Predictive call-out method and device, equipment and storage medium

Abstract

The embodiment of the invention discloses a predictive outbound method and device, equipment and a storage medium, and the method comprises the steps: obtaining historical telephone traffic data at preset intervals, inputting the historical telephone traffic data into a target long-short-term memory network, and obtaining a corresponding target outbound number of the next outbound period; based on the target call-out number, executing call-out operation on clients in the list to be called out, and determining a corresponding call-out result; and under the condition that the outbound call result is that answering is agreed, performing voice dialogue operation with the answering customer based on the target large language model. According to the technical scheme provided by the embodiment of the invention, the problems of low prediction precision and high call loss rate in the existing predictive call-out technology are solved, the call-out initiation quantity can be accurately predicted by combining the time sequence prediction capability of the long-short-term memory network, and the accuracy and efficiency of predictive call-out are improved; and flexible and natural voice interaction with clients is performed based on the intelligent interaction capability of the large model, so that the participation degree and satisfaction degree of the clients are improved, and the call loss rate is reduced.

Description

Technical Field

[0001] Embodiments of the present invention relate to the technical field of banking services, and in particular to a predictive outbound calling method, apparatus, device, and storage medium. Background Art

[0002] Predictive outbound calling is an outbound calling technology based on data analysis and intelligent algorithms. It dynamically predicts the number of outbound calls to be made based on historical connection rates, agent idle time, and call duration, allowing it to execute targeted calls and improve agent utilization. Compared to traditional outbound calling, predictive outbound calling can adjust call speed and allocation strategies based on historical data and real-time information, ensuring a good match between agents and customers.

[0003] While predictive outbound calling has significantly improved outbound call efficiency, existing predictive outbound calling systems still have shortcomings in algorithmic flexibility and intelligent interaction capabilities. Most current predictive outbound calling systems use traditional algorithms such as linear formulas or linear regression. While these algorithms can capture basic outbound call patterns, their prediction accuracy is limited when dealing with complex outbound call environments. Furthermore, due to a lack of dynamic adjustment capabilities, traditional predictive models cannot quickly adapt to different business scenarios and are therefore inflexible. Regarding intelligent interaction capabilities, current predictive outbound calling systems often fail to engage with customers after the call is connected, or use fixed scripts. This increases the probability of customer hang-ups and, consequently, call loss rates. Furthermore, the interaction process is rigid, unable to accurately identify customer intent and struggling to meet personalized needs. Summary of the Invention

[0004] The embodiments of the present invention provide a predictive outbound calling method, apparatus, device, and storage medium. The technical solution of the embodiments of the present invention solves the problems of low prediction accuracy and high call loss rate in existing predictive outbound calling technologies. The number of outbound calls initiated can be accurately predicted by combining the time series prediction capability of the long short-term memory network, thereby improving the accuracy and efficiency of predictive outbound calling. Flexible and natural voice interaction with customers can also be carried out based on the intelligent interaction capability of the large model, thereby improving customer participation and satisfaction and reducing the call loss rate.

[0005] In a first aspect, an embodiment of the present invention provides a predictive outbound calling method, the method comprising:

[0006] Historical call traffic data is acquired at preset intervals and input into a target long-short-term memory network to obtain a target outbound call quantity for the next outbound call cycle. Based on the target outbound call quantity, outbound call operations are performed on customers in the outbound call list, and corresponding outbound call results are determined. If the outbound call result indicates that the customer agrees to answer the call, a voice dialogue operation is conducted with the answering customer based on the target large language model.

[0007] In a second aspect, an embodiment of the present invention provides a predictive outbound calling device, the device comprising:

[0008] The outbound call quantity determination module is used to obtain historical call traffic data at every preset period and input the historical call traffic data into the target long-short-term memory network to obtain the target outbound call quantity for the next outbound call cycle; the outbound call execution module is used to execute outbound call operations on customers in the outbound call list based on the target outbound call quantity and determine the corresponding outbound call results; the voice dialogue module is used to conduct voice dialogue operations with the answering customer based on the target large language model when the outbound call result is that the customer agrees to answer the call.

[0009] In a third aspect, an embodiment of the present invention provides a computer device, the computer device comprising:

[0010] one or more processors;

[0011] a memory for storing one or more programs;

[0012] When the one or more programs are executed by the one or more processors, the one or more processors implement the predictive outbound calling method described in any embodiment.

