Self-adaptive telephone traffic distribution method and system based on customer service emotion

Abstract

The invention discloses a self-adaptive telephone traffic distribution method and system based on customer service emotion. Comprising the following steps: acquiring a first emotional state data set of a plurality of first customer service staff in an idle state in a first time period and a first customer evaluation value after each first customer service staff answers a first incoming call in a second time period, and acquiring a call answering skill value of each first customer service staff; for each first customer service, analyzing the first emotional state data set and the first customer evaluation value of the first customer service by using a target emotional evaluation model to obtain a first emotional value of the first customer service in a first time period, and determining the call receiving priority of each first customer service for answering the second incoming call according to the call receiving skill value and the first emotional value; and in response to an answering request for the second incoming call, determining a target customer service answering the second incoming call according to the call answering priority of each first customer service. According to the invention, the technical problem that the telephone traffic distribution result cannot adapt to the actual working state of the customer service because the analysis of the emotional state of the customer service is neglected in the related telephone traffic distribution technology is solved.

Description

Technical Field

[0001] The present application relates to the technical field of call management. Specifically, it relates to an adaptive call distribution method and system based on customer service emotions. Background Art

[0002] The voice customer service system is an integrated service platform that combines communication technology and information technology and is widely used in multiple industries such as banking, telecommunications, retail, and healthcare. As a core link in the voice customer service system, the efficiency of call distribution management is directly related to the workload of customer service representatives and the quality of the customer service experience. Therefore, adopting an efficient and scientific call distribution method is crucial for maintaining the smooth operation of voice customer service work. Currently, call distribution management mainly focuses on improving the utilization rate of call resources by shortening the artificial call queuing time and increasing the artificial call connection rate, and there are the following limitations in the customer experience in terms of humanistic care:

[0003] (1) In terms of the access layer, an Interactive Voice Response (IVR) system, an AI customer service system, etc. are introduced on the basis of traditional hardware devices such as Private Branch Exchange (PBX), media gateways, and signaling gateways, and customer self-service is provided by integrating the business system data sources related to the customer service system. Therefore, situations such as "system misjudgment" and "technical barriers" for elderly customers may occur.

[0004] (2) In terms of queuing machines and intelligent routing management, an Automatic Call Distribution (ACD) system is usually used to implement functions such as incoming call queuing, queue notification, priority queuing, and skill group queuing. Although the combined use of queuing methods and strategies can maximize the role of customer service representative resources and retain customer value, it ignores the reduction in the customer service experience caused by the negative emotions of customer service representatives.

[0005] (3) In terms of real-time monitoring and historical data analysis, currently, the analysis mainly focuses on the status of users in the real-time call interaction process and the historical behaviors and needs of users, and lacks the analysis of customer service representatives, thus ignoring the value of the emotional continuity of customer service representatives.

[0006] Therefore, there is an urgent need for an effective adaptive call distribution method based on the emotions of customer service representatives to improve the customer experience in terms of humanistic care in call distribution management. Summary of the Invention

[0007] An embodiment of the present application provides an adaptive call traffic allocation method and system based on customer service emotions, so as to at least solve the technical problem that relevant call traffic allocation technologies ignore the analysis of the emotional state of customer service, resulting in the call traffic allocation result being unable to adapt to the actual working state of customer service.

[0008] According to one aspect of the embodiments of the present application, an adaptive call traffic allocation method based on customer service emotions is provided, including: obtaining a first emotional state data set of each of a plurality of first customer services in an idle state during a first time period, and a first customer evaluation value corresponding to each first customer service answering a first incoming call during a second time period, and obtaining the call answering skill value of each first customer service, where the second time period is a time period before the first time period; for each first customer service, using a pre-trained target emotion evaluation model to analyze the first emotional state data set and the first customer evaluation value to obtain a first emotion value of the first customer service during the first time period, and determining the call answering priority of the first customer service answering a second incoming call based on the call answering skill value and the first emotion value; in response to a call answering request for the second incoming call, determining a target customer service for answering the second incoming call based on the call answering priorities of each first customer service.

[0009] Optionally, obtaining a first emotional state data set of each of a plurality of first customer services in an idle state during a first time period and a first customer evaluation value corresponding to each first customer service answering a first incoming call during a second time period includes: obtaining an initial emotional state data set of each of a plurality of first customer services in an idle state during the first time period, where the initial emotional state data set includes emotional state data of a plurality of first dimensions, and the first dimension includes at least one of the following: expression, volume, pitch, semantics, blood pressure, heart rate; preprocessing each initial emotional state data set to obtain a first emotional state data set of each first customer service during the first time period, where the preprocessing includes at least one of the following: data cleaning, normalization processing; for each first customer service, obtaining customer evaluation information corresponding to the first customer service answering the first incoming call during the second time period and customer evaluation information corresponding to a plurality of historical incoming calls answered by the first customer service respectively, where the customer evaluation information includes evaluation information of a plurality of second dimensions, and the second dimension includes at least one of the following: customer satisfaction, problem handling speed, problem solving rate; according to a preset first quantization rule table, respectively determining a first quantization result of each evaluation information in the customer evaluation information corresponding to the first incoming call and a first quantization result of each evaluation information in the customer evaluation information corresponding to each historical incoming call, where the first quantization rule table includes first quantization rules corresponding to each evaluation information of a plurality of second dimensions, and the first quantization rule is used to convert the fuzzy definition of the evaluation information into a corresponding numerical value; according to the first quantization result of each evaluation information in the customer evaluation information corresponding to the first incoming call and the first quantization result of each evaluation information in the customer evaluation information corresponding to each historical incoming call, determining the first customer evaluation value corresponding to the first incoming call according to the following formula: In the formula, X represents the first customer evaluation value, and x i represents the first quantization result corresponding to the i-th type of customer evaluation information corresponding to the first incoming call, n represents the number of pieces of information of the customer evaluation information corresponding to the first incoming call, respectively represent the maximum value and the minimum value of the first quantization results of the i-th type of customer evaluation information corresponding to multiple historical incoming calls, and α i represents the weight coefficient corresponding to the i-th type of customer evaluation information.

[0010] Optionally, obtaining the call-answering skill value of the first customer service includes: obtaining the call-answering skill information of the first customer service, where the call-answering skill information includes multiple pieces of skill evaluation information in a third dimension, and the third dimension includes at least one of the following: the human resource status in the skill field to which the customer service belongs, the maturity of skill mastery; determining the second quantization results corresponding to the respective skill evaluation information in the call-answering skill information according to a preset second quantization rule table, where the second quantization rule table includes the second quantization rules corresponding to multiple pieces of skill evaluation information in the third dimension, and the second quantization rule is used to convert the fuzzy definition of the skill evaluation information into a corresponding numerical value; according to the second quantization results corresponding to the respective skill evaluation information in the call-answering skill information, determining the call-answering skill value of the first customer service according to the following formula: In the formula, Y represents the call-answering skill value, and y j represents the second quantization result of the i-th skill evaluation information corresponding to the first customer service, m represents the number of pieces of information of the skill evaluation information in the call-answering skill information, respectively represent the maximum value and the minimum value of the second quantization results of the i-th skill evaluation information, and θ i represents the weight coefficient corresponding to the i-th skill evaluation information.

