Outbound call control method and device, electronic equipment and readable storage medium

Abstract

The application discloses an outbound call control method and device, electronic equipment and a readable storage medium. The method of the application trains a model by using real-time online data, predicts the number of idle human operators in advance by using the trained model, adjusts the outbound call speed control parameter based on the prediction result, can help the call center system to respond to changes in time, adjust the appropriate outbound call speed, ensure that the human operator has enough time and resources to handle each call, keep the idle operator at a low level all the time, and does not increase the number of call losses, realizes the automatic control of the call loss, can provide more timely and accurate service, and improves the customer satisfaction.

Description

Technical Field

[0001] This application relates to the field of outbound call control technology, and in particular to an outbound call control method, device, electronic device and readable storage medium. Background Technology

[0002] For some service industries such as insurance, banking, and telecommunications, a large number of outbound calls are needed to communicate with customers. In outbound calling scenarios, a robot is first used to talk to the connected customers to screen for potential customers. Once it is determined that the customer has the intention to buy, they are then transferred to a human agent for sales.

[0003] In existing technologies, outbound calls are usually made at a fixed speed. However, the distribution of incoming calls at different times of the day is unpredictable. During peak periods of increased incoming calls, making outbound calls at a fixed speed often results in failures to transfer to human agents, leading to a large number of call detail records lost. This not only results in the loss of target customers but also leads to low customer conversion rates for intelligent outbound marketing. Summary of the Invention

[0004] In view of this, this application provides an outbound call control method, device, electronic device, and readable storage medium. The main purpose is to solve the technical problem in the prior art where outbound calls are made at a fixed speed, often resulting in failure to transfer to a human agent, leading to a large number of call detail records lost. This not only causes the loss of target customers but also results in a low customer conversion rate for intelligent outbound marketing.

[0005] According to a first aspect of this application, an outbound call control method is provided, the method comprising:

[0006] In response to the outbound call initiation command, the first preset outbound call speed control parameter is obtained;

[0007] According to the first preset outbound call speed control parameters, control multiple robot agents to make outbound calls;

[0008] Obtain multiple outbound call connection rates and multiple idle human agent seats at multiple moments;

[0009] Based on multiple outbound call connection rates, multiple idle agent seats, outbound call capacity data, and the first preset outbound call speed control parameters, a preset machine learning model is trained to generate an outbound call speed control parameter determination model.

[0010] Obtain multiple sets of second preset outbound call speed control parameters;

[0011] The model is determined using outbound call speed control parameters, and the target outbound call speed control parameters are determined from multiple sets of second preset outbound call speed control parameters.

[0012] Based on the target outbound call speed control parameters, multiple robot agents are controlled to make outbound calls.

[0013] Optionally, the step of training a preset machine learning model to generate a model for determining the outbound call speed control parameters based on multiple outbound call connection rates, multiple idle agent seats, outbound call capacity data, and a first preset outbound call speed control parameter specifically includes:

[0014] Based on multiple outbound call connection rates, number of idle human agents, outbound call capacity data, and first preset outbound call speed control parameters at each first moment, feature data is generated;

[0015] Generate labeled data based on the number of available human agents at each second time point;

[0016] Use feature data and label data as a dataset;

[0017] Get the preset model parameters of the preset machine learning model;

[0018] The preset machine learning model is trained using the dataset and preset model parameters to generate the trained outbound call speed control parameters to determine the model.

[0019] Optionally, the step of training a preset machine learning model using a dataset and preset model parameters to generate trained outbound call speed control parameters and determine the model specifically includes:

[0020] The dataset is divided into a training dataset and a validation dataset;

[0021] During model training, a preset machine learning model is trained using a training dataset and preset model parameters;

[0022] Using the validation dataset, calculate the mean squared error of the preset machine learning model;

[0023] If the mean squared error is less than the preset error threshold, stop model training and generate a model to determine the outbound call speed control parameters.

[0024] Optionally, the step of determining the target outbound call speed control parameter from multiple sets of second preset outbound call speed control parameters, using the outbound call speed control parameter determination model, specifically includes:

[0025] The second preset outbound call speed control parameters of each group are sequentially input into the outbound call speed control parameter determination model, and the predicted number of idle human agents at the third time point is output.

[0026] The target outbound call speed control parameters are determined based on multiple predicted idle agent seats and preset idle agent seat number thresholds from multiple sets of second preset outbound call speed control parameters.

[0027] Optionally, the step of determining the target outbound call speed control parameter based on multiple predicted idle agent numbers and preset idle agent number thresholds of multiple sets of second preset outbound call speed control parameters specifically includes:

[0028] Each predicted number of available human agents is compared with a preset threshold for the number of available human agents.

[0029] Based on the comparison results, among the multiple predicted idle agent seats, the target number of idle agent seats that is closest to the preset idle agent seat number threshold is determined.

[0030] The second preset outbound call speed control parameter corresponding to the target number of available human agents is used as the target outbound call speed control parameter.

[0031] Optionally, the outbound call capacity data includes the total number of agents, and the method further includes:

[0032] Based on the total number of seats, a preset threshold for the number of available human seats is generated.