[0013] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the predictive outbound calling method described in any embodiment.

[0014] The technical solution provided by the embodiment of the present invention obtains historical call data at every preset period, and inputs the historical call data into the target long short-term memory network to obtain the target outbound call number for the next outbound call cycle; performs outbound call operations on customers in the outbound call list based on the target outbound call number, and determines the corresponding outbound call results; when the outbound call result is that the customer agrees to answer the call, conducts a voice dialogue operation with the answering customer based on the target large language model. The technical solution of the embodiment of the present invention solves the problems of low prediction accuracy and high call loss rate in existing predictive outbound call technology. It can accurately predict the number of outbound calls initiated by combining the time series prediction capability of the long short-term memory network, thereby improving the accuracy and efficiency of predictive outbound calls. It can also conduct flexible and natural voice interactions with customers based on the intelligent interaction capability of the large model, thereby improving customer participation and satisfaction, and thus reducing the call loss rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 This is a flow chart of a predictive outbound calling method provided by an embodiment of the present invention;

[0016] Figure 2 This is a flow chart of another predictive outbound calling method provided by an embodiment of the present invention;

[0017] Figure 3 This is a workflow diagram for predictive outbound calling provided by an embodiment of the present invention;

[0018] Figure 4 This is a structural diagram of a predictive outbound calling device provided by an embodiment of the present invention;

[0019] Figure 5 1 is a schematic structural diagram of a predictive outbound calling device provided by an embodiment of the present invention;

[0020] Figure 6 It is a structural diagram of a computer device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0021] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. The acquisition, storage, use, processing, etc. of data in the technical solutions of the embodiments of the present invention comply with the relevant provisions of national laws and regulations.

[0022] Figure 1 This is a flow chart of a predictive outbound calling method provided by an embodiment of the present invention. The embodiment of the present invention is applicable to scenarios where predictive outbound calling is performed on a customer list. The method can be executed by a predictive outbound calling device, which can be implemented by software and / or hardware.

[0023] like Figure 1 As shown, the predictive outbound calling method includes the following steps:

[0024] S110 , acquiring historical call traffic data at every preset period, and inputting the historical call traffic data into a target long short-term memory network to obtain a corresponding target outbound call quantity for the next outbound call cycle.

[0025] Among them, the preset period can be a period for performing customer outbound call operations. The technical solution of the embodiment of the present invention can estimate the number of outbound customers at intervals of the preset period, and then perform customer outbound call operations based on the estimated number of outbound customers. The historical traffic data can be relevant data on outbound call operations in a historical period. For example, the historical traffic data includes: the number of idle seats, call duration, number of initiated outbound calls, outbound call ratio, connection rate and call loss rate in the historical period. Specifically, the traffic data within the n preset periods before the current moment can be used as historical traffic data. The embodiment of the present invention can analyze the historical traffic data to determine the number of outbound customers that matches the carrying capacity of the outbound call system, so as to improve the efficiency of outbound operations, reduce the outbound call loss rate, and maximize resource utilization.

[0026] The target long-short-term memory network can be used to predict the number of outbound customers. Specifically, the long-short-term memory network (LSTM) is a special type of recurrent neural network that is particularly well-suited for processing time series data. Through its input gate, forget gate, output gate, and cell state mechanism, LSTM effectively addresses the vanishing gradient problem of traditional RNNs in long-sequence learning. LSTM can effectively learn and memorize patterns in time-varying patterns such as customer answer rates and agent idle rates during outbound calls, capturing long-term dependencies and improving forecasting flexibility. Furthermore, by updating state information in real time, LSTM can dynamically adjust outbound call strategies and enhance forecasting accuracy.

[0027] Furthermore, the target number of outbound calls may be the number of customers currently performing outbound call operations. Specifically, historical traffic data may be input into the target long short-term memory network so that the network analyzes the historical traffic data and determines the target number of outbound calls.

[0028] Optionally, the target LSTM network can be obtained through pre-training. The training process of the target LSTM network includes: obtaining a preliminary LSTM network and sample traffic data; and training the preliminary LSTM network based on the sample traffic data to obtain the target LSTM network.

[0029] The initial LSTM network can be a raw, untrained LSTM network. Furthermore, the sample traffic data can be pre-set traffic data used as model training samples. The sample traffic data can include corresponding outbound call quantity labels for evaluating the output of the initial LSTM network and adjusting the model based on the evaluation results.