[0011] Optionally, the training process of the target emotion evaluation model includes: constructing an initial emotion evaluation model; obtaining multiple groups of first training samples, where each first training sample includes: a second emotion state data set of the second customer answering calls during a first historical time period and the second customer evaluation value corresponding to the second customer answering a third incoming call during a second historical time period, and the second emotion value of the second customer during the first historical time period, and the second historical time period is a time period before the first historical time period; for each group of first training samples, inputting the second emotion state data set and the second customer evaluation value in the first training sample into the initial emotion evaluation model to obtain a predicted emotion value output by the initial emotion evaluation model; constructing a first target loss function of the initial emotion evaluation model according to the second emotion value and the predicted emotion value in each group of first training samples, and updating the model parameters of the initial emotion evaluation model by minimizing the first target loss function until a preset first convergence condition is satisfied, so as to obtain the target emotion evaluation model that has completed training.

[0012] Optionally, determine the answering priority of the first customer service agent for answering the second incoming call based on the answering skill value and the first emotion value, including: based on the answering skill value and the first emotion value, and determine the answering priority of the first customer service agent for answering the second incoming call according to the following formula: In the formula, P k represents the k-th first customer service agent in the idle state, V k represents the first emotion value when the k-th first customer service agent in the idle state answers the first incoming call, Y k represents the answering skill value of the k-th first customer service agent in the idle state, ω respectively represent the weight coefficients.

[0013] Optionally, determine the target customer service agent for answering the second incoming call based on the answering priorities of each first customer service agent, including: taking the first customer service agent with the highest answering priority as the target customer service agent for answering the second incoming call; or, based on the answering priorities of each first customer service agent and in combination with a preset allocation strategy, determine the target customer service agent for answering the second incoming call, where the allocation strategy includes at least one of the following: the least idle time strategy for allocating a new incoming call to the target customer service agent with the shortest processing time for handling the current incoming call or the longest idle time, and the skill routing strategy for allocating a new incoming call to the target customer service agent with the highest answering skill score.

[0014] Optionally, based on the answering priorities of each first customer service agent and in combination with a preset allocation strategy, determine the target customer service agent for answering the second incoming call, including: in the case where the allocation strategy is the least idle time strategy, determine the processing time of each first customer service agent for handling the first incoming call or the idle time in the idle state, and take the first customer service agent with the shortest processing time or the longest idle time and the highest answering priority as the target customer service agent for answering the second incoming call; in the case where the allocation strategy is the skill routing strategy, determine the answering skill scores of each first customer service agent, and take the first customer service agent with the highest answering skill score and the highest answering priority as the target customer service agent for answering the second incoming call.

[0015] According to another aspect of the embodiments of the present application, there is also provided an adaptive call distribution system based on customer service emotions, including: an acquisition module, configured to acquire a first emotion state dataset of multiple first customer services in an idle state during a first time period, a first customer evaluation value corresponding to the first incoming calls answered by each first customer service during a second time period, and an incoming call skill value of each first customer service, where the second time period is a time period before the first time period; a determination module, configured to, for each first customer service, analyze the first emotion state dataset and the first customer evaluation value by using a pre-trained target emotion evaluation model to obtain a first emotion value of the first customer service during the first time period, and determine an incoming call priority for the first customer service to answer a second incoming call based on the incoming call skill value and the first emotion value of the first customer service; and an allocation module, configured to, in response to a request to answer a second incoming call, determine a target customer service for answering the second incoming call based on the incoming call priorities of the respective first customer services.

[0016] According to another aspect of the embodiments of the present application, there is also provided a computer program product, which includes: a computer program, where when the computer program is executed by a processor, it implements the above-mentioned adaptive call distribution method based on customer service emotions.

[0017] According to another aspect of the embodiments of the present application, there is also provided an electronic device, which includes: a memory and a processor, where a computer program is stored in the memory, and the processor is configured to execute the above-mentioned adaptive call distribution method based on customer service emotions through the computer program.

[0018] In the embodiments of the present application, by collecting and analyzing a first emotion state dataset of multiple first customer services in an idle state during a first time period, a first customer evaluation value corresponding to the first incoming calls answered by each first customer service during a second time period, and using a pre-trained emotion evaluation model to dynamically determine a first emotion value of each first customer service during the first time period; determining an incoming call priority for answering a second incoming call based on the incoming call skill value of each first customer service and the first emotion value of each first customer service during the first time period; and when receiving a request to answer a second incoming call, determining a target customer service for answering the second incoming call based on the incoming call priorities of the respective first customer services. The entire solution achieves the technical effect of intelligent call distribution based on the emotion state of customer service representatives, achieves the purpose of optimizing customer service quality, improving customer experience satisfaction, and reasonably allocating human resources, and further solves the technical problem that the relevant call distribution technology ignores the analysis of the emotion state of customer service, resulting in the call distribution result being unable to adapt to the actual working state of customer service. Description of the Drawings

[0019] The accompanying drawings described herein are used to provide a further understanding of the present application and form a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation of the present application. In the drawings:

[0020] Figure 1 is a schematic flowchart of an optional method for adaptive call traffic allocation based on customer service emotion according to an embodiment of the present application;

[0021] Figure 2 is a schematic structural diagram of an optional system for adaptive call traffic allocation based on customer service emotion according to an embodiment of the present application;

[0022] Figure 3 is a schematic structural diagram of an optional electronic device according to an embodiment of the present application. Detailed implementation manners

[0023] In order to enable those skilled in the art to better understand the solution of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

[0024] It should be noted that the terms "first", "second", etc. in the specification, claims and drawings of the present application are used to distinguish similar objects and do not necessarily need to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units does not necessarily have to be limited to those clearly listed steps or units, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0025] To better understand the embodiments of the present application, some nouns or terms that appear in the description process of the embodiments of the present application are translated and explained as follows:

[0026] Call traffic allocation: It is the process of managing and allocating incoming calls, online requests, and other forms of customer contacts in a call center or voice customer service system. Its goal is to ensure that customer requests can be processed quickly and effectively, and resources are optimally configured.

[0027] PCI (Peripheral Component Interconnect) sound card: A sound card that connects to the computer motherboard through the PCI interface to provide audio input and output functions for the computer. It can usually be applied to various scenarios such as voice calls, music playback, and games.

[0028] Embodiment 1

[0029] According to an embodiment of the present application, an adaptive call traffic allocation method based on customer service emotions is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. And although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here.

[0030] Figure 1 It is a schematic flowchart of an adaptive call traffic allocation method based on customer service emotions provided according to an embodiment of the present application. As Figure 1 shown, the method includes the following steps:

[0031] Step S102, obtain the first emotion state data sets of multiple first customer services in the idle state during the first time period and the first customer evaluation values corresponding to the first incoming calls answered by each first customer service during the second time period, and obtain the call handling skill values of each first customer service.

[0032] In the technical solution provided in the above step S102, the first customer service refers to the customer service representative who is in the idle state (i.e., not handling call traffic at the current moment) during the current time period (i.e., the first time period). The first emotion state data set refers to the emotion-related physiological data and perception data of the customer service representatives in the idle state during the current time period (i.e., the first emotion state data set) to comprehensively reflect the emotion states of each first customer service. And the first customer evaluation value can directly reflect the evaluation feedback given by the customer to the service of the customer representative after the first incoming call is answered by each first customer service in the previous time period of the current time period. In addition, the call handling skill value refers to the ability level of each first customer service in handling incoming calls of a specific type or skill requirement.

[0033] Step S104, for each first customer service, analyze the first emotion state data set and the first customer evaluation value by using a pre-trained target emotion evaluation model to obtain the first emotion value of the first customer service during the first time period, and determine the call handling priority of the first customer service for answering the second incoming call based on the call handling skill value and the first emotion value.

