time period estimation model is adjusted according to the loss value.

[0095] In an embodiment of the present application, the time period prediction model can be trained offline in the server. After the training is completed, the time period prediction model is deployed online, that is, deployed to the outbound call system. The obtained time period prediction model can predict and score multiple dialable time periods, and determine the time period with the highest score as the target time period. When the target time period is reached, the outbound call system is used to dial the corresponding target user number.

[0096] Furthermore, in the training phase, the embodiment of the present application adopts a time period segmentation design for the dialable time throughout the day, and takes each natural hourly time period as a learning target, such as 00:00 to 1:00, 1:00 to 2:00, ..., 23:00 to 24:00. This is not only conducive to the fitting of the time period prediction model to the training samples and enhancing the generalization performance of the time period prediction model, but also can prevent the problem that the predicted time period loses its effect and fails to improve the connection rate due to the too large dialable time period design. The embodiment of the present application adopts a design of natural hourly time period segmentation, which can effectively improve the robustness of the offline time period prediction model and help the time period prediction model converge quickly.

[0097] In the embodiments of this application, refer to Figure 6 In addition to providing an outbound calling method, an outbound calling device 60 is also provided. The outbound calling device 60 includes: a receiving module 61, which is used to receive an outbound calling task issued by an upstream task system, and the outbound calling task includes a target user number; an obtaining module 62, which is used to obtain a user calling behavior feature of the target user number, and the user calling behavior feature is obtained based on historical call data mining of the target user number, and the user calling behavior feature includes: at least one of the number of connections, the number of hang-ups, the total connection rate, the total connection time, the average connection index, and the connection rate of each time period of the target user number in historical calls; a processing module 63, which is used to input the user calling behavior feature into a pre-trained time period estimation model to estimate the dialing time period, and obtain a target time period, and the target time period indicates that the connection rate of calling the target user number in the target time period is the highest; an outbound calling module 64, which is used to initiate an outbound call to the target user number within the target time period.

[0098] In an optional embodiment, the processing module 63 is specifically used to: input the user call behavior characteristics into the encoding layer of the time period prediction model for encoding processing to obtain a coding vector; input the coding vector into the estimation layer of the time period prediction model for estimation processing to obtain a target time period.

[0099] In an optional embodiment, the processing module 63 is specifically used to: obtain at least one of the caller's call behavior characteristics, user portrait characteristics and time period global characteristics as target additional characteristics; input the user call behavior characteristics and the target additional characteristics into the time period estimation model to estimate the dialing time period to obtain the target time period; wherein, the outbound call task also includes a target calling number, and the caller's call behavior characteristics are obtained based on the historical call data mining of the target calling number, and the caller's call behavior characteristics include: at least one of the number of calls, the number of connections, the number of connections in each time period, the number of hang-ups in each time period and the connection rate of each time period of the target calling number in the historical time period; the user portrait characteristics are obtained based on the historical call data mining of the target user number, and the user portrait characteristics include: at least one of the user age, gender, occupation and education level of the user corresponding to the target user number; the time period global characteristics are obtained based on the historical call data mining of the outbound call system in each time period, and the time period global characteristics include: at least one of the number of calls, the total connection rate, the total call duration and the average call duration of the outbound call system in each historical time period.

[0100] In an optional embodiment, the processing module 63 is also used to: when an outbound call task corresponding to a target user number is received for the first time, the global characteristics of the time period are input into a time period estimation model to estimate the dialing time period to obtain a target time period, and the global characteristics of the time period are obtained based on historical call data mining of the outbound call system in each time period.

[0101] In an optional embodiment, the outbound calling module 64 is specifically configured to automatically initiate an outbound call to a target user number within a target time period.

[0102] In an optional embodiment, the outbound call task also includes: multiple dialable time periods, and the processing module 63 is specifically used to: input the user's call behavior characteristics and multiple dialable time periods into the time period prediction model to estimate the dialing time period, and obtain the target time period, which is one of the multiple dialable time periods.

[0103] In an optional embodiment, the outbound call device 60 also includes: an acquisition module (not shown), which is used to acquire historical outbound call data corresponding to the outbound call system, the outbound call system corresponds to multiple caller numbers, and each caller number is used to make outbound calls to the user number in the corresponding time period; a mining module (not shown), which is used to perform data mining on the historical outbound call data to obtain global characteristics of the time period, user call behavior characteristics and user portrait characteristics of each user number, and caller call behavior characteristics of each caller number. The global characteristics of the time period include: call characteristics of the outbound call system in each time period.