[0030] Specifically, the sample traffic data can be input into the preliminary long short-term memory network to obtain the sample outbound call number; the sample outbound call number is compared with the preset outbound call number label of the sample traffic data, and the analysis loss function is determined according to the comparison result; the preliminary long short-term memory network is adjusted according to the analysis loss function to obtain the target long short-term memory network.

[0031] The sample outbound call quantity may be the outbound call quantity predicted by the preliminary long short-term memory network based on sample traffic data. The preset outbound call quantity label may be the outbound call quantity label pre-annotated in the sample traffic data. By comparing the sample outbound call quantity with the preset outbound call quantity label of the sample traffic data, the error in the outbound call quantity prediction made by the preliminary long short-term memory network can be determined.

[0032] Furthermore, an analysis loss function can be determined based on the comparison results. The analysis loss function can be a loss function corresponding to the analysis error. The analysis loss function can be used to determine the magnitude of the error in the model analysis. Subsequently, the preliminary long short-term memory network can be adjusted based on the analysis loss function to obtain a target long short-term memory network. Specifically, the model parameters of the preliminary long short-term memory network can be adjusted based on the analysis loss function to obtain the target long short-term memory network.

[0033] S120: Perform an outbound call operation on the customers in the outbound call list based on the target outbound call quantity, and determine a corresponding outbound call result.

[0034] The pending outbound call list can be a list of customers who require outbound calls. Specifically, a list of customers in a preset outbound call list that has not been called can be used as the pending outbound call list. After determining the target number of outbound calls, a list of customers with the same number of pending outbound calls as the target customer list can be selected and subsequently used to make outbound calls to the target customer list. The outbound call results can include either a rejection or acceptance. Specifically, if the customer chooses to answer the outbound call, the outbound call result can be determined as an acceptance; otherwise, a rejection.

[0035] S130: If the outbound call result indicates that the call is accepted, conduct a voice dialogue operation with the answering customer based on the target large language model.

[0036] The target large language model can be a large language model used for voice dialogue with the outbound call customer. The large language model (LLM) has strong natural language understanding and context reasoning capabilities, and can provide flexible and natural voice or text responses during customer interaction. Typical LLMs include ChatGPT, DeepSeek, etc. They can automatically generate appropriate communication tactics based on customer historical data and current context, improve interaction experience, support multi-round dialogue, accurately identify customer intent, and flexibly adjust interaction strategies to provide more intelligent services to customers.

[0037] In the case that the outbound call result is to agree to answer, the voice dialogue operation between the target large language model and the answering customer can be performed through the large language model to realize product marketing, customer willingness and customer business information collection, etc. The voice interaction is flexible and natural, which improves customer participation and satisfaction while reducing the communication cost of the agent and improving the outbound call efficiency.

[0038] Optionally, after the voice dialogue operation based on the target large language model and the answering customer, in the case that a transfer to human instruction is received from the answering customer, an idle agent is selected as a target agent, and the answering customer is transferred to the target agent.

[0039] The technical scheme provided by the embodiment of the application obtains historical traffic data every interval of a preset period, inputs the historical traffic data into a target long short-term memory network to obtain a target outbound call quantity corresponding to a next outbound call period, performs an outbound call operation on customers in a to-be-called list based on the target outbound call quantity, and determines a corresponding outbound call result. In the case that the outbound call result is to agree to answer, a voice dialogue operation between the target large language model and the answering customer is performed. The technical scheme of the embodiment of the application solves the problems of low prediction accuracy and high call loss rate in the existing predictive outbound call technology, can accurately predict the outbound call initiation quantity by combining the time series prediction capability of the long short-term memory network, improves the accuracy and efficiency of the predictive outbound call, can flexibly and naturally interact with the customer based on the intelligent interaction capability of the large model, improves customer participation and satisfaction, and further reduces the call loss rate.

[0040] Figure 2This is another flow chart of a predictive outbound calling method provided by an embodiment of the present invention. The embodiment of the present invention can be applied to scenarios where predictive outbound calling is performed on a customer list. Based on the above embodiment, this embodiment further supplements the target long-short-term memory network, including: a data processing module, a feature extraction module, and an outbound call prediction module. The large language model includes: a grammar generator, a context understander, and a dialogue guide. It further explains how to input the historical traffic data into the target long-short-term memory network to obtain the corresponding target outbound call quantity for the next outbound call cycle; and how to conduct voice dialogue operations with the answering customer based on the target large language model. The device can be implemented by software and / or hardware and integrated into a computer device with application development capabilities.