[0034] In the technical solution provided in the above step S104, the above-mentioned target emotion assessment model analyzes the real-time emotion state data set of each customer service representative and the customer evaluation value after the customer service representative answers the previous call, and obtains the current emotion value of each customer service representative. Therefore, the call distribution system calls the pre-trained target emotion assessment model to immediately identify the change in the emotion state of the customer service representative, and combines the customer evaluation value to obtain the emotion value reflecting the current emotion state of the customer service representative.

[0035] Among them, the above-mentioned first emotion state data set is the subjective emotion data of the first customer service, and the first customer evaluation value is the objective feedback of the impact of the first customer service's emotion on work performance. These data are used as the verification and supplement for the evaluation of the first customer service's emotion value in the model analysis, ensuring the comprehensiveness and reliability of the evaluation result. The first emotion value reflects the emotion state of the first customer service in the first time period, and is an important indicator for measuring whether it is currently suitable to answer the next call (i.e., the second call) of the first call. Subsequently, the call connection skill values (their professional capabilities and the efficiency of handling specific types of calls) of each first customer service are combined with the first emotion values of each first customer service calculated above to determine the call connection priority of each first customer service for answering the second call.

[0036] Therefore, the system can fully consider the multi-dimensional capabilities of the customer service representative (including emotion value, call connection skill value) to adaptively adjust the call connection priority of the customer service representative. This means that when the emotion value of the customer service representative is lower than the average level, the system will automatically reduce its call connection priority, avoiding assigning calls to customer service representatives with poor emotions to reduce the negative impact on the customer experience. On the contrary, customer service representatives with higher emotion values will be given priority to receive calls, thereby maximizing the contribution of positive emotions to service quality and optimizing the overall service quality and customer experience.

[0037] Step S106, in response to the call connection request for the second call, determine the target customer service for answering the second call according to the call connection priority of each first customer service.

[0038] In the technical solution provided in the above step S106, when the system receives the call connection request for the second call, it can first query multiple first customer services that are currently idle, and call the previously calculated call connection priority information to sort all idle first customer services to obtain a sorted queue. Among them, customer services with higher priorities are ranked at the front of the queue, indicating that they are more suitable to connect to the upcoming second call. Furthermore, the system can select the customer service representative with the highest priority in the sorted queue as the target customer service to answer the second call, so as to ensure that customer service representatives with the best emotion state and high skill matching degree are given priority to receive call distribution, improving the call processing efficiency and customer service quality.

[0039] Based on the solution defined in the above steps S102 to S106, it can be learned that in the embodiment of the present application, by collecting and analyzing the first emotional state datasets of multiple first customer service representatives in the idle state during the first time period, the first customer evaluation values corresponding to the first incoming calls answered by each first customer service representative during the second time period, and using the pre-trained emotion evaluation model to dynamically determine the first emotion values of each first customer service representative during the first time period; determining the call answering priority for answering the second incoming call according to the call answering skill values of each first customer service representative and the first emotion values of each first customer service representative during the first time period; when receiving the answering request for the second incoming call, determining the target customer service representative for answering the second incoming call according to the call answering priorities of each first customer service representative. The entire solution realizes the technical effect of intelligent call distribution based on the emotional state of customer service representatives, achieves the purpose of optimizing customer service quality, improving customer experience satisfaction, and reasonably allocating human resources, and further solves the technical problem that the relevant call distribution technology ignores the analysis of the emotional state of customer service representatives, resulting in the call distribution result being unable to adapt to the actual working state of customer service representatives.

[0040] The following describes each step of the adaptive call distribution method based on customer service emotion in combination with a specific implementation process.

[0041] As an optional implementation manner, in the technical solution provided in the above step S102, the method may include:

[0042] The first step: Obtain the initial emotional state datasets of multiple first customer service representatives in the idle state during the first time period.

[0043] Among them, the above initial emotional state datasets include emotional state data of multiple first dimensions, and the first dimensions include but are not limited to: facial expressions, volume, pitch, semantics, blood pressure, heart rate, etc. Specifically, a camera can capture behavioral characteristics such as the facial expressions of the first customer service representative, a PCI voice card can record the voice characteristics of the first customer service representative (such as volume, rhythm, speech rate), and a smart watch can monitor the physiological indicators of the first customer service representative (such as blood pressure, heart rate, respiratory rate), etc.

[0044] The second step: Preprocess each initial emotional state dataset to obtain the first emotional state dataset of each first customer service representative during the first time period.

[0045] Among them, the above-mentioned preprocessing includes but is not limited to: data cleaning, standardization processing, format conversion, etc., to eliminate noise, handle missing values, and convert data formats, ensuring data quality and suitability for subsequent analysis. Among them: data cleaning can remove or correct inaccurate, incomplete, inappropriate or irrelevant data; format conversion is to convert emotion parameters using preset professional formulas, where the professional formulas include but are not limited to: heart rate variability rate, mean arterial pressure, sound pressure level, respiratory rate, body temperature and other related professional calculation formulas; while standardization processing can make data from different sources comparable on the same scale, facilitating subsequent model processing.

[0046] Step 3: For each first customer service agent, obtain the customer evaluation information corresponding to the first incoming call received by the first customer service agent during the second time period and the customer evaluation information corresponding to each of the multiple historical incoming calls received by the first customer service agent.

[0047] Among them, the above-mentioned customer evaluation information is usually collected through the IVR (Interactive Voice Response) system or online surveys after the call ends to intuitively reflect the service of the first customer service agent. Therefore, the customer evaluation information includes multiple evaluation information in the second dimension, and the second dimension includes but is not limited to: customer satisfaction, problem handling speed, problem resolution rate, etc.

[0048] Step 4: According to the preset first quantization rule table, determine the first quantization results of each evaluation information in the customer evaluation information corresponding to the first incoming call and the first quantization results of each evaluation information in the customer evaluation information corresponding to each historical incoming call respectively.

[0049] Among them, the above-mentioned first quantization rule table includes the first quantization rules corresponding to multiple evaluation information in the second dimension, and the first quantization rule is used to convert the fuzzy definition of the evaluation information into the corresponding numerical value. That is to say, the first quantization rule table defines the quantization standards for different evaluation information, ensuring the objectivity and comparability of the evaluation information.

[0050] Step 5: According to the first quantization results of each evaluation information in the customer evaluation information corresponding to the first incoming call and the first quantization results of each evaluation information in the customer evaluation information corresponding to each historical incoming call, determine the first customer evaluation value corresponding to the first incoming call according to the following formula:

[0051]

[0052] In the formula, X represents the first customer evaluation value, x i represents the first quantization result corresponding to the i-th type of customer evaluation information corresponding to the first incoming call, and n represents the number of information in the customer evaluation information corresponding to the first incoming call. respectively represent the maximum value and the minimum value of the first quantization result corresponding to the i-th type of customer evaluation information for multiple historical incoming calls, and α i represents the weight coefficient corresponding to the i-th type of customer evaluation information (which reflects the relative importance of the customer evaluation information in the customer evaluation value).

[0053] For example, if the customer evaluation information corresponding to the first incoming call answered by customer service representative A is: customer satisfaction is: "satisfied", problem resolution rate is "90%", and problem handling speed is "average". Then, according to the preset first quantization rule indicators, the above customer evaluation information is quantized. Among them, the customer satisfaction is quantized according to 4 levels of "bad attitude, average, satisfied, very satisfied", so the customer satisfaction of "satisfied" corresponding to customer service representative A answering the first incoming call is quantized to 8 points (with a full score of 10 points); the problem resolution rate can be quantized according to 10 levels of "0 - 10%, 10% - 20%,..., 80% - 90%, 90 - 100%", so the problem resolution rate of "90%" corresponding to customer service representative A answering the first incoming call can be quantized to 9 (with a full score of 10 points); the problem handling speed can be quantized according to 5 levels of "extremely slow, slower, average, faster, extremely fast", so the problem handling speed of "faster" corresponding to customer service representative A answering the first incoming call can be quantized to 6 (with a full score of 10 points).