[0104] In an optional embodiment, the outbound call device 60 also includes: a training module (not shown), which is used to obtain multiple training samples, and the training samples include: a user number sample, a caller number sample, a call time period and a call result in a historical call, and the call result indicates whether the call is connected during the call time period; obtaining a user historical call feature sample and / or a user portrait feature sample corresponding to the user number sample; obtaining a caller call behavior feature sample corresponding to the caller number sample; inputting at least one of the user historical call feature sample, the user portrait feature sample, the caller call behavior feature sample and the time period global feature into a time period estimation model to estimate the dialing time period to obtain an estimated time period; adjusting the time period estimation model according to the estimated time period and the call result to obtain a trained time period estimation model.

[0105] The outbound call device provided in the embodiment of the present application receives an outbound call task issued by an upstream task system, wherein the outbound call task includes a target user number; obtains user call behavior characteristics of the target user number, wherein the user call behavior characteristics are obtained based on historical call data mining of the target user number; inputs the user call behavior characteristics into a pre-trained time period prediction model to estimate the dialing time period, and obtains a target time period, wherein the target time period indicates that the connection rate of calling the target user number during the target time period is the highest; and initiates an outbound call to the target user number during the target time period. The embodiment of the present application can accurately estimate the target time period with the highest connection rate through the user call behavior characteristics, so initiating an outbound call to the target user number during the target time period can improve the connection rate of the outbound call.

[0106] In addition, in some of the processes described in the above embodiments and the accompanying drawings, multiple operations that appear in a specific order are included, but it should be clearly understood that these operations may not be executed in the order in which they appear in this article or executed in parallel, and are only used to distinguish between different operations, and the sequence number itself does not represent any execution order. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this article are used to distinguish different messages, devices, modules, etc., do not represent the order of precedence, and do not limit the "first" and "second" to different types.

[0107] Figure 7 This is a schematic diagram of the structure of an electronic device provided by an exemplary embodiment of the present application. Figure 7 As shown, the electronic device 70 includes: a processor 71, and a memory 72 communicatively connected to the processor 71, and the memory 72 stores computer-executable instructions.

[0108] The processor executes the computer-executable instructions stored in the memory to implement the outbound calling method provided by any of the above method embodiments, and the specific functions and technical effects that can be achieved are not repeated here.

[0109] An embodiment of the present application further provides a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, they are used to implement the outbound calling method provided by any of the above method embodiments.

[0110] An embodiment of the present application also provides a computer program product, which includes: a computer program, which is stored in a readable storage medium. At least one processor of an electronic device can read the computer program from the readable storage medium, and at least one processor executes the computer program so that the electronic device executes the outbound call method provided by any of the above method embodiments.

[0111] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0112] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0113] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0114] The above-mentioned integrated unit implemented in the form of a software functional unit can be stored in a computer-readable storage medium. The above-mentioned software functional unit is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor to perform some steps of the methods of each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), disk or optical disk and other media that can store program codes.

[0115] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional modules is used as an example for illustration. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. The specific working process of the device described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0116] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the invention disclosed herein. The present application is intended to cover any modification, use or adaptation of the present application, which follows the general principles of the present application and includes common knowledge or customary techniques in the art that are not disclosed in the present application. The specification and examples are intended to be exemplary only, and the true scope and spirit of the present application are indicated by the following claims.

[0117] It should be understood that the present application is not limited to the precise structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present application is limited only by the appended claims.

Claims

1. An outbound calling method, characterized in that: include: Receiving an outbound call task issued by an upstream task system, wherein the outbound call task includes a target user number; Acquire a user calling behavior feature of the target user number, wherein the user calling behavior feature is obtained based on historical call data mining of the target user number, and the user calling behavior feature includes: at least one of the number of connection times, number of hang-up times, total connection rate, total connection time, average connection index, and connection rate in each time period of the target user number in historical calls; Inputting the user's calling behavior characteristics into a pre-trained time period prediction model to estimate the dialing time period, and obtaining a target time period, wherein the target time period indicates that the call connection rate of calling the target user number during the target time period is the highest; Initiate an outbound call to the target user number within the target time period.

2. The outbound calling method according to claim 1, characterized in that: The step of inputting the user's calling behavior characteristics into a pre-trained time period prediction model to predict the dialing time period to obtain a target time period includes: Inputting the user call behavior feature into the encoding layer of the time period prediction model for encoding processing to obtain an encoding vector; The encoding vector is input into the estimation layer of the time period estimation model for estimation processing to obtain the target time period.