[0041] like Figure 2 As shown, the predictive outbound calling method includes the following steps:

[0042] S210 , acquiring historical traffic data at preset intervals, and inputting the historical traffic data into the data processing module to obtain processed traffic data.

[0043] The processed traffic data is data that has been time-windowed and normalized. Specifically, the historical traffic data can be input into the data processing module so that the network performs time-windowed and normalized processing on the historical traffic data to obtain the processed traffic data.

[0044] S220: Input the processed traffic data into the feature extraction module to obtain corresponding outbound call timing features.

[0045] The outbound call timing feature may be a timing feature corresponding to historical traffic data. The outbound call prediction module may analyze the outbound call timing feature to determine a correlation between the number of outbound calls and traffic data, and further determine a target number of outbound calls corresponding to the historical traffic data.

[0046] S230: Input the outbound call time series features into the outbound call prediction module to obtain the target number of outbound calls.

[0047] S240: Perform an outbound call operation on the customers in the outbound call list based on the target outbound call quantity, and determine a corresponding outbound call result.

[0048] S250. When the outbound call result indicates that the caller agrees to answer the call, the speech generator obtains the dialogue scenario prompt information fed back by the answering customer and determines the corresponding target speech template.

[0049] The conversation scenario prompt information may be information representing the conversation business scenario. Specifically, the conversation scenario prompt information may be determined based on the business needs of the customer. For example, the business needs may be determined based on the customer's voice or the service code provided, and the conversation scenario prompt information may be determined accordingly. The target speech template may be a speech template used for conversations with the customer. Specifically, the speech generator may determine the target speech template corresponding to the conversation scenario prompt information based on the association between the conversation scenario and the speech template.

[0050] S260: The context understander obtains corresponding context conversation features in advance based on the current conversation content and historical conversation content of the answering customer.

[0051] The contextual conversation feature may be a feature representing the contextual conversation between the caller and the client. Specifically, the context understander may extract semantic features from the current conversation content and the historical conversation content with the caller to obtain the contextual conversation feature.

[0052] S270. The dialogue guide guides the customer to complete the voice dialogue service according to the target speech template, the context dialogue features and preset information points.

[0053] Preset information points can include preset voice content, preset customer background information, and preset business rules. The conversation guide can understand the customer's current needs and business progress based on the context of the conversation. It then provides voice prompts to the customer based on the target speech template and preset information points, guiding the customer through the voice conversation.

[0054] For example, in order to better understand the technical solution provided by the present invention, a specific embodiment is introduced below: Figure 3 This is a workflow diagram for predictive outbound calling provided by an embodiment of the present invention. Figure 3 As shown, the workflow for predictive outbound calling includes the following steps: Step 1: The system collects, slices, and normalizes data. The processed data is used for model training until the model reaches the expected accuracy;

[0055] Step 2: The system collects actual outbound call data, inputs it into the trained LSTM model, and obtains the prediction results;

[0056] Step 3: The system obtains the outbound call list from the upstream and initiates an outbound call based on the prediction results;

[0057] Step 4: If the customer answers the call, the call will be transferred to the large model intelligent interaction process;

[0058] Step 5: If the customer does not answer, repeat steps 2-4;

[0059] Step 6: Use the large-scale intelligent interactive process to promote products, collect information in advance, and inquire about the customer's willingness to transfer.

[0060] Step 7: If the customer agrees to transfer, the agent will be connected to complete the manual connection;

[0061] Step 8: If the customer disagrees, repeat steps 2-7.

[0062] Further, Figure 4 This is a system structure diagram of a predictive outbound call provided by an embodiment of the present invention. Figure 4 As shown, the device mainly includes four units: data collection and training unit, LSTM prediction unit, outbound call execution unit, and large-model intelligent interaction unit. The data collection and training unit is responsible for collecting, cleaning, and feature engineering multi-source heterogeneous data, as well as initial and iterative training of the LSTM model. The LSTM prediction unit is responsible for intelligent prediction and decision-making, predicting the number of outbound calls that should be initiated at this moment through actual traffic data. The outbound call execution unit is responsible for dynamic outbound calls. After obtaining the list issued by the upstream, it initiates outbound calls based on the predicted number of outbound calls, and monitors and collects outbound call data in real time through a monitor. If the customer intends to transfer to an agent, a bridging operation is performed. The large-model intelligent interaction unit is responsible for intelligent interaction. After the customer is connected, it generates speech, understands the context, and guides the conversation. It also performs operations such as product promotion and preliminary data collection to reduce call loss rate and improve outbound call efficiency.