[0054] And according to the above method, the customer evaluation information corresponding to the historical incoming calls of customer service representative A is quantized, and its historical minimum value of satisfaction is 5, and the maximum value is 10; the historical minimum value of the problem resolution rate is 6, and the maximum value is 10; the historical minimum value of the problem handling speed is 3, and the maximum value is 10; the historical minimum value of the problem handling speed is 5, and the maximum value is 10.

[0055] In addition, for each type of customer evaluation information, the weight value of customer satisfaction can be preset to 0.3, the weight value of the problem resolution rate to 0.4, and the weight value of the problem handling speed to 0.3.

[0056] Therefore, combining the above information and using the following formula to calculate the customer evaluation value corresponding to customer service representative A answering the first incoming call:

[0057]

[0058] Furthermore, as an optional implementation manner, in the technical solution provided in the above step S102, the call distribution system can obtain the call answering skill value of the first customer service according to the following method, including:

[0059] The first step: Obtain the call answering skill information of the first customer service.

[0060] Among them, the above answering skill information is usually stored in the skill file of the customer service representative and maintained by the human resources department or through self-assessment and training records. The skill file includes the training records, historical handling records, and on-the-job performance of the customer service representative, etc. Specifically, the answering skill information encompasses multiple skill assessment information in the third dimension, and the third dimension includes but is not limited to: the human resource status in the skill field to which the customer service belongs (such as shortage of human resources, sufficient human resources), and the maturity of skill mastery (such as beginner, intermediate, advanced).

[0061] Step 2: Determine the second quantization results corresponding to each skill assessment information in the answering skill information according to the preset second quantization rule table.

[0062] Among them, the above second quantization rule table also includes the second quantization rules corresponding to multiple skill assessment information in the third dimension, and the second quantization rule is used to convert the fuzzy definition of the skill assessment information into the corresponding numerical value. For example, the beginner, intermediate, and advanced levels of "maturity of skill mastery" are respectively assigned the values of 1, 2, and 3, and the shortage and sufficiency of "human resource status" are respectively assigned the values of 0.5 and 1. Through such quantization rules, the quantifiability and comparability of the skill information are ensured, which is convenient for subsequent analysis.

[0063] Step 3: According to the second quantization results corresponding to each skill assessment information in the answering skill information, determine the answering skill value of the first customer service according to the following formula:

[0064]

[0065] In the formula, Y represents the answering skill value, and y j represents the second quantization result of the i-th skill assessment information corresponding to the first customer service, m represents the number of pieces of information of the skill assessment information in the answering skill information, respectively represent the maximum value and the minimum value of the second quantization results of the i-th skill assessment information, and θ i represents the weight coefficient corresponding to the i-th skill assessment information (which reflects the relative importance of the skill assessment index in the answering skill value).

[0066] For example, if the human resource status in the skill field to which the customer service representative A belongs in the answering skill information is: normal, and the "maturity of skill mastery" is relatively proficient. The human resource status can usually be divided into extremely short, relatively short, normal, saturated, and redundant, and the second quantization rule corresponding to the human resource status is: "extremely short = 1, relatively short = 0.75, normal = 0.5, saturated = 0.25, redundant = 0"; the maturity of skill mastery can usually be divided into proficient, relatively proficient, and unskilled, and the second quantization rule corresponding to the maturity of skill mastery is: "proficient = 0.9, relatively proficient = 0.6, unskilled = 0.3".

[0067] Based on the above second quantization rule and the call - answering skill information of customer service representative A, the following two quantization results can be obtained: the quantization result of the human resource status is 0, and the quantization result of the skill mastery maturity is 2. At the same time, assume that the weight of the human resource status is 0.2 and the weight of the technical mastery maturity is 0.8. Then, based on the above information, the call - answering skill value of customer service representative A is calculated through the following formula:

[0068]

[0069] After determining the first customer evaluation value through the above steps, the system can then call the target emotion evaluation model to analyze the first emotion state dataset of the first customer service and the first customer evaluation value, so as to output the latest emotion value of the first customer service.

[0070] Optionally, the training process of the above - mentioned target emotion evaluation model includes:

[0071] The first step: Construct an initial emotion evaluation model. Among them, the construction of this initial emotion evaluation model is usually based on machine - learning algorithms such as neural networks, support vector machines, and decision trees. In the embodiment of the present application, a neural network model is preferably used. The model structure includes an input layer, multiple hidden layers, and an output layer. Among them, the input layer is used to receive input variables, so the input dimension of the input layer is set to be equal to the number of input variables; the output layer is used to output prediction results, so the number of neurons in the output layer takes a value of 1.

[0072] The second step: Obtain multiple groups of first training samples.

[0073] Among them, the above - mentioned first training samples include: the second emotion state dataset of the second customer answering calls during the first historical time period, the second customer evaluation value corresponding to the second customer answering the third incoming call during the second historical time period, and the second emotion value of the second customer during the first historical time period. Among them, both the first historical time period and the second historical time period are time periods within the first time period, and the second historical time period is a time period before the first historical time period, and the second customer evaluation value can be obtained by means of "expert scoring" or "self - evaluation by customer service".

[0074] The third step: For each group of first training samples, input the second emotion state dataset and the second customer evaluation value in the first training sample into the initial emotion evaluation model to obtain the predicted emotion value output by the initial emotion evaluation model.

[0075] Specifically, this step is implemented through the forward - propagation algorithm. The input data is transmitted through the network, undergoes linear transformation and non - linear activation in each layer until the output layer, generating a predicted emotion value. Among them, for a group of first training samples, its corresponding predicted emotion value can be obtained according to the following formula:

[0076] For the i-th layer (i = 1, 2, 3, ..., I, where I represents the last layer):

[0077] z (i) = W (i) a (i-1) + b (i)

[0078] a (i) = f(z (i) )

[0079] For the output layer (assuming a linear activation):

[0080]

[0081] Wherein, W (i) represents the weight matrix between all nodes of the i-th layer and the previous layer, b (i) represents the bias vector corresponding to the i-th layer, and a (i) represents the activation value corresponding to the i-th layer.

[0082] Step 4: Construct a first objective loss function of the initial emotion evaluation model based on the second emotion value and the predicted emotion value in each group of first training samples, and update the model parameters of the initial emotion evaluation model by minimizing the first objective loss function until the preset first convergence condition is satisfied, so as to obtain the trained target emotion evaluation model.

[0083] Among them, the above first objective loss function is used to measure the difference between the model prediction value and the true value. Common loss functions include mean square error (MSE), cross-entropy loss, etc. In the embodiment of the present application, taking the mean square error as an example, the expression of the first objective loss function corresponding to the predicted emotion value and the second emotion value of a group of first training samples can be written as:

[0084]

[0085] Where n represents the total number of first training samples, V i represents the second emotion value in the i-th group of first training samples, represents the predicted emotion value corresponding to the i-th group of first training samples.