3. The outbound calling method according to claim 1, characterized in that: The step of inputting the user's calling behavior characteristics into a pre-trained time period prediction model to predict the dialing time period to obtain a target time period includes: Acquire at least one of the caller's calling behavior feature, the user portrait feature, and the time period global feature as a target additional feature; Inputting the user calling behavior feature and the target additional feature into the time period estimation model to estimate the dialing time period to obtain the target time period; Among them, the outbound call task also includes a target calling number, and the calling call behavior characteristics are obtained based on the historical call data mining of the target calling number, and the calling call behavior characteristics include: at least one of the number of calls, the number of connections, the number of connections in each time period, the number of hang-ups in each time period and the connection rate in each time period of the target calling number in the historical time period; the user portrait characteristics are obtained based on the historical call data mining of the target user number, and the user portrait characteristics include: at least one of the user age, gender, occupation and education level of the user corresponding to the target user number; the time period global characteristics are obtained based on the historical call data mining of the outbound call system in each time period, and the time period global characteristics include: at least one of the number of calls, total connection rate, total call duration and average call duration of the outbound call system in each historical time period.

4. The outbound calling method according to claim 1, characterized in that: Also includes: When an outbound call task corresponding to the target user number is received for the first time, the global characteristics of the time period are input into the time period estimation model to estimate the dialing time period to obtain the target time period. The global characteristics of the time period are obtained by mining the historical call data of the outbound call system in each time period.

5. The outbound calling method according to claim 1, characterized in that: Initiating an outbound call to the target user number within the target time period includes: In the target time period, an outbound call is automatically initiated to the target user number.

6. The outbound calling method according to any one of claims 1 to 5, characterized in that: The outbound call task further includes: a plurality of available time periods for dialing, wherein the user's calling behavior characteristics are input into a pre-trained time period prediction model to predict the dialing time period to obtain a target time period, including: The user calling behavior characteristics and the multiple dialable time periods are input into the time period estimation model to estimate the dialing time period to obtain the target time period, which is one of the multiple dialable time periods.

7. The outbound calling method according to any one of claims 1 to 5, characterized in that: Before receiving the outbound call task sent by the upstream task system, the method further includes: Acquire historical outbound call data corresponding to an outbound call system, wherein the outbound call system corresponds to multiple caller numbers, each of which is used to make an outbound call to a user number in a corresponding time period; Data mining is performed on the historical outbound call data to obtain global characteristics of the time period, user calling behavior characteristics and user portrait characteristics of each user number, and calling behavior characteristics of each calling number. The global characteristics of the time period include: calling characteristics of the outbound call system in each time period.

8. The outbound calling method according to claim 7, characterized in that: The time period estimation model is trained in the following way: Acquire multiple training samples, the training samples including: a user number sample, a calling number sample, a call period and a call result in a historical call, the call result indicating whether the call is connected during the call period; Obtain user history call feature samples and / or user portrait feature samples corresponding to the user number samples; Obtain a caller calling behavior feature sample corresponding to the caller number sample; Inputting at least one of the user's historical call feature sample, the user portrait feature sample, the caller's call behavior feature sample, and the time period global feature into the time period estimation model to estimate the dialing time period, thereby obtaining an estimated time period; The time period estimation model is adjusted according to the estimated time period and the call result to obtain a trained time period estimation model.

9. An outbound calling device, characterized in that: include: A receiving module, used for receiving an outbound call task issued by an upstream task system, wherein the outbound call task includes a target user number; An acquisition module, configured to acquire a user call behavior feature of the target user number, wherein the user call behavior feature is obtained based on historical call data mining of the target user number, and the user call behavior feature includes: a behavior feature of a user corresponding to the target user number in historical calls; A processing module, used for inputting the user calling behavior characteristics into a pre-trained time period prediction model to predict the dialing time period, and obtaining a target time period, wherein the target time period indicates that the dialing rate of calling the target user number during the target time period is the highest; The outbound calling module is used to initiate an outbound call to the target user number within the target time period.

10. An electronic device, characterized in that: include: A processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor implements the outbound calling method according to any one of claims 1 to 8 when executing the computer program.

11. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed on an electronic device, the electronic device executes the outbound calling method according to any one of claims 1 to 8.

Images