Agent outbound calling method, device, equipment and storage medium

Abstract

This application relates to the field of financial technology and provides a method, apparatus, device, and storage medium for outbound calls by agents. The method comprises: obtaining user characteristic data of multiple customers to be called outbound, and obtaining agent data for making outbound calls to the multiple customers to be called outbound; inputting each user characteristic data into a pre-trained outbound call time prediction model for prediction to determine the optimal outbound call time for making outbound calls to each customer to be called outbound; inputting the agent data and the multiple optimal outbound call time periods into a pre-trained agent allocation model for processing to determine a target outbound call time period for making outbound calls to each customer to be called outbound; and making outbound calls to the multiple customers to be called outbound based on the target outbound call time period corresponding to each customer to be called outbound. This application also relates to blockchain, aiming to rationally allocate agent resources and improve outbound call efficiency and marketing conversion rate.

Description

Technical Field

[0001] The present application relates to the technical field of financial technology, and in particular to an agent outbound calling method, device, equipment and storage medium. Background Art

[0002] Telephone call systems are an integral component of many products and services, such as insurance sales, financial services, wealth management services, real estate sales, and event invitations. Many banks, insurance companies, and securities firms have established their own telemarketing platforms, leveraging them to promote applications for loans and credit cards, as well as purchases of insurance and wealth management products, thereby expanding and retaining customer base and increasing business profitability.

[0003] Taking insurance companies' underwriting business as an example, agents primarily prioritize outbound calls based on customer level and concentrate calls during popular hours. This approach doesn't reflect customer willingness to answer calls and places high demands on agent resource allocation, resulting in low outbound call efficiency and marketing conversion rates. Therefore, improving outbound call efficiency and marketing conversion rates has become a pressing issue. Summary of the Invention

[0004] The main purpose of this application is to provide an agent outbound calling method, device, equipment, and storage medium, aiming to improve outbound calling efficiency and marketing conversion rate through artificial intelligence-related technical means. This application can be applied to financial technology fields such as insurance sales services, financial services, and wealth management services.

[0005] In a first aspect, the present application provides an agent outbound calling method, comprising:

[0006] Acquire user feature data of a plurality of customers to be called outbound, and acquire agent data for making outbound calls to the plurality of customers to be called outbound;

[0007] Inputting each of the user feature data into a pre-trained outbound call time period prediction model for prediction, so as to determine the optimal outbound call time period for each of the outbound call customers;

[0008] Inputting the agent data and the plurality of optimal outbound call time periods into a pre-trained agent allocation model for processing to determine a target outbound call time period for making outbound calls to each of the customers to be called;

[0009] Based on the target outbound calling time period corresponding to each of the outbound calling customers, outbound calls are made to a plurality of the outbound calling customers.

[0010] In a second aspect, the present application further provides an agent outbound calling device, the agent outbound calling device comprising:

[0011] An acquisition module, configured to acquire user characteristic data of a plurality of customers to be called outbound, and acquire seat data for making outbound calls to the plurality of customers to be called outbound;

[0012] A prediction module, configured to input each of the user feature data into a pre-trained outbound call time period prediction model for prediction, so as to determine the optimal outbound call time period for each of the outbound call-waiting customers;

[0013] a processing module, configured to input the agent data and the plurality of optimal outbound call time periods into a pre-trained agent allocation model for processing, so as to determine a target outbound call time period for making an outbound call to each of the customers to be called;

[0014] The outbound calling module is used for making outbound calls to a plurality of the outbound calling customers based on the target outbound calling time period corresponding to each of the outbound calling customers.

[0015] In a third aspect, the present application further provides a computer device comprising a processor, a memory, and a computer program stored in the memory and executable by the processor, wherein when the computer program is executed by the processor, the steps of the agent outbound call method as described above are implemented.

[0016] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the steps of the agent outbound call method as described above are implemented.

[0017] The present application provides an agent outbound calling method, apparatus, equipment and storage medium. The present application obtains user characteristic data of multiple customers to be called outbound, and obtains agent data for making outbound calls to multiple customers to be called outbound; each user characteristic data is input into a pre-trained outbound call time prediction model for prediction to determine the optimal outbound call time for making outbound calls to each customer to be called outbound; the agent data and multiple optimal outbound call time periods are input into a pre-trained agent allocation model for processing to determine the target outbound call time period for making outbound calls to each customer to be called outbound; and outbound calls are made to multiple customers to be called outbound based on the target outbound call time period corresponding to each customer to be called outbound. The present application takes into account both the personalized characteristics of the customer and the allocation of agent resources, and can improve the customer's willingness to connect and reasonably allocate agent resources, thereby improving the outbound call efficiency and marketing conversion rate in the field of financial technology. The present application can be applied to the insurance business of an insurance company, thereby reasonably allocating the customer resources of insurance salesmen and improving the customer's willingness to purchase insurance. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the description of the embodiments. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0019] Figure 1 A schematic diagram of the steps of an agent outbound calling method provided in an embodiment of the present application;

[0020] Figure 2 for Figure 1 Schematic diagram of the sub-step flow of the agent outbound call method in FIG;

[0021] Figure 3 A schematic diagram of the optimal outbound call time distribution data provided in an embodiment of the present application;

[0022] Figure 4 Another schematic diagram of the optimal outbound call time distribution data provided in the embodiment of the present application

[0023] Figure 5 A schematic block diagram of an agent outbound calling device provided in an embodiment of the present application;

[0024] Figure 6 for Figure 5 A schematic block diagram of a submodule of an agent outbound call device;

[0025] Figure 7 A schematic block diagram of the structure of a computer device provided in an embodiment of the present application.

[0026] The realization of the objectives, functional features and advantages of this application will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0027] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0028] The flowcharts shown in the accompanying drawings are illustrative only and do not necessarily include all content and operations / steps, nor do they necessarily need to be executed in the order described. For example, some operations / steps may be decomposed, combined, or partially merged, so the actual order of execution may vary depending on the actual situation. In addition, although the functional modules are divided in the device schematics, in some cases, the module division may be different from that shown in the device schematics.

[0029] Outbound calls are a key channel for product marketing services. In a telesales system, when a human agent confirms a user's interest in insurance, the insurance company creates an application form based on the conversation between the agent and the user and delivers it offline to the user. Once the user confirms the application form and pays the premium, the insurance system generates a corresponding policy and delivers it to the user, completing the entire insurance process.