[0086] Therefore, for all the first training samples, the sum of the loss functions represents the quality of the overall prediction ability of the model. Thus, the goal of model optimization is to minimize this sum. Generally, the model parameters (including the weight matrix W and the bias vector b) can be updated according to the gradient of the first objective loss function through the backpropagation algorithm to reduce the value of the loss function. This process usually involves selecting an optimization algorithm, such as Stochastic Gradient Descent (SGD), Adam, etc., to iteratively adjust the parameters. Taking gradient descent as an example, the model parameters can be updated according to the following formula:

[0087]

[0088] where γ represents the learning rate.

[0089] In each iteration, the system calculates the new loss value and checks whether the preset first convergence condition is met, such as the change in the loss value being less than a threshold or reaching the maximum number of iterations. When the update iteration of the model parameters meets the first convergence condition, that is, the value of the loss function reaches the minimum or the change tends to be stable, the model training is completed, and the trained target emotion evaluation model is obtained. The model parameters at this time are considered to be optimal and can accurately predict the emotion state of the customer service representative when answering calls based on the input second emotion state dataset and the second customer evaluation value.

[0090] Through the above process, the target emotion evaluation model is trained and optimized, which can accurately evaluate the emotion state of the customer service representative, provide emotion intelligence for call distribution, ensure that calls are assigned to the customer service representative with the best emotion state, and thus improve the operation efficiency and customer satisfaction of the call center.

[0091] It should be noted that the above target emotion evaluation model uses a supervised learning algorithm for model adaptive learning. This means that with the completion of each call, the system adjusts the parameters of the emotion evaluation model according to the latest emotion state of the customer service representative and the customer evaluation situation, thereby improving the accuracy of the model's prediction of the customer service representative's emotion. This learning ability enables the system to gradually optimize the call distribution strategy over time and with the accumulation of data to adapt to the changes in the customer service representative's emotion and the fluctuations in customer expectations.

[0092] As an alternative implementation, in the technical solution provided in the above step S104, the call distribution system can determine the answering priority of the first customer service representative answering the second incoming call according to the answering skill value and the first emotion value, and according to the following formula:

[0093]

[0094] where P k represents the kth first customer service representative in the idle state, and V kDenote the first emotional value when the k-th first customer service in the idle state answers the first incoming call, Y k Denote the call answering skill value of the k-th first customer service in the idle state, ω respectively denote the weight coefficients.

[0095] Specifically, when the call distribution system processes an incoming call, it usually goes through the following processing links:

[0096] (1) The call distribution system first receives the incoming call and preliminarily classifies the current incoming call through preset rules, such as whether it is a repeated call, whether it is a call from a high-value customer, etc.;

[0097] (2) Allocate the current incoming call to the appropriate queue through the incoming call number, incoming time, incoming channel, etc., so as to direct the incoming call to the most suitable customer service representative or service queue;

[0098] (3) Adjust and set the priority of the current incoming call through the number attributes or queue load conditions;

[0099] (4) If there is no available customer service representative to immediately answer the incoming call, the system places the incoming call in the queue to wait and queues according to the first-come, first-served principle. At the same time, it may play waiting music or pre-recorded information to notify the customer of the waiting time;

[0100] (5) The system allocates the call traffic to the most suitable customer service representative according to the nature and priority of the incoming call, as well as the skills and available status of the customer service representative;

[0101] (6) Once the best routing scheme is determined, the system will perform the routing operation and allocate the call traffic request in the queue to the selected customer service representative.

[0102] Therefore, the call answering priority determined in step S104 above can be applied to the above link (5) to improve resource matching through the call answering priority of each customer service representative.

[0103] As an optional implementation manner, in the technical solution provided in step S106 above, the call distribution system can determine the target customer service for answering the second incoming call according to the following two methods, including:

[0104] Method 1: Use the call answering priority as an independent strategy for routing decision-making.

[0105] Optionally, the call distribution system can use the first customer service with the highest call answering priority as the target customer service for answering the second incoming call.

[0106] Therefore, the corresponding expression of this method can be written as:

[0107] M = argmax i (Pi )

[0108] where \(i = \{1, 2, \ldots, n\}\), and \(P\) i represents the call answering priority of the \(i\)-th first customer service in the sorting queue obtained by sorting the call answering priorities of each first customer service in descending order.

[0109] Method 2: Combine the call answering priority as a variable such as a coefficient or weight with other routing strategies.

[0110] Optionally, the call distribution system can determine the target customer service for answering the second incoming call based on the call answering priorities of each first customer service and in combination with a preset distribution strategy. Among them, the distribution strategy includes at least one of the following: the least idle time strategy for allocating new incoming calls to the target customer service with the shortest processing time for handling the current incoming call or the longest idle time, and the skill routing strategy for allocating new incoming calls to the target customer service with the highest call answering skill score.

[0111] Specifically:

[0112] In the case where the distribution strategy is the least idle time strategy, the call distribution system can determine the processing time of each first customer service for handling the first incoming call or the idle time in the idle state, and use the first customer service with the shortest processing time or the longest idle time and the highest call answering priority as the target customer service for answering the second incoming call. Therefore, the expression corresponding to this method can be written as:

[0113] M = argmax i (T i δP i )

[0114] where \(i = \{1, 2, \ldots, n\}\), and \(P\) i represents the call answering priority of the \(i\)-th first customer service in the sorting queue obtained by sorting the call answering priorities of each first customer service in descending order, \(\delta\in(0, 1)\) represents a weight factor or adjustment coefficient, and \(T\) i represents the cumulative call time or idle time of the \(i\)-th first customer service in the sorting queue.

[0115] In the case where the distribution strategy is the skill routing strategy, the call distribution system can determine the call answering skill scores of each first customer service, and use the first customer service with the highest call answering skill score and the highest call answering priority as the target customer service for answering the second incoming call.

[0116] Through the above solution, it is not difficult to see that compared with the existing call distribution methods, the adaptive call distribution method based on customer service emotions provided by the embodiments of the present application has the following technical advantages:

[0117] (1) By introducing the "emotional value of the customer service" as a consideration factor, the solution of this application makes up for the deficiency in the prior art of neglecting the emotional state of customer service representatives. The addition of the emotional value can identify and preferentially allocate customer service representatives with better emotional states, avoiding customer service representatives with poor moods from affecting the customer experience, thereby improving the overall service quality and customer satisfaction.

[0118] (2) In addition to considering the emotional value of the customer service representative, the solution of this application also considers dimensions such as the call answering skill value and customer evaluation value of the customer service representative. By comprehensively evaluating multi-dimensional data to calculate the call answering priority, this makes the call traffic allocation more comprehensive and scientific, avoiding over-reliance on customer service representatives with better emotional states, ensuring the reasonable allocation and utilization of the entire call center resources, thereby improving the service quality while also improving the operation efficiency and reducing the operation cost.

[0119] (3) The solution of this application adopts an adaptive learning algorithm, which can continuously optimize and adjust the emotion prediction model according to new data, and dynamically calculate the emotional state of each customer service representative. This mechanism improves the prediction accuracy and flexibility of the model, ensures that the call traffic allocation strategy can be adjusted in a timely manner with the change of the emotional state of the customer service representative, and enhances the intelligence and adaptability of the system.