[0063] The data collection and training unit is primarily responsible for collecting, preprocessing, and normalizing outbound call data. This unit uses this dimensionalized data to train an LSTM model for predictive outbound calling. This unit primarily comprises a basic data collector, a data time series slicer, a data normalizer, and a model trainer. The basic data collector collects key business data such as the number of available agents, ringing duration, outbound call duration, and the number of customers in queue. The data time series slicer divides the collected data into time windows to form data slices arranged in time series, which are suitable for LSTM model input modeling. The data normalizer normalizes the sliced ​​data, converting data of different dimensions to the same dimensional range, thereby improving LSTM model training accuracy and predictive stability. Finally, the organized dataset is divided into a training set and a validation set in an 8:2 ratio. This data is then fed into the model trainer for LSTM model training. Training stops when the model reaches the specified accuracy, at which point the trained LSTM model is exported.

[0064] The LSTM prediction unit is the core decision-making engine for predictive outbound calling. It consists of a data input unit, an LSTM modeler, and a prediction output unit. The data input unit feeds time-series sliced ​​and normalized data collected during actual outbound calling into the LSTM model. The LSTM modeler uses LSTM to process time-series data and learns the dynamic patterns of indicators such as outbound call volume and connection rate based on historical data characteristics. The prediction output unit, based on the LSTM model's predictions, outputs the optimal outbound call volume at the current moment. This ensures an optimal balance between agent resources and connected customers at the next moment, effectively improving outbound calling efficiency and reducing the risk of customer churn.

[0065] The outbound call execution unit is responsible for executing outbound call tasks based on the LSTM model's prediction results. It comprises a list acquirer, an outbound call executor, an outbound call monitor, and a call bridge. The list acquirer retrieves the target customer list from the upstream system. After receiving the prediction results, it pushes the corresponding number of outbound call lists to the outbound call executor. The outbound call executor automatically initiates outbound calls based on the predicted number and, after load balancing, distributes the outbound call requests across multiple servers within the FS cluster. The outbound call monitor monitors outbound call progress in real time, recording call status, including connections, rejections, and hang-ups. It also collects call data to optimize the LSTM prediction model. Once a customer connects and wishes to be transferred to a live agent, the call bridge connects the customer to an available agent, providing timely and efficient service.

[0066] The large-model intelligent interaction unit provides intelligent conversation guidance after the customer connects, effectively improving the customer experience. The system primarily consists of three components: a speech generator, a context understander, and a conversation guide. The speech generator, based on a large language model, generates personalized speech that meets business requirements and dynamically adjusts interaction and follow-up strategies based on different scenarios. The context understander analyzes the customer's voice content in real time, combining historical interaction data for semantic understanding to ensure conversation coherence and accuracy. Furthermore, if it detects emotional fluctuations or repetition of similar speech, it promptly and politely directs the customer to a human agent. The conversation guide, based on contextual information and pre-set information points, guides the customer through steps such as information confirmation and business introduction. It also pre-collects business-related information for interested customers and transfers them to a human agent, streamlining agent interaction costs. This improves customer satisfaction while ensuring business success rates and enhancing agent conversation efficiency.

[0067] The technical solution provided by the embodiment of the present invention obtains historical traffic data at preset intervals, inputs the historical traffic data into the data processing module to obtain processed traffic data; inputs the processed traffic data into the feature extraction module to obtain corresponding outbound call timing features; inputs the outbound call timing features into the outbound call prediction module to obtain the target outbound call number; performs outbound call operations on customers in the outbound call list based on the target outbound call number, and determines the corresponding outbound call results; when the outbound call result is that the customer agrees to answer the call, the speech generator obtains the dialogue scene prompt information fed back by the answering customer and determines the corresponding target speech template; the context understander advances the corresponding context dialogue features based on the current dialogue content and historical dialogue content of the answering customer; the dialogue guide guides the customer to complete the voice dialogue service based on the target speech template, the context dialogue features and preset information points. The technical solution of the embodiment of the present invention solves the problems of low prediction accuracy and high call loss rate in existing predictive outbound calling technology. It can accurately predict the number of outbound calls initiated by combining the time series prediction capability of the long short-term memory network, thereby improving the accuracy and efficiency of predictive outbound calling. It can also conduct flexible and natural voice interaction with customers based on the intelligent interaction capability of the large model, thereby improving customer participation and satisfaction, and thus reducing the call loss rate.