[0030] Currently, insurance outbound call scenarios face low connection rates and uneven agent resource allocation. The industry currently addresses these issues by prioritizing outbound calls based on customer level and concentrating outbound calls during popular hours. However, these solutions present several challenges: customer priority doesn't reflect their willingness to answer the call, nor can it accurately determine their acceptance of outbound call times, leading to low customer willingness to purchase insurance. Furthermore, concentrating outbound calls during popular hours lacks personalized consideration for customers and places greater demands on agent resource allocation.

[0031] The present invention provides an agent outbound calling method, apparatus, device, and storage medium. The agent outbound calling method can be applied to a terminal device or server. The terminal device can be an electronic device such as a mobile phone, tablet computer, laptop computer, desktop computer, personal digital assistant, or wearable device. The server can be a single server or a server cluster consisting of multiple servers.

[0032] The embodiments of this application utilize artificial intelligence technology to acquire and process relevant data, while also taking into account customer personalization and agent resource allocation. This can enhance customer connection willingness, rationally allocate agent resources, and improve outbound call efficiency and marketing conversion rates. Artificial intelligence (AI) refers to the theories, methods, technologies, and application systems that utilize digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to achieve optimal results. This includes areas such as machine learning and deep learning.

[0033] The following describes some embodiments of the present application in detail with reference to the accompanying drawings. In the absence of conflict, the following embodiments and features therein may be combined with each other.

[0034] Please refer to Figure 1 , Figure 1 A schematic diagram of the steps of an agent outbound calling method provided in an embodiment of the present application.

[0035] like Figure 1 As shown, the agent outbound calling method includes steps S101 to S104.

[0036] Step S101: Acquire user feature data of a plurality of customers to be called outbound, and acquire agent data for making outbound calls to the plurality of customers to be called outbound.

[0037] User profile data can include product purchase data and historical call data. Historical call data can include outbound call connection information and connection duration, with the latter indicating whether the call was connected or not. Agent data includes data related to agents used to make outbound calls to multiple customers, such as the number of agents and agent roster information. Agent data can also include agent-related data over a period of time, such as the number of agents or agent roster information scheduled over multiple days or time periods.

[0038] Furthermore, user characteristic data may also include information used to characterize the personal attributes and social attributes of the outbound customers, such as the age, family members, interests and hobbies, terminal devices used (such as mobile phones, tablets, laptops, household appliances), and other related information, as well as information such as the city where they are located, occupation type, income status, etc.

[0039] In a telephone call system, any outbound call recipient must establish a communication connection with a corresponding call agent before the call can proceed. Call agents can be staff members providing business services in the outbound call service system, such as bank clerks, securities traders, insurance salespeople, course salespeople, and other staff who provide business consulting services.

[0040] For example, insurance services may include accident insurance, health insurance, and illness insurance. Agent data may include the agent's name, employee number, and contact information. In this embodiment, each type of insurance has one or more agents responsible for handling policies. Policies for different types of insurance can be handled by the same agent or different agents. For example, Agent A can handle accident and health insurance policies, while Agent B can handle accident and illness insurance policies.

[0041] In one embodiment, user feature data and agent data may be stored in the same memory or database, or in separate memories or databases. The memory in this embodiment may include a storage medium and internal memory, and the database in this embodiment refers to a database for storing user feature data or agent data for batches of outbound call recipients.

[0042] In one embodiment, the timing for acquiring user feature data and agent data can be flexibly determined based on actual circumstances. For example, user feature data for multiple customers waiting for outbound calls can be acquired first, followed by acquisition of agent data for agents making outbound calls to the multiple customers waiting for outbound calls. Alternatively, agent data for agents making outbound calls to the multiple customers waiting for outbound calls can be acquired first, followed by acquisition of user feature data for the multiple customers waiting for outbound calls. The timing for acquiring agent data for agents making outbound calls to the multiple customers waiting for outbound calls can be after step S102, but this embodiment does not impose any specific limitations on this.

[0043] It should be noted that in order to further ensure the privacy and security of the above-mentioned user feature data, seat data and other related information, the above-mentioned user feature data, seat data and other related information can also be stored in a blockchain node. The technical solution of this application can also be applied to adding other data files stored on the blockchain. The blockchain referred to in this application is a new application model of computer technologies such as distributed data storage, point-to-point transmission, consensus mechanism, and encryption algorithm.

[0044] Step S102: Input each user's characteristic data into a pre-trained outbound call time period prediction model for prediction, so as to determine the optimal outbound call time period for each customer to be called outbound.

[0045] The outbound call time prediction model is used to predict the optimal time for making outbound calls to each customer. This optimal time prediction is a multi-classification problem, and the outbound call time prediction model can be a multi-layer fully connected neural network. Alternatively, other prediction models designed for multi-classification problems, such as decision trees and recurrent convolutional neural networks, can be used.

[0046] In one embodiment, if Figure 2 As shown, step S102 includes: sub-step S1021 to sub-step S1022.

[0047] Sub-step S1021: Inputting the user feature data into a pre-trained outbound call time period prediction model for prediction to obtain connection probabilities for multiple outbound call time periods.

[0048] There can be multiple outbound call time periods for each prospective outbound caller. These time periods can be divided based on actual circumstances, for example, every two or three hours, or even every hour or half an hour. It should be noted that the connection probability refers to the probability that the user will connect with the outbound call, and the connection probability in different outbound call time periods can be the same or different. Predicting the outbound call time period based on the input user feature data using a pre-trained outbound call time period prediction model can accurately determine the connection probability for multiple outbound call time periods for the prospective outbound caller.

[0049] In one embodiment, the outbound call time period prediction model includes an input layer, a hidden layer, and an output layer; user feature data is input into a pre-trained outbound call time period prediction model for prediction to obtain connection probabilities for multiple outbound call time periods, including: inputting user feature data into the input layer for feature extraction to obtain first feature vector data; inputting the first feature vector data into the hidden layer for activation function operation to obtain second feature vector data; inputting the second feature vector data into the output layer for linear transformation to obtain connection probabilities for multiple outbound call time periods.