[0120] Embodiment 2

[0121] According to the embodiment of this application, there is also provided an adaptive call traffic allocation device based on customer service emotion for implementing the adaptive call traffic allocation method based on customer service emotion in Embodiment 1, as Figure 2 shown. The adaptive call traffic allocation device based on customer service emotion at least includes: an acquisition module 22, a determination module 24, and an allocation module 26, where:

[0122] The acquisition module 22 is configured to acquire a first emotional state data set of a plurality of first customer services in an idle state during a first time period, a first customer evaluation value corresponding to the first incoming call answered by each first customer service during a second time period, and acquire the call answering skill value of each first customer service, where the second time period is a time period before the first time period;

[0123] The determination module 24 is configured to, for each first customer service, analyze the first emotional state data set and the first customer evaluation value by using a pre-trained target emotion evaluation model to obtain a first emotional value of the first customer service during the first time period, and determine the call answering priority of the first customer service for answering the second incoming call according to the call answering skill value and the first emotional value of the first customer service;

[0124] The allocation module 26 is configured to, in response to a call answering request for the second incoming call, determine a target customer service for answering the second incoming call according to the call answering priorities of each first customer service.

[0125] The functions of each module of the adaptive call distribution device based on customer service emotions will be described below in combination with specific implementation processes.

[0126] As an optional implementation manner, the obtaining module 22 may obtain the first emotion state data set and the first customer evaluation value of each first customer service according to the following method, including:

[0127] The first step: Obtain the initial emotion state data sets of multiple first customer services in the idle state during the first time period respectively.

[0128] Among them, the above initial emotion state data set includes emotion state data of multiple first dimensions, and the first dimension includes but is not limited to: expression, volume, pitch, semantics, blood pressure, heart rate, etc. Specifically, the camera can capture the behavioral characteristics of the first customer service such as facial expressions, the PCI voice card can record the voice characteristics of the first customer service (such as volume, rhythm, speech rate), and the smart watch can monitor the physiological indicators of the first customer service (such as blood pressure, heart rate, breathing rate), etc.

[0129] The second step: Preprocess each initial emotion state data set to obtain the first emotion state data set of each first customer service during the first time period.

[0130] Among them, the above preprocessing includes but is not limited to: data cleaning, normalization processing, format conversion, etc., to eliminate noise, process missing values, and convert data formats to ensure data quality and suitability for subsequent analysis.

[0131] The third step: For each first customer service, obtain the customer evaluation information corresponding to the first incoming call received by the first customer service during the second time period and the customer evaluation information corresponding to each of the multiple historical incoming calls received by the first customer service respectively.

[0132] Among them, the above customer evaluation information is usually collected through the IVR system or online survey after the call ends to intuitively reflect the service of the first customer service. Therefore, the customer evaluation information includes multiple second-dimension evaluation information, and the second dimension includes but is not limited to: customer satisfaction, problem handling speed, problem resolution rate, etc.

[0133] The fourth step: According to the preset first quantization rule table, determine the first quantization results of each evaluation information in the customer evaluation information corresponding to the first incoming call and the first quantization results of each evaluation information in the customer evaluation information corresponding to each historical incoming call respectively.

[0134] Among them, the above first quantization rule table includes first quantization rules corresponding to multiple pieces of evaluation information in the second dimension, and the first quantization rule is used to convert the fuzzy definition of the evaluation information into corresponding numerical values. That is to say, the first quantization rule table defines the quantization criteria for different evaluation information, ensuring the objectivity and comparability of the evaluation information.

[0135] Fifth step: According to the first quantization results of each piece of evaluation information in the customer evaluation information corresponding to the first incoming call and the first quantization results of each piece of evaluation information in the customer evaluation information corresponding to each historical incoming call, determine the first customer evaluation value corresponding to the first incoming call according to the following formula:

[0136]

[0137] In the formula, X represents the first customer evaluation value, and x i represents the first quantization result corresponding to the i-th type of customer evaluation information corresponding to the first incoming call, n represents the number of pieces of information in the customer evaluation information corresponding to the first incoming call, respectively represent the maximum value and the minimum value of the first quantization results of the i-th type of customer evaluation information corresponding to multiple historical incoming calls, and α i represents the weight coefficient corresponding to the i-th type of customer evaluation information (which reflects the relative importance of the customer evaluation information in the customer evaluation value).

[0138] As an optional implementation manner, the obtaining module 22 can also obtain the incoming call skill value of the first customer service according to the following method, including:

[0139] First step: Obtain the incoming call skill information of the first customer service.

[0140] Among them, the above incoming call skill information is usually stored in the skill file of the customer service representative and is maintained by the human resources department or through self-evaluation and training records. The skill file includes the training records, historical processing records, and on-the-job performance, etc., of the customer service representative.

[0141] Second step: Determine the second quantization results corresponding to each skill evaluation information in the incoming call skill information according to the preset second quantization rule table.

[0142] Among them, the above second quantization rule table also includes second quantization rules corresponding to multiple pieces of skill evaluation information in the third dimension, and the second quantization rule is used to convert the fuzzy definition of the skill evaluation information into corresponding numerical values. For example, assign 1, 2, and 3 to the beginner, intermediate, and advanced levels of "skill mastery maturity" respectively, and assign 0.5 and 1 to the shortage and sufficiency of "manpower status" respectively. Through such quantization rules, the quantizability and comparability of the skill information are ensured, facilitating subsequent analysis.

[0143] Step 3: According to the second quantization results corresponding to each skill evaluation information in the answering skill information, determine the answering skill value of the first customer service according to the following formula:

[0144]

[0145] In the formula, Y represents the answering skill value, and y j represents the second quantization result of the i-th skill evaluation information corresponding to the first customer service, m represents the number of pieces of skill evaluation information in the answering skill information, respectively represent the maximum value and the minimum value of the second quantization results of the i-th skill evaluation information, and θ i represents the weight coefficient corresponding to the i-th skill evaluation information (which reflects the relative importance of the skill evaluation index in the answering skill value).

[0146] Optionally, the determination module 24 can pre-train the target emotion evaluation model according to the following method, including:

[0147] Step 1: Construct an initial emotion evaluation model. Among them, the construction of the initial emotion evaluation model is usually based on machine learning algorithms such as neural networks, support vector machines, and decision trees. In the embodiments of the present application, a neural network model is preferably used. The model structure includes an input layer, multiple hidden layers, and an output layer. Among them, the input layer is used to receive input variables, so the input dimension of the input layer is set to be equal to the number of input variables; the output layer is used to output prediction results, so the number of neurons in the output layer takes the value of 1.

[0148] Step 2: Obtain multiple groups of first training samples. Among them, the above first training samples include: the second emotion state data set of the second customer answering calls during the first historical period, the second customer evaluation value corresponding to the second customer answering the third incoming call during the second historical period, and the second emotion value of the second customer during the first historical period.

[0149] Step 3: For each group of first training samples, input the second emotion state data set and the second customer evaluation value in the first training sample into the initial emotion evaluation model to obtain the predicted emotion value output by the initial emotion evaluation model.

[0150] Step 4: Construct the first target loss function of the initial emotion evaluation model according to the second emotion value and the predicted emotion value in each group of first training samples, and update the model parameters of the initial emotion evaluation model by minimizing the first target loss function until the preset first convergence condition is satisfied, so as to obtain the trained target emotion evaluation model.

[0151] It should be noted that the above-mentioned target emotion evaluation model uses a supervised learning algorithm for model adaptive learning. Therefore, after the first emotion value of the first customer service representative after answering the first incoming call is output by the target emotion evaluation model, it is necessary to continue to update the model parameters of the target emotion evaluation model by using the first emotion state data set of the first customer service representative, the first customer evaluation value, and the first emotion value of the first customer service representative when answering the first incoming call, so as to improve the accuracy of the model in predicting the emotions of customer service representatives.