[0033] Optionally, the method further includes:

[0034] Based on the target outbound call speed control parameters, generate a prompt message for adjusting the outbound call speed;

[0035] The notification message will be sent to the relevant personnel's terminals.

[0036] According to a second aspect of this application, an outbound call control device is provided, the device comprising:

[0037] The first acquisition module is used to acquire the first preset outbound call speed control parameters in response to the outbound call initiation command;

[0038] The control module is used to control multiple robot agents to make outbound calls according to the first preset outbound call speed control parameters;

[0039] The second acquisition module is used to acquire multiple outbound call connection rates and multiple idle human agent seats at multiple first moments;

[0040] The generation module is used to train a preset machine learning model based on multiple outbound call connection rates, multiple idle human agent seats, outbound call capacity data and a first preset outbound call speed control parameter, and generate an outbound call speed control parameter determination model.

[0041] The third acquisition module is used to acquire multiple sets of second preset outbound call speed control parameters;

[0042] The determination module is used to determine the model using outbound call speed control parameters, and to determine the target outbound call speed control parameters from multiple sets of second preset outbound call speed control parameters;

[0043] The control module is also used to control multiple robot agents to make outbound calls based on the target outbound call speed control parameters.

[0044] According to a third aspect of this application, an electronic device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of any of the methods in the first aspect.

[0045] According to a fourth aspect of this application, a readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described in any one of the first aspects.

[0046] By employing the above technical solution, this application provides an outbound call control method, device, electronic device, and readable storage medium. Upon receiving an outbound call initiation command, it controls a robot agent to make outbound calls based on a first preset outbound call speed control parameter. During the outbound call process, the outbound call connection rate and the number of idle human agents at each first moment are combined with outbound call capacity data and the initial outbound call speed control parameter to train a preset machine learning model, generating an outbound call speed control parameter determination model. Subsequently, based on the number of idle human agents corresponding to each set of second preset outbound call speed control parameters predicted by the outbound call speed control parameter determination model, a target outbound call speed control parameter that best matches the ideal number of idle human agents is selected. Finally, the robot agent is controlled to make outbound calls according to the determined target outbound call speed control parameter. Compared to the existing technology that uses fixed outbound call speed control parameters, this method lacks flexibility and personalization, cannot adapt to real-time changes in the environment and conditions of outbound call activities, makes it impossible to guarantee the incoming call speed, increases the number of call losses, and thus leads to a decrease in the effectiveness and efficiency of outbound call activities. In this embodiment, during the outbound call process, a model is trained using real-time online data. The trained model is then used for advance prediction, and the outbound call speed control parameters are adjusted based on the prediction results. This helps the call center system respond to changes in a timely manner to adjust the appropriate outbound call speed. It ensures that human agents have sufficient time and resources to handle each call, keeping the number of idle agents at a low level without increasing the number of call losses. This achieves automatic control of call losses, enabling the provision of more timely and accurate service and improving customer satisfaction.

[0047] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0048] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the scope of this application. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0049] Figure 1 This paper illustrates a flowchart of an outbound call control method provided in an embodiment of this application.

[0050] Figure 2 This paper illustrates a schematic flowchart of another outbound call control method provided in an embodiment of this application.

[0051] Figure 3 A schematic block diagram of the training dataset of the training model provided in the embodiments of this application is shown;

[0052] Figure 4 A schematic diagram of the structure of an outbound call control device provided in an embodiment of this application is shown. Detailed Implementation

[0053] Exemplary embodiments of the present application will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present application are shown in the drawings, it should be understood that the present application may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this application will be thorough and complete, and will fully convey the scope of the present application to those skilled in the art.

[0054] This application provides an outbound call control method, such as... Figure 1 As shown, the method includes:

[0055] S101. In response to the outbound call start command, obtain the first preset outbound call speed control parameter.

[0056] The outbound call control method provided in this application embodiment can be applied to call center systems. For service industries such as insurance sales, financial services, wealth management services, real estate sales, and event invitations, the call center system controls robot agents to screen interested customers and then transfer them to human agents for sales. This enables marketing and promotion such as loan applications, credit card applications, or the purchase of insurance and wealth management products, helping to screen customers with higher intent and thus increasing enterprise efficiency.

[0057] In existing technologies, the outbound call speed control parameters for robot agents are typically pre-set by relevant personnel. These fixed parameters are used to control the robot agents throughout the outbound call process. However, fixed outbound call speed control parameters cannot flexibly respond to risks and changes. If the robot agent's outbound call speed is too fast, human agents may not be able to handle all calls in time, leading to an increase in hang-ups and customers not receiving human service. Some customers may even give up waiting and hang up, resulting in increased call loss and impacting service quality and customer satisfaction. Conversely, if the robot agent's outbound call speed is too slow, human agents may experience slow incoming calls, leaving them idle for extended periods. This means that call center resources are not fully utilized, resulting in a waste of human and equipment resources. Therefore, to better balance call loss and incoming call speed, and to reduce call loss while achieving faster incoming calls on the agent side, this application proposes predicting the number of idle human agents at a future time based on actual conditions during the outbound call process, and then adjusting the outbound call speed control parameters according to the prediction results.