[0050] It should be noted that there can be multiple input layers, and the number of input layers can be correlated with the number of data types contained in the user feature data. Each node in the next input layer is connected to all nodes in the previous input layer, thereby integrating the extracted features to obtain the first feature vector data. The hidden layer can include multiple neurons, and activation functions are run on the neurons. The neurons in the hidden layer can perform activation function operations (nonlinear operations) on the input first feature vector data to obtain the second feature vector data. The output layer is used to perform a linear transformation on the second feature vector data to obtain the connection probabilities for multiple outbound call time periods. For example, the output layer can include a SoftMax layer, which outputs the connection probabilities for outbound calls in each time period. The sum of the connection probabilities for multiple outbound call time periods can be 1.

[0051] In one embodiment, a pre-trained outbound call time prediction model can be obtained by iteratively training using sample user feature data as training samples. For example, the outbound call time prediction model is iteratively trained using multiple sample user feature data until the outbound call time prediction model converges. The convergence conditions for the outbound call time prediction model can be set based on actual circumstances, such as when the number of iterations reaches a preset number, the iteration duration exceeds a preset duration, and the model loss value is less than a preset loss value. This embodiment does not impose specific limitations on these conditions.

[0052] Sub-step S1022: Determine the outbound call time period corresponding to the maximum connection probability from the multiple outbound call time periods as the optimal outbound call time period.

[0053] Among them, the connection probabilities of multiple outbound call time periods can be different. The maximum connection probability is determined from the connection probabilities of multiple outbound call time periods, and the outbound call time period corresponding to the maximum connection probability is used as the optimal outbound call time period, thereby increasing customers' willingness to connect and improving marketing conversion rate.

[0054] In another embodiment, user feature data is input into a pre-trained outbound call time period prediction model for prediction. After obtaining the connection probabilities of multiple outbound call time periods, the optimal outbound call time period is determined from the multiple outbound call time periods based on the connection probabilities of the multiple outbound call time periods. The optimal outbound call time period may be composed of one or more outbound call time periods. When the optimal outbound call time period includes multiple outbound call time periods, the multiple outbound call time periods may be continuous or discontinuous. Alternatively, the optimal outbound call time period may be composed of partial time periods of at least two outbound call time periods, such as the optimal outbound call time period includes partial time periods of two adjacent outbound call time periods, and the partial time periods of the two adjacent outbound call time periods may be equal or unequal in duration. This embodiment does not specifically limit this.

[0055] In one embodiment, user feature data is input into a pre-trained outbound call time prediction model to predict the optimal outbound call time for each customer to be called. It should be noted that the outbound call time prediction model in this embodiment can directly determine the optimal outbound call time based on the input user feature data.

[0056] Exemplarily, the outbound call time prediction model includes an input layer, a hidden layer, an output layer, and a screening layer. User feature data is input into the input layer for feature extraction to obtain first feature vector data. The first feature vector data is input into the hidden layer for activation function operation to obtain second feature vector data. The second feature vector data is input into the output layer for linear transformation to obtain connection probabilities for multiple outbound call time periods. The connection probabilities for multiple outbound call time periods are input into the screening layer for selection to determine the optimal outbound call time period from the multiple outbound call time periods.

[0057] Step S103: Input the agent data and the multiple optimal outbound call time periods into a pre-trained agent allocation model for processing to determine a target outbound call time period for each customer to be called outbound.

[0058] Seat data includes seat quantity information and / or seat list information. Seat data may also include time information, which is used to characterize the time attributes of the seat quantity information or seat list information. For example, seat data may include seat list information for a week, or seat quantity information for multiple time periods within a 24-hour period.

[0059] The agent allocation model is used to adjust and allocate the optimal outbound call time slots for multiple potential outbound call customers based on agent data, thereby determining the target outbound call time slot for each potential outbound call customer. Adjusting and allocating the optimal outbound call time slots is a fine-tuning problem, so the agent allocation model can be a single-layer fully connected neural network, or other models designed for fine-tuning.

[0060] In one embodiment, before inputting the agent data and multiple optimal outbound call time periods into a pre-trained agent allocation model for processing to determine a target outbound call time period for each outbound call to a customer to be called, the process further includes: splicing the agent data and the multiple optimal outbound call time periods according to a preset format, wherein the preset format can be flexibly set according to actual conditions; and inputting the spliced ​​agent data and the multiple optimal outbound call time periods into the pre-trained agent allocation model for processing, thereby improving the stability of the parameters input into the agent allocation model and ensuring the effectiveness of the model.

[0061] In one embodiment, the agent allocation model includes a statistical layer, a scheduling layer, and a mapping layer; the optimal outbound call time periods of multiple customers to be called out are input into the statistical layer for statistics to obtain optimal outbound call distribution data for different time periods; the agent data and the optimal outbound call distribution data are input into the scheduling layer to schedule the optimal outbound call time periods of the customers to be called out in the optimal outbound call distribution data to obtain updated outbound call time period distribution data; the updated outbound call time period distribution data is input into the mapping layer for mapping to obtain a target outbound call time period corresponding to each customer to be called out.

[0062] It should be noted that the statistical layer in the agent allocation model can calculate the optimal outbound call time periods for multiple prospective outbound call customers within different time periods, thereby generating optimal outbound call distribution data. For example, the statistical layer can aggregate the optimal outbound call time periods for prospective outbound call customers within the same time period and output optimal outbound call time period distribution data. The scheduling layer in the agent allocation model then schedules this optimal outbound call time period distribution data based on agent data. This scheduling can be balanced scheduling, i.e., evenly distributing the optimal outbound call time periods for prospective outbound call customers across different time periods or dates based on agent data. This generates updated outbound call time period distribution data, thereby rationally allocating agent resources and improving outbound call efficiency. The mapping layer outputs this updated outbound call time period distribution data as the target outbound call time period for each prospective outbound call customer. The target outbound call time period for each prospective outbound call customer can be obtained by adjusting the optimal outbound call time period. For example, if the optimal outbound call time period for a prospective outbound call customer is 7:00 PM to 8:00 PM, then after processing by the scheduling and mapping layers, the target outbound call time period for the prospective outbound call customer is output as 6:00 PM to 7:00 PM.