[0068] Figure 5 This is a structural diagram of a predictive outbound calling device provided by an embodiment of the present invention. The embodiment of the present invention is applicable to scenarios where predictive outbound calling is performed on a customer list. The device can be implemented by software and / or hardware and integrated into a computer device with application development capabilities.

[0069] like Figure 5 As shown, the predictive outbound calling device includes: an outbound call quantity determination module 310 , an outbound call execution module 320 and a voice dialogue module 330 .

[0070] Among them, the outbound call quantity determination module 310 is used to obtain historical call data at every preset period, and input the historical call data into the target long short-term memory network to obtain the target outbound call quantity of the corresponding next outbound call cycle; the outbound call execution module 320 is used to perform outbound call operations on customers in the outbound call list based on the target outbound call quantity, and determine the corresponding outbound call results; the voice dialogue module 330 is used to conduct voice dialogue operations with the answering customer based on the target large language model when the outbound call result is that the call is accepted.

[0071] The technical solution provided by the embodiment of the present invention obtains historical call data at every preset period, and inputs the historical call data into the target long short-term memory network to obtain the target outbound call number for the next outbound call cycle; performs outbound call operations on customers in the outbound call list based on the target outbound call number, and determines the corresponding outbound call results; when the outbound call result is that the customer agrees to answer the call, conducts a voice dialogue operation with the answering customer based on the target large language model. The technical solution of the embodiment of the present invention solves the problems of low prediction accuracy and high call loss rate in existing predictive outbound call technology. It can accurately predict the number of outbound calls initiated by combining the time series prediction capability of the long short-term memory network, thereby improving the accuracy and efficiency of predictive outbound calls. It can also conduct flexible and natural voice interactions with customers based on the intelligent interaction capability of the large model, thereby improving customer participation and satisfaction, and thus reducing the call loss rate.

[0072] In an optional embodiment, the target long short-term memory network includes: a data processing module, a feature extraction module and an outbound call prediction module, and the outbound call quantity determination module 310 is specifically used to: input the historical traffic data into the data processing module to obtain processed traffic data; wherein the processed traffic data is data after time window division and normalization; input the processed traffic data into the feature extraction module to obtain corresponding outbound call timing features; input the outbound call timing features into the outbound call prediction module to obtain the target outbound call quantity.

[0073] In an optional embodiment, the target large language model includes: a speech generator, a context understander and a dialogue guide, and the voice dialogue module 330 is specifically used for: the speech generator obtains the dialogue scene prompt information of the answering customer feedback and determines the corresponding target speech template; the context understander advances the corresponding context dialogue features based on the current dialogue content and historical dialogue content of the answering customer; the dialogue guide guides the customer to complete the voice dialogue service based on the target speech template, the context dialogue features and preset information points.

[0074] In an optional embodiment, the predictive outbound calling device further includes: a manual transfer module configured to: upon receiving a manual transfer instruction from the answering customer, select an idle agent as a target agent and transfer the answering customer to the target agent.

[0075] In an optional embodiment, the predictive outbound calling device further includes: a long short-term memory network training module, which is used to: obtain a preliminary long short-term memory network and sample traffic data; train the preliminary long short-term memory network based on the sample traffic data to obtain the target long short-term memory network.

[0076] In an optional embodiment, the long short-term memory network training module includes: a model adjustment unit, used to: input the sample traffic data into the preliminary long short-term memory network to obtain the sample outbound call number; determine the analysis loss function based on the sample outbound call number and the preset outbound call number label of the sample traffic data; adjust the preliminary long short-term memory network according to the analysis loss function to obtain the target long short-term memory network.

[0077] In an optional implementation, the historical traffic data includes at least one of the number of idle seats, call duration, number of initiated outbound calls, outbound call ratio, connection rate and call loss rate during a historical period.

[0078] The predictive outbound calling device provided in the embodiment of the present invention can execute the predictive outbound calling method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0079] Figure 6 A schematic structural diagram of a computer device provided in an embodiment of the present invention. Figure 6 A block diagram of an exemplary computer device 12 suitable for use in implementing embodiments of the present invention is shown. Figure 6 The computer device 12 shown is only an example and should not limit the functionality and scope of use of the embodiments of the present invention. The computer device 12 can be any terminal device with computing capabilities and can be configured in a predictive outbound calling device.