[0058] The first preset outbound call speed control parameter is a parameter setting for controlling the calling speed of the robot agents, including the outbound call frequency (i.e., the number of calls per minute or hour) and concurrent call volume (i.e., the data of outbound calls made simultaneously). Before the call center system controls multiple robot agents to start making outbound calls, relevant personnel can set the first preset outbound call speed control parameter based on the system's historical outbound call data and current business objectives (e.g., marketing outbound calls may be faster, while customer service outbound calls may need to be relatively stable). This parameter serves as the initial outbound call speed control parameter for the robot agents. When the system receives an outbound call start command initiated by relevant personnel, it can obtain the initial outbound call speed control parameter input by those personnel.

[0059] S102. Control multiple robot agents to make outbound calls according to the first preset outbound call speed control parameters.

[0060] In this step, after receiving the outbound call start command, the first preset outbound call speed control parameter set by relevant personnel can be used as the initial control parameter to control each robot agent to make outbound calls.

[0061] S103. Obtain multiple outbound call connection rates and multiple idle agent seats at multiple first moments.

[0062] In practical applications, call center systems include a Computer Telephony Integration (CTI) system. The CTI system communicates with multiple robotic and human agents. It returns a series of online data for both robotic and human agents at a pre-set frequency (e.g., every 60 seconds), such as call volume, call success rate, average call duration, call start time, end time, call duration, call waiting time, and agent status. Subsequently, based on the collected real-time data, the outbound call connection rate of the robotic agents and the number of available human agents are determined at each given moment.

[0063] Furthermore, the multiple "first moments" are determined based on the data collection frequency. Specifically, the system starts making outbound calls at 9:00 AM and collects online data every 60 seconds. Therefore, the multiple "first moments" are 9:01, 9:02, and up to the current moment. Subsequently, the online data at each time interval is analyzed to calculate the outbound call connection rate of the robot agents and the number of available human agents at each "first moment".

[0064] S104. Based on multiple outbound call connection rates, multiple idle agent seats, outbound call capacity data, and the first preset outbound call speed control parameters, train the preset machine learning model to generate an outbound call speed control parameter determination model.

[0065] In this step, the outbound call speed control parameter determination model is used to predict the number of available agent seats at a certain moment based on different preset outbound call speed control parameters, and then select the most suitable outbound call speed control parameters for the real-time situation based on the prediction results. Specifically, after obtaining the outbound call connection rate and the number of available agent seats over a period of time, the preset machine learning model is trained using multiple real-time outbound call connection rates, multiple numbers of available agent seats, outbound call capacity data, and the first preset outbound call speed control parameters as training data to generate the outbound call speed control parameter determination model.

[0066] Among them, outbound call capacity data refers to the number of outbound call channels for robot agents and the total number of human agents pre-set in the call center system.

[0067] Optionally, the preset machine learning model can be the LightGBM model. LightGBM has a fast training and prediction speed, which is crucial for the real-time outbound call speed control of robot agents, as it is necessary to optimize and make decisions on a large number of outbound calls in a short period of time. LightGBM can quickly train the model and determine the optimal outbound call speed control parameters in a timely and accurate manner.

[0068] S105. Obtain multiple sets of second preset outbound call speed control parameters.

[0069] S106. Determine the model using outbound call speed control parameters, and determine the target outbound call speed control parameters from multiple sets of second preset outbound call speed control parameters.

[0070] In steps S105 and S106, in the outbound call scenario, the ideal number of available agent seats is set based on the actual total number of agent seats, thereby controlling the outbound call speed to keep the number of available agent seats near the ideal value. The multiple sets of second preset outbound call speed control parameters are pre-set outbound call speed control parameters by relevant personnel. During the outbound call process, each set of second preset outbound call speed control parameters is sequentially input into the trained outbound call speed control parameter determination model to obtain the predicted number of available agent seats for each set of second preset outbound call speed control parameters at a set time. Then, the number of available agent seats closest to the ideal number is selected from the multiple predicted numbers of available agent seats, and the second preset outbound call speed control parameter closest to this is used as the target outbound call speed control parameter.

[0071] In practical applications, relevant personnel formulate multiple sets of second preset outbound call speed control parameters based on the goals and requirements of the outbound call activity, the number, capabilities, and working hours of the robot agents, as well as the system's historical outbound call data. When an outbound call is initiated, one of these sets is selected as the first preset outbound call speed control parameters to control the robot agent to make the outbound call.

[0072] By using the above method, the trained model selects the most suitable outbound call speed control parameters from multiple sets of second preset outbound call speed control parameters, so that the idle seats are kept at a low level, while not increasing the number of call losses, which helps to optimize the effectiveness and efficiency of outbound call activities.

[0073] S107. Control multiple robot agents to make outbound calls according to the target outbound call speed control parameters.