[0063] In one embodiment, a pre-trained agent allocation model can be obtained by iteratively training using sample agent data and multiple optimal outbound call time periods as training samples. For example, the agent allocation model is iteratively trained using multiple sample agent data and multiple optimal outbound call time periods until the agent allocation model converges. The convergence conditions for the agent allocation model can be set based on actual circumstances, such as when the number of iterations reaches a preset number, the iteration duration exceeds a preset duration, or the model loss value is less than a preset loss value. This embodiment does not impose specific limitations on these conditions.

[0064] In one embodiment, the agent data includes agent quantity information. The optimal outbound call time slot distribution data is evenly scheduled based on the agent quantity information, thereby evenly allocating the optimal outbound call time slots for the customers to be called. This prevents excessive outbound call demand during popular time slots, resulting in updated outbound call time slot distribution data. This allows for a more rational allocation of agent resources and improved outbound call efficiency.

[0065] For example, Figure 3 As shown, Figure 3 Figure 1 shows the optimal outbound call time distribution data. The agent data includes information about the number of agents, 5. The optimal outbound call time distribution data shows that the optimal outbound call time period is the second outbound call time period (B), with 150 customers waiting to be called; the optimal outbound call time period is the first outbound call time period (A), with 50 customers waiting to be called; and the optimal outbound call time period is the third outbound call time period (C), with 10 customers waiting to be called. The maximum number of outbound calls per agent per hour is 20, and the minimum number of outbound calls per agent per hour is 5. The balanced scheduling strategy ensures that the number of customers waiting to be called during the target outbound call time period does not exceed the maximum number of outbound calls, and the number of customers waiting to be called during the target outbound call time period does not fall below the minimum number of outbound calls.

[0066] Based on this, after the agent allocation model evenly schedules multiple optimal outbound call time periods based on the agent quantity information, the number of customers waiting for outbound calls in the target outbound call time period B is 100, the number of customers waiting for outbound calls in the target outbound call time period A is 85, and the number of customers waiting for outbound calls in the target outbound call time period C is 25. This application can be applied to financial technology fields such as insurance sales services, financial services, and wealth management services, and can also be applied to services such as real estate sales and event invitations. By evenly scheduling multiple optimal outbound call time periods, the task load of agents in different time periods can be balanced, thereby improving agent work efficiency and outbound call efficiency, which is conducive to improving marketing conversion rate.

[0067] In one embodiment, the agent data includes information about the number of agents at different time periods. This information represents the number of agents at different working hours. Based on this information, the optimal outbound call time slots for the outbound customer are evenly distributed across different time periods, resulting in updated outbound call time slot distribution data. This allows for more efficient allocation of agent resources and improved outbound call efficiency.

[0068] For example, Figure 4As shown, the number of agents for different time periods is 5 from 6:00 PM to 7:00 PM (Segment A), 3 from 7:00 PM to 8:00 PM (Segment B), and 2 from 8:00 PM to 9:00 PM (Segment C). The optimal outbound calling time period of 6:00 PM to 7:00 PM (Segment A) has 150 customers waiting for outbound calls, 50 from 7:00 PM to 8:00 PM (Segment B), and 10 from 8:00 PM to 9:00 PM (Segment C). The balanced scheduling formula can be: the number of customers waiting for outbound calls during the target outbound calling time period, K1, = the total number of customers waiting for outbound calls, K, * the number of agents for the corresponding time period, M1, / the total number of agents across multiple time periods, M.

[0069] Based on this, after balancing multiple optimal outbound call time periods using the agent allocation model based on the number of agents in different time periods, the target outbound call time period of 6:00 PM to 7:00 PM (Segment A) has 105 customers waiting for outbound calls, the target outbound call time period of 7:00 PM to 8:00 PM (Segment B) has 63 customers waiting for outbound calls, and the target outbound call time period of 8:00 PM to 9:00 PM (Segment C) has 42 customers waiting for outbound calls. This application can be applied to financial technology fields such as insurance sales services, financial services, and wealth management services, for example, in outbound call scenarios in application fields such as stock trading, fund agency, securities trading, bank card processing, and insurance renewal.

[0070] In one embodiment, the agent data includes agent list data. The optimal outbound call time slot distribution data is evenly scheduled based on the agent list data, thereby evenly allocating the optimal outbound call time slots for the customers to be called. This prevents significant discrepancies in the number of customers assigned to different agents, thereby rationally allocating agent resources and improving outbound call efficiency. For example, within the same outbound call time slot, customers to be called are evenly allocated to Zhang San and Li Si in the agent list data.

[0071] In one embodiment, the agent data includes agent list data for different dates. The optimal outbound call time period distribution data is evenly scheduled according to the agent list data for different dates, thereby evenly arranging the optimal outbound call time periods for customers to be called on different dates, avoiding excessive differences in the number of customers to be called allocated on different dates, thereby rationally allocating agent resources and improving outbound call efficiency.

[0072] In one embodiment, the training process of the outbound call time period prediction model and the agent allocation model includes: obtaining sample feature data and sample agent data of multiple outbound call customers as training samples; training the preset outbound call time period prediction model based on the multiple sample feature data to determine the optimal outbound call time period for making outbound calls to each outbound call customer and the sample connection probability corresponding to each optimal outbound call time period; training the preset agent allocation model based on the multiple optimal outbound call time periods and sample agent data to determine the outbound call volume in different time periods; calculating the target loss function based on the sample connection probability corresponding to each optimal outbound call time period and the sample outbound call volume in different time periods; and updating the model parameters of the outbound call time period prediction model and the agent allocation model based on the target loss function until the outbound call time period prediction model and the agent allocation model converge.

[0073] The outbound call time prediction model can be a multi-layer fully connected neural network, and the agent allocation model can be a single-layer fully connected neural network. It should be noted that the embodiments of the present application optimize the neural network structure and incorporate agent resource allocation constraints into the network structure layer to improve prediction accuracy and effectiveness, thereby increasing the customer's time period connection rate while ensuring a balanced agent resource balance.

[0074] In one embodiment, a target loss function is calculated based on the sample connection probability corresponding to each optimal outbound call time period and the sample outbound call volume in different time periods, including: obtaining the actual connection probability of each optimal outbound call time period, and generating a first loss function based on the product of the sample connection probability and the actual connection probability corresponding to each optimal outbound call time period; obtaining the actual outbound call volume in different time periods, and generating a second loss function based on the deviation between the actual outbound call volume and the sample outbound call volume in different time periods; and determining the target loss function based on the first loss function and the second loss function.