[0080] like Figure 6 As shown, computer device 12 is implemented as a general-purpose computing device. Components of computer device 12 may include, but are not limited to, one or more processors or processing units 16, system memory 28, and a bus 18 that connects various system components (including system memory 28 and processing unit 16).

[0081] The bus 18 may be one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor, or a local bus using any of a variety of bus architectures. Examples of these architectures include, but are not limited to, an Industry Standard Architecture (ISA) bus, a Micro Channel Architecture (MAC) bus, an Enhanced ISA bus, a Video Electronics Standards Association (VESA) local bus, and a Peripheral Component Interconnect (PCI) bus.

[0082] The computer device 12 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by the computer device 12, including volatile and non-volatile media, removable and non-removable media.

[0083] System memory 28 can include computer system readable media in the form of volatile memory, such as random access memory (RAM) 30 and / or cache 32. Computer device 12 can further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, storage system 34 can be provided for reading from and writing to a non-removable, non-volatile magnetic media (not shown and typically called a "hard drive"). Figure 6 Although not shown, computer device 12 can further include a data repository 36 that stores sets of data usable in connection with the embodiments of the present application. Although Figure 6 not shown in FIG. 1, a disk drive can be provided for reading from and writing to a removable, non-volatile magnetic disk (e.g., a "floppy disk"), and an optical disk drive can be provided for reading from or writing to a removable, non-volatile optical disk (e.g., a CD-ROM, DVD-ROM or other optical media). In these instances, each drive can be connected to bus 18 by one or more data media interfaces. System memory 28 can include at least one program product having a set (e.g., at least one) of program modules that are configured to carry out the functions of embodiments of the present application.

[0084] Program / utility 40 having a set (at least one) of program modules 42 can be stored in system memory 28 by way of example, such as an operating system, one or more application programs, other program modules, and program data, each or some combination thereof, which may

[0085] Computer device 12 can also communicate with one or more external devices 14 such as a keyboard, a pointing device, a display 24, etc.; one or more devices that enable a user to interact with computer device 12; and / or one or more devices that enable computer device 12 to communicate with one or more other computing devices. Such communication can be via input / output (I / O) interfaces 22. Still yet, computer device 12 can communicate with one or more networks (such as a local area network (LAN), a wide area network (WAN), and / or the Internet) through network adapter 20. As Figure 6 illustrated, network adapter 20 can communicate with the other components of computer device 12 via bus 18. It should be understood that although Figure 6 not shown in FIG. 1, other hardware and / or software modules can be used in conjunction with computer device 12. Such hardware would include, but is not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data archival storage systems, etc.

[0086] The processing unit 16 executes various functional applications and data processing by running programs stored in the system memory 28, such as implementing the predictive outbound calling method provided by an embodiment of the present invention, which includes:

[0087] Historical call traffic data is acquired at preset intervals and input into a target long-short-term memory network to obtain a target outbound call quantity for the next outbound call cycle. Based on the target outbound call quantity, outbound call operations are performed on customers in the outbound call list, and corresponding outbound call results are determined. If the outbound call result indicates that the customer agrees to answer the call, a voice dialogue operation is conducted with the answering customer based on the target large language model.

[0088] This embodiment provides a computer-readable storage medium having a computer program stored thereon. When the program is executed by a processor, the predictive outbound calling method provided in any embodiment of the present invention is implemented, including:

[0089] Historical call traffic data is acquired at preset intervals and input into a target long-short-term memory network to obtain a target outbound call quantity for the next outbound call cycle. Based on the target outbound call quantity, outbound call operations are performed on customers in the outbound call list, and corresponding outbound call results are determined. If the outbound call result indicates that the customer agrees to answer the call, a voice dialogue operation is conducted with the answering customer based on the target large language model.

[0090] The computer storage medium of the embodiment of the present invention can adopt any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium can be, for example, but not limited to: an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or component, or any combination of the above. More specific examples (non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device.