[0152] Furthermore, the determination module 24 can determine the call answering priority of the first customer service representative for answering the second incoming call according to the call answering skill value and the first emotion value, and according to the following formula:

[0153]

[0154] In the formula, P k represents the kth first customer service representative in the idle state, V k represents the first emotion value of the kth first customer service representative in the idle state when answering the first incoming call, Y k represents the call answering skill value of the kth first customer service representative in the idle state, ω respectively represent weight coefficients.

[0155] Optionally, the allocation module 26 can determine the target customer service representative for answering the second incoming call according to the following two methods, including:

[0156] Method 1: Use the call answering priority as an independent strategy for routing decisions.

[0157] Optionally, the call traffic distribution system can use the first customer service representative with the highest call answering priority as the target customer service representative for answering the second incoming call.

[0158] Method 2: Combine the call answering priority as a coefficient, weight and other variables with other routing strategies.

[0159] Optionally, the call traffic distribution system can determine the target customer service representative for answering the second incoming call based on the call answering priorities of each first customer service representative and in combination with a preset allocation strategy. Among them, the allocation strategy includes at least one of the following: the least idle time strategy for allocating new incoming calls to the target customer service representative with the shortest processing time for handling the current incoming call or the longest idle time in the idle state, and the skill routing strategy for allocating new incoming calls to the target customer service representative with the highest call answering skill score.

[0160] Specifically, in the case where the allocation strategy is the least idle time strategy, the call traffic distribution system can determine the processing time of each first customer service representative for handling the first incoming call or the idle time in the idle state, and use the first customer service representative with the shortest processing time or the longest idle time and the highest call answering priority as the target customer service representative for answering the second incoming call.

[0161] Specifically, in the case where the allocation policy is a skill routing policy, the call distribution system can determine the call answering skill scores of each first customer service agent, and use the first customer service agent with the highest call answering skill score and the highest call answering priority as the target customer service agent for answering the second incoming call.

[0162] It should be noted that each module in the adaptive call distribution device based on customer service agent emotion in the embodiments of the present application corresponds to each implementation step in the adaptive call distribution method based on customer service agent emotion in Embodiment 1. Since detailed descriptions have been made in Embodiment 1, details not shown in this embodiment can be referred to Embodiment 1 and will not be elaborated here.

[0163] Embodiment 3

[0164] According to an embodiment of the present application, there is also provided a computer program product, which includes a computer program. When the computer program is executed by a processor, it implements the adaptive call distribution method based on customer service agent emotion in Embodiment 1.

[0165] According to an embodiment of the present application, there is also provided a non-volatile storage medium, which includes a stored computer program. The device where the non-volatile storage medium is located executes the adaptive call distribution method based on customer service agent emotion in Embodiment 1 by running the computer program.

[0166] According to an embodiment of the present application, there is also provided a processor, which is used to run a computer program. When the computer program runs, it executes the adaptive call distribution method based on customer service agent emotion in Embodiment 1.

[0167] According to an embodiment of the present application, there is also provided an electronic device, which includes: a memory and a processor. A computer program is stored in the memory, and the processor is configured to execute the adaptive call distribution method based on customer service agent emotion in Embodiment 1 through the computer program.

[0168] Optionally, when the computer program runs, it executes the following steps: obtaining the first emotion state data sets of multiple first customer service agents in an idle state during a first time period and the first customer evaluation values corresponding to the first incoming calls answered by each first customer service agent during a second time period, and obtaining the call answering skill values of each first customer service agent, where the second time period is the time period before the first time period; for each first customer service agent, analyzing the first emotion state data set and the first customer evaluation value by using a pre-trained target emotion evaluation model to obtain the first emotion value of the first customer service agent during the first time period, and determining the call answering priority of the first customer service agent for answering the second incoming call based on the call answering skill value and the first emotion value; in response to a call answering request for the second incoming call, determining the target customer service agent for answering the second incoming call based on the call answering priorities of each first customer service agent.

[0169] As an alternative implementation, the above-mentioned electronic device may exist in the form of a mobile terminal, a computer terminal, or a similar computing device. Figure 3 The following shows a hardware structure block diagram of an electronic device for implementing an adaptive call traffic allocation method based on customer service emotions. As Figure 3 shown, the electronic device 30 may include one or more processors 302 (shown as 302a, 302b, ……, 302n in the figure) (the processor 302 may include, but is not limited to, a processing device such as a microprocessor MCU or a programmable logic device FPGA), a memory 304 for storing data, and a transmission device 306 for communication functions. In addition, it may further include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports in the BUS bus), a network interface, a power supply, and / or a camera. Those of ordinary skill in the art can understand that Figure 3 the structure shown is only schematic and does not limit the structure of the above-mentioned electronic device. For example, the electronic device 30 may further include more or fewer components than those Figure 3 shown, or have a different configuration from that Figure 3 shown.

[0170] It should be noted that the above one or more processors 302 and / or other data processing circuits are generally referred to as "data processing circuits" in this article. The data processing circuit may be embodied in software, hardware, firmware, or any combination thereof, in whole or in part. In addition, the data processing circuit may be a single independent processing module, or be incorporated in whole or in part into any one of the other elements in the electronic device 30. As involved in the embodiments of the present application, the data processing circuit is a kind of processor control (such as the selection of a variable resistor terminal path connected to an interface).

[0171] The memory 304 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the adaptive call traffic allocation method based on customer service emotions in the embodiments of the present application. The processor 302 executes various functional applications and data processing by running the software programs and modules stored in the memory 304, that is, implements the vulnerability detection method of the above-mentioned application program. The memory 304 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memories, or other non-volatile solid-state memories. In some instances, the memory 304 may further include a memory remotely set relative to the processor 302, and these remote memories may be connected to the electronic device 30 through a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an enterprise internal network, a local area network, a mobile communication network, and their combinations.

[0172] The transmission device 306 is used to receive or send data via a network. Specific examples of the above-mentioned network may include a wireless network provided by the communication provider of the electronic device 30. In one example, the transmission device 306 includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 306 can be a Radio Frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0173] The display can be, for example, a touch-screen liquid crystal display (LCD), which enables the user to interact with the user interface of the electronic device 30.

[0174] The above-mentioned serial numbers of the embodiments are only for description and do not represent the advantages or disadvantages of the embodiments.

[0175] In the above embodiments of the present application, the descriptions of the respective embodiments have their own emphases. For the parts not detailed in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0176] In several embodiments provided in the present application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only illustrative. For example, the division of units can be a logical function division. In actual implementation, there can be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces. The indirect couplings or communication connections of units or modules can be in electrical or other forms.

[0177] The units described as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they can be located in one place or distributed to multiple units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0178] In addition, the functional units in each embodiment of the present application can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above-mentioned integrated units can be implemented in the form of hardware or in the form of software functional units.

[0179] When the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods of various embodiments of this application. The foregoing storage medium includes: various media that can store program codes, such as USB flash drives, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), mobile hard disks, magnetic disks, or optical discs.

[0180] The above are only the preferred embodiments of this application. It should be noted that for those of ordinary skill in the art of this technology, without departing from the principle of this application, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of this application.

Claims

1. An adaptive traffic distribution method based on customer service emotions, characterized in that: include: Obtain a first emotional state data set of each of a plurality of first customer service representatives in an idle state within a first time period and a first customer evaluation value corresponding to each of the first customer service representatives answering a first incoming call within a second time period, and obtain a call answering skill value of each of the first customer service representatives, wherein the second time period is a time period before the first time period; For each of the first customer service representatives, the first emotional state data set and the first customer evaluation value are analyzed using a pre-trained target emotion assessment model to obtain a first emotional value of the first customer service representative in a first time period, and a call answering priority of the first customer service representative for answering the second incoming call is determined according to the call answering skill value and the first emotional value; In response to the answering request for the second incoming call, a target customer service for answering the second incoming call is determined according to the answering priorities of the first customer service staff.