[0075] Among them, the target loss function consists of two parts. The first part is the difference between the optimal time period predicted by the individual and the actual answering time period; the second part is the deviation between the predicted outbound call volume in each time period and the agent resources in each time period. It should be noted that the conventional neural network target optimization function only fits the outbound call probability with the actual answering, while the target loss function of the embodiment of the present application also needs to consider the deviation fitting between the predicted total outbound call volume in each time period and the agent resource allocation. Deviation fitting refers to minimizing the loss function through multiple rounds of data calculation iterations. The goal is to ensure that the total number of outbound calls in each time period is consistent with the allocated agent resources, so as to avoid the problem of unbalanced resource allocation.

[0076] In one embodiment, the objective loss function is ; is the first loss function, i.e. the loss function of the target outbound call probability and the actual answering probability, is the second loss function, i.e., the deviation fitting between the total outbound call volume predicted in each period and the agent resource allocation. is the regularization term, a and b are the adjustment coefficients of the two loss functions, which can be set manually.

[0077] It should be noted that the weight coefficients a and b in the target loss function are fixed as constants after the initial setting and do not need to be optimized; the sizes of a and b represent the weights of connection prediction accuracy and agent resource optimization allocation, respectively; if a is set to be larger, the connection prediction accuracy will be strengthened in the optimized loss function, and vice versa, the agent resource allocation will be strengthened. The embodiment of the present application optimizes the target loss function, adds the tuning of resource allocation deviation to the traditional prediction result loss optimization, adjusts the weights of the two loss functions by coefficients, and outputs the predicted optimal outbound call time period and agent outbound call resource allocation at the same time, avoiding the agent resource allocation problem caused by the results of the traditional prediction model, improving the customer connection rate, and overall the outbound call time period prediction based on the improved neural network is more stable and accurate.

[0078] In one embodiment, the model parameters of the outbound call time period prediction model and the agent allocation model are updated based on the target loss function, including: updating the model parameters of the agent allocation model based on the target loss function to obtain a new agent allocation model; training the new agent allocation model based on multiple optimal outbound call time periods and sample agent data to determine new outbound call volumes in different time periods; calculating a candidate loss function based on the sample connection probability corresponding to each optimal outbound call time period and the new sample outbound call volumes in different time periods; updating the model parameters of the outbound call time period prediction model based on the candidate loss function to obtain a new outbound call time period prediction model.

[0079] It should be noted that the model continuously trains samples, optimizes the target loss function, and ultimately outputs a target outbound call time period for each customer. The target loss function is used to update the model parameters of the agent allocation model to obtain a new agent allocation model. This new agent allocation model is then used to calculate a candidate loss function. Based on this candidate loss function, the model parameters of the outbound call time period prediction model are updated to obtain a new outbound call time period prediction model. This updates the agent allocation model and the outbound call time period prediction model, improving model training results and, consequently, the accuracy of outbound call time period predictions and the effectiveness of agent resource allocation.

[0080] Step S104: making outbound calls to multiple customers to be called based on the target outbound calling time period corresponding to each customer to be called.

[0081] It should be noted that after determining a target outbound calling time period for each customer to be called, outbound calls are made to multiple customers based on their corresponding target outbound calling time period. In other words, during different working hours, agents can sequentially call customers on the list of customers to be called during the target outbound calling time period, thereby improving call connection rates and, in turn, agent outbound calling efficiency.

[0082] For example, taking the example of telephone renewal in the insurance field, insurance sales personnel can serve as outbound call agents. Through the outbound call time prediction model, they can accurately determine the optimal outbound call time for each outbound customer. Through the agent allocation model, they can accurately adjust the target outbound call time corresponding to each outbound customer, so that they can use the target outbound call time to call the corresponding outbound customers. They can consider the customer's willingness to connect and reasonably allocate agent resources, greatly improving outbound call efficiency and marketing conversion rate.

[0083] For example, when a human agent reaches an insurance agreement with a user through a telephone sales system, the insurance system obtains the insurance agreement reached between the human agent and the user and generates an insurance application based on the insurance agreement. It should be noted that during the process of the human agent and the user reaching an insurance agreement through the telephone sales system, the human agent will enter the user's insurance information, the insured's information, the beneficiary's information, and other information into the insurance system to form an insurance agreement. The insurance application, also known as an insurance application form or insurance proposal, is a written offer from the insured to the insurance company to enter into an insurance contract. The insurance application includes information such as the name and address of the insured, the name and storage location of the subject matter of insurance, the type of insurance, the start and end dates of the insurance liability, the insurance value, and the insurance amount.

[0084] Once the insurance system confirms that the user has paid the full premium, it generates an insurance policy corresponding to the application. An insurance policy, also known as an insurance policy, must clearly and completely describe the rights and obligations of both parties involved in the insurance. It primarily contains the names of the insurer and the insured, the subject matter of insurance, the insured amount, the premium, the insurance term, the scope of liability for compensation or payment, and other specified matters. In this embodiment, once the user pays the premium, the insurance system generates the application online. Once the insurance system generates the insurance policy corresponding to the application, it sends it to the user.

[0085] The agent outbound calling method provided in the above embodiment obtains user characteristic data of multiple customers to be called outbound, and obtains agent data for making outbound calls to multiple customers to be called outbound; inputs each user characteristic data into a pre-trained outbound calling time period prediction model for prediction to determine the optimal outbound calling time period for making outbound calls to each customer to be called outbound; inputs the agent data and multiple optimal outbound calling time periods into a pre-trained agent allocation model for processing to determine the target outbound calling time period for making outbound calls to each customer to be called outbound; and makes outbound calls to multiple customers to be called outbound based on the target outbound calling time period corresponding to each customer to be called outbound. This application takes into account both the personalized characteristics of customers and the allocation of agent resources, and can improve customers' willingness to connect and reasonably allocate agent resources, thereby improving outbound call efficiency and marketing conversion rate.

[0086] The agent outbound calling method provided in the embodiment of the present application can be applied to the insurance business of an insurance company, thereby rationally allocating customer resources of insurance salesmen and increasing customers' willingness to purchase insurance.