[0074] In this step, the environment and conditions of outbound calling activities can change at any time, such as the number of customers, agent availability, and network conditions. After determining the target outbound calling speed control parameters using the trained model, the system controls multiple robot agents to make outbound calls according to the newly determined target outbound calling speed control parameters. By adjusting the outbound calling speed control parameters in real time, the system can flexibly adjust the outbound calling speed according to different environments and conditions, ensuring that agents have sufficient time and resources to handle each call, keeping the number of idle agents at a low level, and not increasing the number of call drops. This helps the call center system respond to changes and adjust activity strategies in a timely manner, ensuring call quality and improving customer satisfaction.

[0075] The outbound call control method provided in this application, upon receiving an outbound call initiation command, controls a robot agent to make outbound calls based on a first preset outbound call speed control parameter. During the outbound call process, the outbound call connection rate and the number of idle human agents at each first moment are combined with outbound call capacity data and the initial outbound call speed control parameter to train a preset machine learning model, generating an outbound call speed control parameter determination model. Subsequently, based on the number of idle human agents corresponding to each set of second preset outbound call speed control parameters predicted by the outbound call speed control parameter determination model, a target outbound call speed control parameter that is closest to the ideal number of idle human agents is selected. Finally, the robot agent is controlled to make outbound calls according to the determined target outbound call speed control parameter. Compared with the existing technology that uses fixed outbound call speed control parameters, this method lacks flexibility and personalization, cannot adapt to real-time changes in the environment and conditions of outbound call activities, makes it impossible to guarantee the incoming call speed, increases the number of call losses, and thus leads to a decrease in the effectiveness and efficiency of outbound call activities. In this embodiment, during the outbound call process, a model is trained using real-time online data. The trained model is then used to predict the number of available human agents in advance. Based on the prediction results, the outbound call speed control parameters are adjusted. This helps the call center system respond to changes in a timely manner to adjust the appropriate outbound call speed. It ensures that human agents have sufficient time and resources to handle each call, keeping the number of available agents at a low level without increasing the number of call losses. This achieves automatic control of call losses, enabling the provision of more timely and accurate services and improving customer satisfaction.

[0076] Furthermore, such as Figure 2 As shown, as a refinement and extension of the specific implementation of the above embodiments, in order to fully illustrate the specific implementation process of this embodiment, this application provides another outbound call control method, which includes:

[0077] S201. In response to the outbound call start command, obtain the first preset outbound call speed control parameter.

[0078] In existing technologies, the outbound call speed control parameters for robot agents are typically pre-set by relevant personnel. These fixed parameters are used to control the robot agents throughout the outbound call process. However, fixed outbound call speed control parameters cannot flexibly respond to risks and changes. If the robot agent's outbound call speed is too fast, human agents may not be able to handle all calls in time, leading to an increase in hang-ups and customers not receiving human service. Some customers may even give up waiting and hang up, resulting in increased call loss and impacting service quality and customer satisfaction. Conversely, if the robot agent's outbound call speed is too slow, human agents may experience slow incoming calls, leaving them idle for extended periods. This means that call center resources are not fully utilized, resulting in a waste of human and equipment resources. Therefore, to better balance call loss and incoming call speed, and to reduce call loss while achieving faster incoming calls on the agent side, this application proposes predicting the number of idle human agents at a future time based on actual conditions during the outbound call process, and then adjusting the outbound call speed control parameters according to the prediction results.

[0079] The first preset outbound call speed control parameter is a parameter setting for controlling the calling speed of the robot agents, including the outbound call frequency (i.e., the number of calls per minute or hour) and concurrent call volume (i.e., the data of outbound calls made simultaneously). Before the call center system controls multiple robot agents to start making outbound calls, relevant personnel can set the first preset outbound call speed control parameter based on the system's historical outbound call data and current business objectives (e.g., marketing outbound calls may be faster, while customer service outbound calls may need to be relatively stable). This parameter serves as the initial outbound call speed control parameter for the robot agents. When the system receives an outbound call start command initiated by relevant personnel, it can obtain the initial outbound call speed control parameter input by those personnel.

[0080] S202. Control multiple robot agents to make outbound calls according to the first preset outbound call speed control parameters.

[0081] In this step, after receiving the outbound call start command, the first preset outbound call speed control parameter set by relevant personnel can be used as the initial control parameter to control each robot agent to make outbound calls.

[0082] S203. Obtain multiple outbound call connection rates and multiple idle agent seats at multiple first moments.

[0083] In practical applications, call center systems include a Computer Telephony Integration (CTI) system. The CTI system communicates with multiple robotic and human agents. It returns a series of online data for both robotic and human agents at a pre-set frequency (e.g., every 60 seconds), such as call volume, call success rate, average call duration, call start time, end time, call duration, call waiting time, and agent status. Subsequently, based on the collected real-time data, the outbound call connection rate of the robotic agents and the number of available human agents are determined at each given moment.

[0084] Furthermore, the multiple "first moments" are determined based on the data collection frequency. Specifically, the system starts making outbound calls at 9:00 AM and collects online data every 60 seconds. Therefore, the multiple "first moments" are 9:01, 9:02, and up to the current moment. Subsequently, the online data at each time interval is analyzed to calculate the outbound call connection rate of the robot agents and the number of available human agents at each "first moment".