[0091] A computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0092] Program code embodied on a computer-readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0093] The computer program code for performing the operations of the present invention can be written in one or more programming languages, or a combination thereof, including object-oriented programming languages ​​such as C, Java, Smalltalk, C++, C#, and Python, and also conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the client computer, partially on the client computer, as a separate software package, partially on the client computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer can be connected to the client computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0094] Those skilled in the art will appreciate that the modules or steps of the present invention described above can be implemented using a general-purpose computing device. They can be centralized on a single computing device or distributed across a network of multiple computing devices. Alternatively, they can be implemented using program code executable by a computer device, which can then be stored in a storage device and executed by the computing device. Alternatively, they can be fabricated into separate integrated circuit modules, or multiple modules or steps can be fabricated into a single integrated circuit module. Thus, the present invention is not limited to any specific combination of hardware and software.

[0095] Note that the above merely describes preferred embodiments of the present application and the principles of the technology applied. Those skilled in the art will understand that the present application is not limited to the specific embodiments described herein, and that various obvious changes, modifications and substitutions can be made thereto without departing from the scope of the present application. Therefore, although the present application has been described in detail through the above embodiments, the present application is not limited to the above embodiments, and can include more other equivalent embodiments without departing from the concept of the present application, and the scope of the present application is determined by the scope of the claims.

Claims

1. A predictive outbound calling method, characterized in that: include: Acquire historical traffic data at preset intervals, and input the historical traffic data into the target long-short-term memory network to obtain the target number of outbound calls for the next outbound call cycle; Performing outbound call operations on customers in the outbound call list based on the target outbound call quantity, and determining corresponding outbound call results; If the outbound call result indicates that the customer agrees to answer the call, a voice conversation operation is performed with the answering customer based on the target large language model.

2. The method according to claim 1, characterized in that The target long-short-term memory network includes: a data processing module, a feature extraction module and an outbound call prediction module. The historical traffic data is input into the long-short-term memory network in the target outbound call model to obtain the target number of outbound calls for the next outbound call cycle. Inputting the historical traffic data into the data processing module to obtain processed traffic data; wherein the processed traffic data is data that has been divided into time windows and normalized; Inputting the processed traffic data into the feature extraction module to obtain corresponding outbound call timing features; The outbound call timing characteristics are input into the outbound call prediction module to obtain the processed call traffic data.

3. The method according to claim 1, characterized in that The target large language model includes: a grammar generator, a context understander, and a dialogue guide. The voice dialogue operation with the answering customer based on the target large language model includes: The speech generator obtains the dialogue scene prompt information of the customer feedback and determines the corresponding target speech type; The context understander generates corresponding context conversation features in advance based on the current conversation content and historical conversation content of the answering customer; The dialogue guide guides the customer to complete the voice dialogue service according to the context dialogue features and preset information points.

4. The method according to claim 1, further comprising: When receiving a call transfer instruction from the customer, an agent in an idle state is used as a target agent, and the customer is transferred to the target agent.

5. The method according to claim 1, wherein The training process of the target long short-term memory network includes: Obtain preliminary long short-term memory network and sample traffic data; The preliminary long short-term memory network is trained based on the sample traffic data to obtain the target long short-term memory network.

6. The method according to claim 5, characterized in that The training of the preliminary long short-term memory network based on the sample traffic data to obtain the target long short-term memory network includes: Inputting the sample traffic data into the preliminary long short-term memory network to obtain the number of sample outbound calls; Determine an analysis loss function based on the sample outbound call quantity and a preset outbound call quantity label of the sample traffic data; The preliminary long short-term memory network is adjusted according to the analysis loss function to obtain the target long short-term memory network.

7. The method according to claim 1, characterized in that The historical traffic data includes at least one of the number of idle seats, call duration, number of initiated outbound calls, outbound call ratio, connection rate and call loss rate within a historical period.

8. A predictive outbound calling device, characterized in that: The device comprises: An outbound call quantity determination module is configured to obtain historical call traffic data at predetermined intervals and input the historical call traffic data into a target long-short-term memory network to obtain a target outbound call quantity for the next outbound call cycle. An outbound call execution module, configured to execute an outbound call operation on the customers in the outbound call list based on the target outbound call quantity, and determine the corresponding outbound call result; The voice dialogue module is used to conduct a voice dialogue with the answering customer based on the target large language model when the outbound call result indicates that the customer agrees to answer the call.

9. A computer device, characterized in that: The computer device comprises: one or more processors; a memory for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the predictive outbound calling method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the predictive outbound calling method according to any one of claims 1 to 7 is implemented.

Images