[0087] Please refer to Figure 5 , Figure 5 A schematic block diagram of an agent outbound calling device provided in an embodiment of the present application.

[0088] like Figure 5 As shown, the agent outbound calling device 200 includes:

[0089] An acquisition module 201 is configured to acquire user characteristic data of a plurality of customers to be called outbound, and acquire agent data for making outbound calls to the plurality of customers to be called outbound;

[0090] Prediction module 202, configured to input each user feature data into a pre-trained outbound call time prediction model to perform prediction, so as to determine the optimal outbound call time for each of the outbound call-waiting customers;

[0091] The processing module 203 is configured to input the agent data and the plurality of optimal outbound call time periods into a pre-trained agent allocation model for processing, so as to determine a target outbound call time period for making outbound calls to each of the customers to be called outbound;

[0092] The outbound calling module 204 is configured to make outbound calls to a plurality of the outbound calling customers based on the target outbound calling time period corresponding to each of the outbound calling customers.

[0093] In one embodiment, Figure 6 As shown, the prediction module 202 includes:

[0094] Prediction submodule 2021, configured to input the user feature data into a pre-trained outbound call time period prediction model to perform prediction and obtain connection probabilities for multiple outbound call time periods;

[0095] The determination submodule 2022 is configured to determine, from the multiple outbound call time periods, an outbound call time period corresponding to the maximum connection probability as the optimal outbound call time period.

[0096] In one embodiment, the outbound call time period prediction model includes an input layer, a hidden layer, and an output layer; the prediction module 202 is further configured to:

[0097] Inputting the user feature data into the input layer for feature extraction to obtain first feature vector data;

[0098] Inputting the first feature vector data into the hidden layer to perform activation function operation to obtain second feature vector data;

[0099] The second feature vector data is input into the output layer for linear transformation to obtain connection probabilities of multiple outbound call time periods.

[0100] In one embodiment, the seat allocation model includes a statistical layer, a scheduling layer, and a mapping layer; the processing module 203 is further used to:

[0101] Inputting the optimal outbound calling time periods of the plurality of customers to be called outbound into the statistical layer for statistical analysis to obtain optimal outbound calling distribution data for different time periods;

[0102] Inputting the seat data and the optimal outbound call distribution data into the scheduling layer to schedule the optimal outbound call time period of the customer to be called in the optimal outbound call distribution data to obtain updated outbound call time period distribution data;

[0103] The updated outbound call time period distribution data is input into the mapping layer for mapping to obtain the target outbound call time period corresponding to each of the customers to be called outbound.

[0104] In one embodiment, the agent outbound calling device 200 further includes a training module, which is used to:

[0105] Obtaining sample feature data and sample agent data of multiple outbound customers as training samples;

[0106] Training a preset outbound call time prediction model based on the plurality of sample feature data to determine an optimal outbound call time for each of the outbound call customers and a sample connection probability corresponding to each of the optimal outbound call time periods;

[0107] Training a preset agent allocation model based on the plurality of optimal outbound call time periods and the sample agent data to determine the outbound call volume in different time periods;

[0108] Calculating a target loss function based on the sample connection probability corresponding to each of the optimal outbound call time periods and the sample outbound call volume in different time periods;

[0109] Model parameters of the outbound call time period prediction model and the seat allocation model are updated based on the target loss function until the outbound call time period prediction model and the seat allocation model converge.

[0110] In one embodiment, the training module is further configured to:

[0111] Obtaining a true connection probability for each of the optimal outbound call time periods, and generating a first loss function based on the product of the sample connection probability and the true connection probability corresponding to each of the optimal outbound call time periods;

[0112] Obtain the actual outbound call volume in different time periods, and generate a second loss function based on the deviation between the actual outbound call volume and the sample outbound call volume in different time periods;

[0113] A target loss function is determined based on the first loss function and the second loss function.

[0114] In one embodiment, the training module is further configured to:

[0115] updating the model parameters of the seat allocation model based on the target loss function to obtain a new seat allocation model;

[0116] Training a new agent allocation model based on the multiple optimal outbound call time periods and the sample agent data to determine new outbound call volumes in different time periods;

[0117] Calculating a candidate loss function based on the sample connection probability corresponding to each of the optimal outbound call time periods and the new sample outbound call volume in different time periods;

[0118] The model parameters of the outbound call time period prediction model are updated based on the candidate loss function to obtain a new outbound call time period prediction model.

[0119] It should be noted that those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the above-described device and each module and unit can refer to the corresponding processes in the aforementioned embodiment of the agent outbound call method, and will not be repeated here.

[0120] The apparatus provided in the above embodiment can be implemented in the form of a computer program. The computer program can be used in Figure 7 Runs on the computer equipment shown.

[0121] See also Figure 7 , Figure 7 This is a schematic block diagram of the structure of a computer device provided in an embodiment of the present application. The computer device can be a server or a terminal device.

[0122] like Figure 7 As shown, the computer device includes a processor, a memory and a network interface connected through a system bus, wherein the memory may include a storage medium and an internal memory, and the storage medium may be non-volatile or volatile.

[0123] The storage medium can store an operating system and a computer program. The computer program includes program instructions, and when the program instructions are executed, the processor can execute any one of the agent outbound calling methods.

[0124] The processor is used to provide computing and control capabilities and support the operation of the entire computer equipment.

[0125] The internal memory provides an environment for the operation of the computer program in the storage medium. When the computer program is executed by the processor, the processor can execute any agent outbound calling method.

[0126] The network interface is used for network communication, such as sending assigned tasks, etc. Those skilled in the art will understand that Figure 7 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0127] It should be understood that the processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc.

[0128] In one embodiment, the processor is configured to execute a computer program stored in the memory to implement the following steps:

[0129] Acquire user feature data of a plurality of customers to be called outbound, and acquire agent data for making outbound calls to the plurality of customers to be called outbound;

[0130] Inputting each of the user feature data into a pre-trained outbound call time period prediction model for prediction, so as to determine the optimal outbound call time period for each of the outbound call customers;

[0131] Inputting the agent data and the plurality of optimal outbound call time periods into a pre-trained agent allocation model for processing to determine a target outbound call time period for making outbound calls to each of the customers to be called;

[0132] Based on the target outbound calling time period corresponding to each of the outbound calling customers, outbound calls are made to a plurality of the outbound calling customers.