[0085] S204. Generate feature data based on the multiple outbound call connection rates, number of idle agent seats, outbound call capacity data, and the first preset outbound call speed control parameters for each first moment.

[0086] S205. Generate tag data based on the number of available human agents at each second time point.

[0087] S206. Use the feature data and label data as a dataset.

[0088] In steps S204 to S206, before model training, a dataset for training the model needs to be prepared. This dataset includes feature data, which serves as input data for training and predicting the model. It contains various feature information that the model needs to utilize, describing various attributes and influencing factors related to the target variable. Based on the outbound call connection rate and the number of available seats at each first moment, combined with the number of outbound call channels, the total number of human agents, and a first preset outbound call speed control parameter, the feature data corresponding to each first moment is used.

[0089] Furthermore, the training dataset also includes labeled data, which is the target variable that the model needs to predict. In supervised learning, labeled data is known data with correct answers, and the model's goal is to use the feature data to predict the values ​​of the labeled data. In outbound call scenarios, based on experience, relevant personnel determine the outbound call speed control parameters for each first moment, which have the greatest impact on t+x*3 (the second moment), where t is the first moment and x is the time interval between two first moments (i.e., the time interval for collecting online data, such as 60 seconds). In other words, based on each first moment t, its corresponding second moment can be determined, and the number of available agent seats at each second moment can be used as the labeled data.

[0090] Subsequently, all feature data and label data are aggregated to obtain the training dataset for the preset machine learning model.

[0091] In practical application scenarios, such as Figure 3 The diagram shown illustrates the training dataset for the training model. Based on a series of online data returned by the CTI system at intervals of x seconds, n first time points are pre-selected. Let t be the first time point, then the multiple first time points are t1, t2, ..., tt... n The characteristic data M for each first time point t is determined by the outbound call connection rate (m). t 1 ) and number of available seats (m t 2 ), the number of outbound call channels (m t 3 ), Total number of human operator seats (m) t 4 ), combined with the first preset outbound call speed control parameter (m t a1 m t1 a2 Composed of, i.e., M t =[m t 1 m t 2 m t 3 m t 4 m t a1 m t a2 ], then the model's feature dataset X t =[M t1 M t2 M tnFurthermore, based on experience, relevant personnel determined the outbound call speed control parameters, which had the greatest impact on the number of available agent seats at time t+x*3 (the second time). Therefore, the number of available agent seats at time t+x*3 (Y) was determined. t () as tag data.

[0092] S207. Obtain the preset model parameters of the preset machine learning model.

[0093] S208. Use the dataset and preset model parameters to train the preset machine learning model and generate the trained outbound call speed control parameters to determine the model.

[0094] In steps S207 and S208, relevant personnel determine the preset model parameters of the preset machine learning model based on the specific task of the preset machine learning model (predicting the number of available human agents at a future moment) and the training dataset, and send the set preset model parameters to the system. The dataset and preset model parameters are then input into the preset machine learning model, which iteratively optimizes itself based on the data and parameters, learning patterns and rules in the data to obtain the trained outbound call speed control parameters and determine the model.

[0095] Optionally, the parameters required by the model, such as tree depth, learning rate, and regularization parameters, can be determined according to the task requirements.

[0096] Optionally, in this embodiment, step S208, training a preset machine learning model using a dataset and preset model parameters to generate a trained outbound call speed control parameter determination model, specifically includes: dividing the dataset into a training dataset and a validation dataset; training the preset machine learning model using the training dataset and preset model parameters during model training; calculating the mean square error of the preset machine learning model using the validation dataset; and stopping model training when the mean square error is less than a preset error threshold, thereby generating the outbound call speed control parameter determination model.

[0097] In this embodiment, the dataset is divided into a training dataset and a validation dataset. First, a preset machine learning model is trained using the training dataset and preset model parameters. During training, mean squared error is selected as the evaluation metric, and the validation dataset is used to evaluate the model's performance. Specifically, the preset machine learning model is used to predict on the validation dataset to obtain predicted label data (i.e., the predicted number of available agent seats at the second time step). The mean squared error is calculated using the actual number of available agent seats at the second time step and the predicted number of available agent seats. If the mean squared error is less than a preset error threshold, the model is considered successfully trained, and the trained outbound call speed control parameters are obtained to confirm the model.

[0098] Optionally, the preset error threshold can be set by relevant personnel based on the specific task of the model, and this application does not impose any specific limitations on it.

[0099] By using the above method, mean squared error as an evaluation metric to validate the preset machine learning model can provide an intuitive and easily interpretable performance measurement standard, and can adapt to the characteristics of regression problems, making the model have high sensitivity and stability.

[0100] S209. Input the second preset outbound call speed control parameters of each group into the outbound call speed control parameter determination model in sequence, and output the predicted number of idle human agents at the third time point.

[0101] S210. Determine the target outbound call speed control parameters based on multiple predicted idle agent numbers and preset idle agent number thresholds of multiple sets of second preset outbound call speed control parameters.