2. The method according to claim 1, characterized in that Acquiring a first emotional state data set of each of a plurality of first customer service representatives in an idle state within a first time period and a first customer evaluation value corresponding to each of the first customer service representatives answering a first incoming call within a second time period, including: Acquire an initial emotional state data set of each of the plurality of first customer service representatives in an idle state within a first time period, wherein the initial emotional state data set includes emotional state data of a plurality of first dimensions, and the first dimensions include at least one of the following: expression, volume, musicality, semantics, blood pressure, and heart rate; Preprocessing each of the initial emotional state data sets to obtain a first emotional state data set of each of the first customer service staff in a first time period, wherein the preprocessing includes at least one of the following: data cleaning and standardization processing; For each of the first customer service representatives, obtaining customer evaluation information corresponding to the first incoming call answered by the first customer service representative in the second time period and customer evaluation information corresponding to multiple historical incoming calls answered by the first customer service representative, wherein the customer evaluation information includes evaluation information of multiple second dimensions, and the second dimension includes at least one of the following: customer satisfaction, problem handling speed, and problem resolution rate; According to a preset first quantization rule table, first quantization results of each evaluation information in the customer evaluation information corresponding to the first incoming call and first quantization results of each evaluation information in the customer evaluation information corresponding to each of the historical incoming calls are respectively determined, wherein the first quantization rule table includes first quantization rules corresponding to each of the evaluation information of the second dimension, and the first quantization rules are used to convert fuzzy definitions of the evaluation information into corresponding numerical values; According to the first quantized result of each evaluation information in the customer evaluation information corresponding to the first incoming call and the first quantized result of each evaluation information in the customer evaluation information corresponding to each of the historical incoming calls, the first customer evaluation value corresponding to the first incoming call is determined according to the following formula: In the formula, X represents the first customer evaluation value, x i represents the first quantization result corresponding to the i-th type of customer evaluation information corresponding to the first incoming call, n represents the number of pieces of customer evaluation information corresponding to the first incoming call, respectively represent the maximum value and the minimum value of the first quantization result of the i-th type of customer evaluation information corresponding to the plurality of historical calls, α i Represents the weight coefficient corresponding to the i-th category of customer evaluation information.

3. The method according to claim 1, characterized in that Obtaining the first customer service's answering skill value, including: Acquiring the answering skill information of the first customer service, wherein the answering skill information includes skill evaluation information of multiple third dimensions, and the third dimension includes at least one of the following: the human resources status and skill mastery maturity of the skill field to which the customer service belongs; Determine a second quantization result corresponding to each skill evaluation information in the answering skill information according to a preset second quantization rule table, wherein the second quantization rule table includes second quantization rules corresponding to each of the skill evaluation information in the third dimension, and the second quantization rule is used to convert the fuzzy definition of the skill evaluation information into a corresponding numerical value; According to the second quantification result corresponding to each skill evaluation information in the answering skill information, the answering skill value of the first customer service is determined according to the following formula: Where, Y represents the call-handling skill value, y j represents the second quantization result of the i-th skill evaluation information corresponding to the first customer service, m represents the number of skill evaluation information in the answering skill information, represents the maximum value of the second quantization result and the minimum value of the second quantization result of the i-th skill evaluation information, θ i Represents the weight coefficient corresponding to the i-th skill evaluation information.

4. The method according to claim 1, characterized in that: The training process of the target emotion assessment model includes: Constructing an initial emotion assessment model; Acquire multiple groups of first training samples, wherein the first training samples include: a second emotional state data set of a second customer answering a call in a first historical time period, a second customer evaluation value corresponding to a third call answered by the second customer in the second historical time period, and a second emotional value of the second customer in the first historical time period, wherein the second historical time period is a time period before the first historical time period; For each group of the first training samples, inputting the second emotional state data set and the second customer evaluation value in the first training samples into the initial emotion evaluation model to obtain a predicted emotion value output by the initial emotion evaluation model; A first target loss function of the initial emotion assessment model is constructed based on the second emotion value and the predicted emotion value in each group of the first training samples, and the model parameters of the initial emotion assessment model are updated by minimizing the first target loss function until a preset first convergence condition is met, thereby obtaining the trained target emotion assessment model.

5. The method according to claim 1, characterized in that Determining the answering priority of the first customer service for answering the second incoming call according to the answering skill value and the first emotion value includes: The answering priority of the first customer service for answering the second incoming call is determined according to the answering skill value and the first emotion value and the following formula: Where P k represents the kth first customer service in idle state, V k represents the first emotion value of the kth first customer service representative in the idle state when answering the first incoming call, Y k Indicates the call-answering skill value of the kth first customer service representative who is in an idle state. ω represents the weight coefficient.

6. The method according to claim 1, characterized in that Determining a target customer service for answering the second incoming call according to the answering priorities of each of the first customer service staff includes: The first customer service representative with the highest call answering priority is selected as the target customer service representative for answering the second incoming call; Alternatively, based on the answering priorities of each of the first customer services and in combination with a preset allocation strategy, a target customer service who answers the second call is determined, wherein the allocation strategy includes at least one of the following: a minimum idle time strategy for assigning new calls to the target customer service who has the shortest processing time for handling the current call or is in the longest idle state, and a skill routing strategy for assigning new calls to the target customer service with the highest answering skill score.

7. The method according to claim 6, characterized in that Determining a target customer service for answering the second incoming call based on the answering priorities of the first customer service staff and in combination with a preset allocation strategy includes: When the allocation strategy is the minimum idle time strategy, determining the processing time of each of the first customer service staff for processing the first incoming call or the idle time in an idle state, and selecting the first customer service staff with the shortest processing time or the longest idle time and the highest call answering priority as the target customer service staff for answering the second incoming call; When the allocation strategy is the skill routing strategy, the answering skill scores of the first customer services are determined, and the first customer service with the highest answering skill score and the highest answering priority is selected as the target customer service for answering the second incoming call.

8. An adaptive call distribution system based on customer service emotions, characterized in that: include: an acquisition module, configured to acquire a first emotional state data set of each of a plurality of first customer service representatives in an idle state within a first time period and a first customer evaluation value corresponding to each of the first customer service representatives answering a first incoming call within a second time period, and to acquire a call answering skill value of each of the first customer service representatives, wherein the second time period is a time period before the first time period; a determination module, configured to analyze the first emotional state data set and the first customer evaluation value for each of the first customer service personnel using a pre-trained target emotion evaluation model to obtain a first emotional value of the first customer service personnel in a first time period, and determine a call answering priority of the first customer service personnel for answering the second incoming call according to the call answering skill value of the first customer service personnel and the first emotional value; The allocation module is used to respond to the answering request for the second incoming call and determine the target customer service to answer the second incoming call according to the answering priority of each of the first customer service.

9. A computer program product, characterized in that include: A computer program, wherein when the computer program is executed by a processor, the method for adaptively allocating call traffic based on customer service emotions as described in any one of claims 1 to 7 is implemented.

10. An electronic device, characterized in that: include: A memory and a processor, wherein the memory stores a computer program, and the processor is configured to execute the adaptive call distribution method based on customer service emotions as described in any one of claims 1 to 7 through the computer program.

Images