[0133] In one embodiment, when inputting each user feature data into a pre-trained outbound call time period prediction model for prediction to determine the optimal outbound call time period for each of the waiting outbound call customers, the processor is configured to implement:

[0134] Inputting the user feature data into a pre-trained outbound call time period prediction model to perform predictions to obtain connection probabilities for multiple outbound call time periods;

[0135] An outbound call time period corresponding to the maximum connection probability is determined from the multiple outbound call time periods as the optimal outbound call time period.

[0136] In one embodiment, the outbound call time prediction model includes an input layer, a hidden layer, and an output layer; when the processor inputs the user feature data into a pre-trained outbound call time prediction model for prediction to obtain connection probabilities for multiple outbound call time periods, it is configured to implement:

[0137] Inputting the user feature data into the input layer for feature extraction to obtain first feature vector data;

[0138] Inputting the first feature vector data into the hidden layer to perform activation function operation to obtain second feature vector data;

[0139] The second feature vector data is input into the output layer for linear transformation to obtain connection probabilities of multiple outbound call time periods.

[0140] In one embodiment, the agent allocation model includes a statistical layer, a scheduling layer, and a mapping layer; when the processor inputs the agent data and the plurality of optimal outbound call time periods into a pre-trained agent allocation model for processing to determine a target outbound call time period for each of the waiting outbound call customers, the processor is configured to:

[0141] Inputting the optimal outbound calling time periods of the plurality of customers to be called outbound into the statistical layer for statistical analysis to obtain optimal outbound calling distribution data for different time periods;

[0142] Inputting the seat data and the optimal outbound call distribution data into the scheduling layer to schedule the optimal outbound call time period of the customer to be called in the optimal outbound call distribution data to obtain updated outbound call time period distribution data;

[0143] The updated outbound call time period distribution data is input into the mapping layer for mapping to obtain the target outbound call time period corresponding to each of the customers to be called outbound.

[0144] In one embodiment, when implementing the training process of the outbound call time period prediction model and the agent allocation model, the processor is configured to implement:

[0145] Obtaining sample feature data and sample agent data of multiple outbound customers as training samples;

[0146] Training a preset outbound call time prediction model based on the plurality of sample feature data to determine an optimal outbound call time for each of the outbound call customers and a sample connection probability corresponding to each of the optimal outbound call time periods;

[0147] Training a preset agent allocation model based on the plurality of optimal outbound call time periods and the sample agent data to determine the outbound call volume in different time periods;

[0148] Calculating a target loss function based on the sample connection probability corresponding to each of the optimal outbound call time periods and the sample outbound call volume in different time periods;

[0149] Model parameters of the outbound call time period prediction model and the seat allocation model are updated based on the target loss function until the outbound call time period prediction model and the seat allocation model converge.

[0150] In one embodiment, when calculating the target loss function based on the sample connection probability corresponding to each optimal outbound call time period and the sample outbound call volume in different time periods, the processor is configured to implement:

[0151] Obtaining a true connection probability for each of the optimal outbound call time periods, and generating a first loss function based on the product of the sample connection probability and the true connection probability corresponding to each of the optimal outbound call time periods;

[0152] Obtain the actual outbound call volume in different time periods, and generate a second loss function based on the deviation between the actual outbound call volume and the sample outbound call volume in different time periods;

[0153] A target loss function is determined based on the first loss function and the second loss function.

[0154] In one embodiment, when the processor updates the model parameters of the outbound call time period prediction model and the agent allocation model based on the target loss function, it is configured to implement:

[0155] updating the model parameters of the seat allocation model based on the target loss function to obtain a new seat allocation model;

[0156] Training a new agent allocation model based on the multiple optimal outbound call time periods and the sample agent data to determine new outbound call volumes in different time periods;

[0157] Calculating a candidate loss function based on the sample connection probability corresponding to each of the optimal outbound call time periods and the new sample outbound call volume in different time periods;

[0158] The model parameters of the outbound call time period prediction model are updated based on the candidate loss function to obtain a new outbound call time period prediction model.

[0159] It should be noted that those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the computer device described above can refer to the corresponding process in the aforementioned embodiment of the agent outbound calling method, and will not be repeated here.

[0160] The present application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, and the like. The present application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, and the like that perform specific tasks or implement specific abstract data types. The present application can also be practiced in distributed computing environments in which tasks are performed by remote processing devices connected via a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media, including storage devices.

[0161] An embodiment of the present application further provides a computer-readable storage medium, on which a computer program is stored. The computer program includes program instructions. The method implemented when the program instructions are executed can refer to the various embodiments of the agent outbound calling method of the present application.

[0162] The computer-readable storage medium may be an internal storage unit of the computer device described in the aforementioned embodiment, such as a hard disk or memory of the computer device. The computer-readable storage medium may also be an external storage device of the computer device, such as a plug-in hard disk, a SmartMedia Card (SMC), a Secure Digital (SD) card, a flash memory card, etc., equipped on the computer device.

[0163] Furthermore, the computer-usable storage medium may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function, etc.; the data storage area may store data created based on the use of blockchain nodes, etc. The blockchain referred to in this application is a new application model of computer technologies such as distributed data storage, point-to-point transmission, consensus mechanism, and encryption algorithm. Blockchain is essentially a decentralized database, a string of data blocks generated by cryptographic methods. Each data block contains information about a batch of network transactions, which is used to verify the validity of its information (anti-counterfeiting) and generate the next block. The blockchain may include the blockchain underlying platform, the platform product service layer, and the application service layer.

[0164] It should be understood that the terms used in this specification are only for the purpose of describing specific embodiments and are not intended to limit the present application. As used in this specification and the appended claims, the singular forms "a", "an", and "the" are intended to include the plural forms unless the context clearly indicates otherwise.

[0165] It should also be understood that the term "and / or" used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, including these combinations. It should be noted that, in this article, the terms "include", "comprise" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or system that includes a series of elements includes not only those elements, but also other elements that are not explicitly listed, or also includes elements that are inherent to such process, method, article or system. In the absence of further limitations, an element defined by the sentence "including a..." does not exclude the presence of other identical elements in the process, method, article or system that includes the element.