[0102] In steps S209 and S210, during the process of controlling the robot agents to make outbound calls, in order to make the outbound call speed control parameters of the robot agents more in line with the current actual situation, after training and generating a model using real-time outbound call data, multiple sets of second preset outbound call speed control parameters pre-set by relevant personnel are input into the trained outbound call speed control parameter determination model, and the predicted number of available human agents at a future preset time (third time) corresponding to each set of control parameters is output. Subsequently, using multiple predicted numbers of available human agents and preset threshold numbers of available human agents, the target outbound call speed control parameters that best fit the current actual environment are selected. Then, the target outbound call speed control parameters are used to control multiple robot agents to make outbound calls, realizing real-time adjustment of the outbound call speed, reducing the number of call losses while ensuring the incoming call speed.

[0103] Optionally, the trained model can be used to predict the number of available agent seats in advance, and the most suitable target outbound call speed idle parameters can be selected based on the prediction results. Therefore, based on the current time, a third time is determined, where the third time is the current time plus the data collection time interval (e.g., 60 seconds) × 3. This enables the prediction of the number of available agent seats three minutes in advance using the model and multiple sets of preset outbound call speed control parameters.

[0104] In this embodiment of the application, optionally, step S210, which involves determining the target outbound call speed control parameter based on multiple predicted idle agent numbers and a preset idle agent number threshold of multiple sets of second preset outbound call speed control parameters, specifically includes: sequentially comparing each predicted idle agent number with the preset idle agent number threshold; determining the target idle agent number that is closest to the preset idle agent number threshold among multiple predicted idle agent numbers based on the comparison results; and using the second preset outbound call speed control parameter corresponding to the target idle agent number as the target outbound call speed control parameter.

[0105] In this embodiment, the preset threshold for the number of idle agent seats is the ideal number of idle agent seats during the current outbound call activity. After predicting multiple idle agent seat numbers at the third time point, each predicted idle agent seat number is compared with the preset threshold for the number of idle agent seats. Among the multiple predicted idle agent seat numbers, the target number of idle agent seats that is closest to the preset threshold is found. Subsequently, a set of second preset outbound call speed control parameters corresponding to the target number of idle agent seats is used as the target outbound call speed control parameters for controlling the robot agents.

[0106] By utilizing the trained model in this way, changes in available agent availability can be detected in advance, allowing for timely adjustments to the outbound call speed of the robotic agents. This reduces call loss while ensuring a certain incoming call speed.

[0107] Optionally, in order to ensure the accuracy of the preset idle seat threshold, the method of this embodiment further includes: generating a preset idle human seat number threshold based on the total number of seats.

[0108] In this embodiment, the total number of human agents included in the outbound call capacity data is obtained, and the ideal number of available agents for this outbound call activity (i.e., the preset threshold for the number of available human agents) is set to 5% of the total number of agents.

[0109] S211. Control multiple robot agents to make outbound calls according to the target outbound call speed control parameters.

[0110] In this step, the environment and conditions of outbound calling activities can change at any time, such as the number of customers, agent availability, and network conditions. After determining the target outbound calling speed control parameters using the trained model, the system controls multiple robot agents to make outbound calls according to the newly determined target outbound calling speed control parameters. By adjusting the outbound calling speed control parameters in real time, the system can flexibly adjust the outbound calling speed according to different environments and conditions, ensuring that agents have sufficient time and resources to handle each call, keeping the number of idle agents at a low level, and not increasing the number of call drops. This helps the call center system respond to changes and adjust activity strategies in a timely manner, ensuring call quality and improving customer satisfaction.

[0111] In practical applications, the model can be optimized at time intervals (e.g., 20 minutes). The optimized model can then be used to adjust the outbound call speed every 20 minutes, ensuring that the robot agent's outbound call speed matches the current situation. This keeps idle agents at a low level without increasing the number of lost calls.

[0112] S212. Generate a prompt message for adjusting the outbound call speed based on the target outbound call speed control parameters; send the prompt message to the terminal of the relevant personnel.

[0113] In this step, after adjusting the outbound calling speeds of multiple robot agents based on the target outbound calling speed control parameters, a prompt message indicating the adjustment is generated. This prompt message is then sent to the terminals of relevant personnel, allowing them to promptly understand the speed changes of the robot agents and identify problems in a timely manner, thus optimizing the customer experience.

[0114] Furthermore, as Figure 1 To specifically implement the method, this application provides an outbound call control device 300, such as... Figure 4 As shown, the device includes:

[0115] The first acquisition module 301 is used to acquire the first preset outbound call speed control parameters in response to the outbound call initiation command;

[0116] Control module 302 is used to control multiple robot agents to make outbound calls according to the first preset outbound call speed control parameters;

[0117] The second acquisition module 303 is used to acquire multiple outbound call connection rates and multiple idle human agent seats at multiple first moments;

[0118] The generation module 304 is used to train a preset machine learning model based on multiple outbound call connection rates, multiple idle human agent seats, outbound call capacity data and a first preset outbound call speed control parameter, and generate an outbound call speed control parameter determination model.