[0166] The serial numbers of the embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments. The above description is only a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in this application, and these modifications or replacements should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. A seat outbound calling method, characterized in that: include: Acquire user feature data of a plurality of customers to be called outbound, and acquire agent data for making outbound calls to the plurality of customers to be called outbound; Inputting each of the user feature data into a pre-trained outbound call time period prediction model for prediction, so as to determine the optimal outbound call time period for each of the outbound call customers; Inputting the agent data and the plurality of optimal outbound call time periods into a pre-trained agent allocation model for processing to determine a target outbound call time period for making outbound calls to each of the customers to be called; making outbound calls to a plurality of the outbound call customers based on the target outbound call time period corresponding to each of the outbound call customers; Among them, the outbound call time period prediction model and the seat allocation model are obtained by iterative training based on the target loss function, and the target loss function includes a first target loss function and a second target loss function. The first target loss function is used to characterize the difference between the optimal outbound call time period predicted by an individual and the actual answering time period, and the second target loss function is used to characterize the deviation between the predicted outbound call volume in each time period and the seat resources in each time period.

2. The agent outbound calling method according to claim 1, wherein: The step of inputting each user feature data into a pre-trained outbound call time period prediction model to perform prediction to determine the optimal outbound call time period for each of the outbound call-receiving customers includes: Inputting the user feature data into a pre-trained outbound call time period prediction model to perform predictions to obtain connection probabilities for multiple outbound call time periods; An outbound call time period corresponding to the maximum connection probability is determined from the multiple outbound call time periods as the optimal outbound call time period.

3. The agent outbound calling method according to claim 2, wherein: The outbound call time prediction model includes an input layer, a hidden layer, and an output layer. The user feature data is input into a pre-trained outbound call time prediction model for prediction to obtain connection probabilities for multiple outbound call time periods, including: Inputting the user feature data into the input layer for feature extraction to obtain first feature vector data; Inputting the first feature vector data into the hidden layer to perform activation function operation to obtain second feature vector data; The second feature vector data is input into the output layer for linear transformation to obtain connection probabilities of multiple outbound call time periods.

4. The agent outbound calling method according to claim 1, wherein: The agent allocation model includes a statistical layer, a scheduling layer, and a mapping layer; the agent data and the plurality of optimal outbound call time periods are input into a pre-trained agent allocation model for processing to determine a target outbound call time period for each of the outbound call-receiving customers, including: Inputting the optimal outbound calling time periods of the plurality of customers to be called outbound into the statistical layer for statistical analysis to obtain optimal outbound calling distribution data for different time periods; Inputting the seat data and the optimal outbound call distribution data into the scheduling layer to schedule the optimal outbound call time period of the customer to be called in the optimal outbound call distribution data to obtain updated outbound call time period distribution data; The updated outbound call time period distribution data is input into the mapping layer for mapping to obtain the target outbound call time period corresponding to each of the customers to be called outbound.

5. The agent outbound calling method according to any one of claims 1 to 4, characterized in that: The training process of the outbound call time prediction model and the agent allocation model includes: Obtaining sample feature data and sample agent data of multiple outbound customers as training samples; Training a preset outbound call time prediction model based on the plurality of sample feature data to determine an optimal outbound call time for each of the outbound call customers and a sample connection probability corresponding to each of the optimal outbound call time periods; Training a preset agent allocation model based on the plurality of optimal outbound call time periods and the sample agent data to determine the outbound call volume in different time periods; Calculating a target loss function based on the sample connection probability corresponding to each of the optimal outbound call time periods and the sample outbound call volume in different time periods; Model parameters of the outbound call time period prediction model and the seat allocation model are updated based on the target loss function until the outbound call time period prediction model and the seat allocation model converge.

6. The agent outbound calling method according to claim 5, characterized in that: The target loss function is calculated based on the sample connection probability corresponding to each optimal outbound call time period and the sample outbound call volume in different time periods, including: Obtaining a true connection probability for each of the optimal outbound call time periods, and generating a first loss function based on the product of the sample connection probability and the true connection probability corresponding to each of the optimal outbound call time periods; Obtain the actual outbound call volume in different time periods, and generate a second loss function based on the deviation between the actual outbound call volume and the sample outbound call volume in different time periods; A target loss function is determined based on the first loss function and the second loss function.

7. The agent outbound calling method according to claim 5, characterized in that: The updating of the model parameters of the outbound call time period prediction model and the seat allocation model based on the target loss function includes: updating the model parameters of the seat allocation model based on the target loss function to obtain a new seat allocation model; Training a new agent allocation model based on the multiple optimal outbound call time periods and the sample agent data to determine new outbound call volumes in different time periods; Calculating a candidate loss function based on the sample connection probability corresponding to each of the optimal outbound call time periods and the new sample outbound call volume in different time periods; The model parameters of the outbound call time period prediction model are updated based on the candidate loss function to obtain a new outbound call time period prediction model.

8. A seat outbound calling device, characterized in that: The seat outbound calling device includes: An acquisition module, configured to acquire user characteristic data of a plurality of customers to be called outbound, and acquire seat data for making outbound calls to the plurality of customers to be called outbound; A prediction module, configured to input each of the user feature data into a pre-trained outbound call time period prediction model for prediction, so as to determine the optimal outbound call time period for each of the outbound call-waiting customers; a processing module, configured to input the agent data and the plurality of optimal outbound call time periods into a pre-trained agent allocation model for processing, so as to determine a target outbound call time period for making an outbound call to each of the customers to be called; An outbound calling module, configured to make outbound calls to a plurality of the outbound calling customers based on the target outbound calling time period corresponding to each of the outbound calling customers; Among them, the outbound call time period prediction model and the seat allocation model are obtained by iterative training based on the target loss function, and the target loss function includes a first target loss function and a second target loss function. The first target loss function is used to characterize the difference between the optimal time period predicted by an individual and the actual answering time period, and the second target loss function is used to characterize the deviation between the predicted outbound call volume in each time period and the seat resources in each time period.

9. A computer device, characterized in that: The computer device includes a processor, a memory, and a computer program stored in the memory and executable by the processor, wherein when the computer program is executed by the processor, the steps of the agent outbound calling method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, wherein when the computer program is executed by a processor, the steps of the agent outbound calling method according to any one of claims 1 to 7 are implemented.

Images