[0119] The third acquisition module 305 is used to acquire multiple sets of second preset outbound call speed control parameters;

[0120] The determination module 306 is used to determine the model using outbound call speed control parameters, and to determine the target outbound call speed control parameters from multiple sets of second preset outbound call speed control parameters;

[0121] The control module 302 is also used to control multiple robot agents to make outbound calls according to the target outbound call speed control parameters.

[0122] Optionally, module 304 is generated, specifically including:

[0123] The first generation unit is used to generate feature data based on multiple outbound call connection rates, number of idle human agents, outbound call capacity data and first preset outbound call speed control parameters at each first moment;

[0124] The second generation unit is used to generate label data based on the number of available human agents at each second time point;

[0125] The third generation unit is used to take feature data and label data as a dataset;

[0126] The acquisition unit is used to acquire the preset model parameters of the preset machine learning model;

[0127] The fourth generation unit is used to train a preset machine learning model using the dataset and preset model parameters, and generate a model for determining the outbound call speed control parameters after training.

[0128] Optionally, the generation module 304 further includes:

[0129] Split units are used to divide a dataset into training and validation datasets;

[0130] The training unit is used to train a preset machine learning model using a training dataset and preset model parameters during the model training process.

[0131] The computing unit is used to calculate the mean squared error of a pre-defined machine learning model using a validation dataset.

[0132] The fourth generation unit is also used to stop model training and generate a model for determining outbound call speed control parameters when the mean square error is less than a preset error threshold.

[0133] Optionally, module 306 is defined, specifically including:

[0134] The output unit is used to sequentially input each group of second preset outbound call speed control parameters into the outbound call speed control parameter determination model, and output the predicted number of idle human agents at the third time point.

[0135] The first determining unit is used to determine the target outbound call speed control parameters based on multiple predicted idle agent numbers and preset idle agent number thresholds of multiple sets of second preset outbound call speed control parameters.

[0136] Optionally, module 306 further includes:

[0137] The comparison unit is used to compare each predicted number of available human agents with a preset threshold number of available human agents in turn.

[0138] The second determining unit is used to determine the target number of idle human agents that is closest to the preset threshold number of idle human agents among multiple predicted numbers of idle human agents based on the comparison results.

[0139] The third determining unit is used to use the second preset outbound call speed control parameter corresponding to the target number of available human agents as the target outbound call speed control parameter.

[0140] Optionally, the outbound call capacity data includes the total number of agents, and the generation module 304 is further used to generate a preset threshold for the number of available human agents based on the total number of agents.

[0141] Optionally, the generation module 304 is also used to generate a prompt message for adjusting the outbound call speed based on the target outbound call speed control parameters.

[0142] Optionally, the device further includes:

[0143] The sending module 307 is used to send the prompt information to the terminal of the relevant personnel.

[0144] The outbound call control device 300 provided in this application embodiment, upon receiving an outbound call start command, controls a robot agent to make outbound calls based on a first preset outbound call speed control parameter. During the outbound call process, the outbound call connection rate and the number of idle human agents at each first moment are combined with outbound call capacity data and the initial outbound call speed control parameter to train a preset machine learning model, generating an outbound call speed control parameter determination model. Subsequently, based on the number of idle human agents corresponding to each group of second preset outbound call speed control parameters predicted by the outbound call speed control parameter determination model, a target outbound call speed control parameter that is closest to the ideal number of idle human agents is selected. Finally, the robot agent is controlled to make outbound calls according to the determined target outbound call speed control parameter. Compared with the existing technology that uses fixed outbound call speed control parameters, this method lacks flexibility and personalization, cannot adapt to real-time changes in the environment and conditions of outbound call activities, makes it impossible to guarantee the incoming call speed, increases the number of call losses, and thus leads to a decrease in the effectiveness and efficiency of outbound call activities. In this embodiment, during the outbound call process, a model is trained using real-time online data. The trained model is then used for advance prediction, and the outbound call speed control parameters are adjusted based on the prediction results. This helps the call center system respond to changes in a timely manner to adjust the appropriate outbound call speed. It ensures that human agents have sufficient time and resources to handle each call, keeping the number of idle agents at a low level without increasing the number of call losses. This achieves automatic control of call losses, enabling the provision of more timely and accurate service and improving customer satisfaction.

[0145] In an exemplary embodiment, this application also provides an electronic device, including a memory and a processor. The memory stores a computer program, and the processor executes the program stored in the memory to perform the outbound call control method described in the above embodiments.

[0146] In an exemplary embodiment, this application also provides a readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the outbound call control method.

[0147] Through the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented in hardware or by using software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solution of this application can be embodied in the form of a software product, which can be stored in a non-volatile readable storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) and includes several instructions to cause an electronic device (such as a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0148] Those skilled in the art will understand that the accompanying drawings are merely schematic diagrams of a preferred embodiment, and the modules or processes shown in the drawings are not necessarily essential for implementing this application.

[0149] Those skilled in the art will understand that the modules in the apparatus of the implementation scenario can be distributed within the apparatus of the implementation scenario as described, or they can be located in one or more apparatuses different from this implementation scenario with corresponding changes. The modules of the above-mentioned implementation scenario can be combined into one module, or they can be further divided into multiple sub-modules.

[0150] The serial numbers in this application are for descriptive purposes only and do not represent the superiority or inferiority of the implementation scenario.

[0151] The above disclosures are only a few specific implementation scenarios of this application. However, this application is not limited to these. Any variations that can be conceived by those skilled in the art should fall within the protection scope of this application.

Claims

1. An outbound call control method, characterized by, include: In response to the outbound call initiation command, the first preset outbound call speed control parameter is obtained; According to the first preset outbound call speed control parameters, control multiple robot agents to make outbound calls; Obtain multiple outbound call connection rates and multiple idle human agent seats at multiple moments; Based on the multiple outbound call connection rates, the multiple number of idle human agents, outbound call capacity data, and the first preset outbound call speed control parameters, a preset machine learning model is trained to generate an outbound call speed control parameter determination model. Obtain multiple sets of second preset outbound call speed control parameters; The model is determined using the outbound call speed control parameters, and the target outbound call speed control parameters are determined from the plurality of sets of second preset outbound call speed control parameters; Based on the target outbound call speed control parameters, the multiple robot agents are controlled to make outbound calls.

2. The method according to claim 1, characterized in that, The step of training a preset machine learning model based on the multiple outbound call connection rates, the multiple idle agent seats, outbound call capacity data, and the first preset outbound call speed control parameters to generate an outbound call speed control parameter determination model specifically includes: Based on multiple outbound call connection rates, number of idle human agents, outbound call capacity data, and first preset outbound call speed control parameters at each first moment, feature data is generated; Generate labeled data based on the number of available human agents at each second time point; The feature data and the label data are used as a dataset; Obtain the preset model parameters of the preset machine learning model; The preset machine learning model is trained using the dataset and the preset model parameters to generate the trained outbound call speed control parameter determination model.

3. The method according to claim 2, characterized in that, The step of training the preset machine learning model using the dataset and the preset model parameters to generate the trained outbound call speed control parameters for determining the model specifically includes: The dataset is divided into a training dataset and a validation dataset; During model training, the preset machine learning model is trained using the training dataset and the preset model parameters; Using the validation dataset, calculate the mean squared error of the preset machine learning model; If the mean square error is less than a preset error threshold, stop model training and generate the outbound call speed control parameter determination model.

4. The method according to claim 1, characterized in that, The step of determining the target outbound call speed control parameter from the plurality of sets of second preset outbound call speed control parameters using the outbound call speed control parameter determination model specifically includes: The second set of preset outbound call speed control parameters for each group are sequentially input into the outbound call speed control parameter determination model, and the predicted number of available human agents at the third time point is output. The target outbound call speed control parameters are determined based on multiple predicted idle agent numbers and preset idle agent number thresholds from multiple sets of second preset outbound call speed control parameters.

5. The method according to claim 4, characterized in that, The step of determining the target outbound call speed control parameter based on multiple predicted idle agent numbers and preset idle agent number thresholds of multiple sets of second preset outbound call speed control parameters specifically includes: Each predicted number of available human agents is compared with a preset threshold for the number of available human agents. Based on the comparison results, among the multiple predicted number of available agent seats, the target number of available agent seats that is closest to the preset threshold number of available agent seats is determined. The second preset outbound call speed control parameter corresponding to the target number of available human agents is used as the target outbound call speed control parameter.

6. The method according to any one of claims 1 to 5, characterized in that, Outbound call capacity data includes the total number of agents, and the method further includes: Based on the total number of seats, a preset threshold for the number of available human seats is generated.

7. The method according to any one of claims 1 to 5, characterized in that, The method further includes: Based on the target outbound call speed control parameters, generate a prompt message for adjusting the outbound call speed; The notification message will be sent to the relevant personnel's terminals.

8. An outbound call control device, characterized in that, include: The first acquisition module is used to acquire the first preset outbound call speed control parameters in response to the outbound call initiation command; The control module is used to control multiple robot agents to make outbound calls according to the first preset outbound call speed control parameters; The second acquisition module is used to acquire multiple outbound call connection rates and multiple idle human agent seats at multiple first moments; The generation module is used to train a preset machine learning model based on the multiple outbound call connection rates, the multiple number of idle human agents, outbound call capacity data, and the first preset outbound call speed control parameters, and generate an outbound call speed control parameter determination model. The third acquisition module is used to acquire multiple sets of second preset outbound call speed control parameters; The determining module is used to determine the model using the outbound call speed control parameters, and to determine the target outbound call speed control parameters from the plurality of sets of second preset outbound call speed control parameters; The control module is also used to control the multiple robot agents to make outbound calls according to the target outbound call speed control parameters.

9. An electronic device comprising a memory and a processor, the memory storing a computer program, characterized in that, The processor executes the steps of any one of claims 1 to 7 when executing a computer program.

10. A readable storage medium having a computer program stored thereon, characterized in that, When a computer program is executed by a processor, it implements the steps of any one of claims 1 to 7.

Images