Customer service seat matching method and device, electronic equipment and storage medium

Abstract

The invention provides a customer service seat matching method and device, electronic equipment and a storage medium. The method comprises the following steps: acquiring current dialogue data of a user; based on a trained large language model, analyzing the current dialogue data to obtain user key features of the user; based on the user key features and the current index value of the agent key index of each customer service agent, the customer service agent matched with the user is determined, and the current index value is updated based on the historical service record of the customer service agent. According to the customer service seat matching method and device, the electronic equipment and the storage medium provided by the invention, the user key features of the user are accurately captured by utilizing the powerful data processing and learning capabilities of the trained large language model, the current index values of the seat key indexes of the customer service seats are updated in time, and the user experience is improved. Multi-dimensional data of user key features and seat key indexes are comprehensively evaluated, optimal matching is achieved, and intelligent and personalized development of customer service distribution is promoted.

Description

Technical Field

[0001] The present invention relates to the technical field of customer service centers, and in particular to a customer service agent matching method, device, electronic equipment and storage medium. Background Art

[0002] With the diversification of customer service needs, how to quickly and accurately match customer demands with the most appropriate customer service agents has become the key to improving service quality.

[0003] Existing customer service allocation systems mostly adopt static or simple dynamic strategies, often focusing only on a single dimension of data (such as waiting time), without fully considering the actual capabilities and real-time status of customer service representatives. As a result, customer requests cannot always be assigned to the most appropriate customer service seat, which not only reduces service efficiency but also affects customer satisfaction. Summary of the invention

[0004] The present invention provides a customer service agent matching method, device, electronic device and storage medium, which are used to solve the defects of static or simple dynamic customer service allocation strategies in the prior art that lead to poor service efficiency and customer satisfaction.

[0005] The present invention provides a customer service agent matching method, comprising: Get the user's current conversation data; Analyze the current conversation data based on the trained large language model to obtain key user features of the user; Based on the key features of the user and the current indicator values ​​of the key indicators of each customer service agent, a customer service agent matching the user is determined, wherein the current indicator values ​​are updated based on the historical service records of the customer service agent.

[0006] According to the customer service agent matching method provided by the present invention, the current conversation data is analyzed based on the trained large language model to obtain the user's key features, including: generating an analysis input text based on the analysis prompt information and the current conversation data, wherein the analysis prompt information includes instructions for determining key features of the user; The analyzed input text is input into the large language model to obtain the user key features output by the large language model.

[0007] According to the customer service agent matching method provided by the present invention, the analysis prompt information also includes question information and expected output for the key features of the user.

[0008] According to the customer service agent matching method provided by the present invention, the customer service agent matching the user is determined based on the key features of the user and the current indicator values ​​of the key indicators of each customer service agent, including: Based on the similarity between the key features of the user and the key indicators of the agent, and the current indicator value, determining a matching score between the customer and each of the customer service agents; A customer service agent matching the user is determined based on the matching score.

[0009] According to the customer service agent matching method provided by the present invention, the matching score between the customer and each of the customer service agents is determined based on the similarity between the key features of the user and the key indicators of the agent, as well as the current indicator value, including: Based on the weight value of each user's key feature, the similarity and the current index value, determine the initial matching score between the customer and each customer service agent, wherein the weight value represents the importance of the user's key feature for agent matching; The matching score is determined based on the current indicator value of the service satisfaction rate indicator in the agent key indicators and the initial matching score.

[0010] According to the customer service agent matching method provided by the present invention, after determining the customer service agent matching the user, the method further includes: Receiving a service evaluation from the user regarding the customer service agent; Based on the service evaluation, the current indicator value of the key indicator of the customer service agent is updated.

[0011] According to the customer service agent matching method provided by the present invention, the user key characteristics include at least one of user demands, emotional state, and urgency; the agent key indicators include at least one of skill level, stress resistance, emotional response ability, and service satisfaction rate.

[0012] The present invention also provides a customer service agent matching device, comprising: A data acquisition unit, used to acquire the user's current conversation data; A data analysis unit, configured to analyze the current conversation data based on a trained large language model to obtain key user features of the user; The agent matching unit is used to determine a customer service agent that matches the user based on the user key features and the current indicator value of the agent key indicator of each customer service agent, wherein the current indicator value is updated based on the historical service record of the customer service agent.

[0013] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, any of the customer service agent matching methods described above is implemented.

[0014] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the customer service agent matching method as described in any one of the above is implemented.

[0015] The present invention also provides a computer program product, including a computer program, wherein when the computer program is executed by a processor, the computer program implements any of the customer service agent matching methods described above.

[0016] The customer service agent matching method, device, electronic device and storage medium provided by the present invention utilize the powerful data processing and learning capabilities of the trained large-scale language model to accurately capture the user's key features. Compared with simple keyword matching to determine the user's service needs, it can accurately extract the user's key features in multiple dimensions; by timely updating the current indicator values ​​of the customer service agent's key indicators, comprehensively evaluating the user's key features and the multi-dimensional data of the agent's key indicators, optimal matching is achieved, and customer service allocation is promoted to develop towards intelligence and personalization. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0018] Figure 1 This is one of the flow charts of the customer service agent matching method provided by the present invention.

[0019] Figure 2 It is a flowchart of step 120 in the customer service agent matching method provided by the present invention.

[0020] Figure 3 This is one of the flowchart diagrams of step 130 in the customer service agent matching method provided by the present invention.

[0021] Figure 4 This is the second flowchart of step 130 in the customer service agent matching method provided by the present invention.

[0022] Figure 5 This is the second flow chart of the customer service agent matching method provided by the present invention.

[0023] Figure 6It is a structural schematic diagram of the customer service seat matching device provided by the present invention.

[0024] Figure 7 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0025] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0026] In the current traditional customer service agent matching process, the main steps usually include the following aspects: ① User request access: When a user needs service, he / she submits a service request to the customer service center via telephone, online chat or other communication methods.

[0027] ② Basic question identification: The system preliminarily analyzes the user's request and determines the user's basic information and service needs through simple keyword matching or preset question and answer pairs.

[0028] ③Static allocation strategy: user requests are allocated to corresponding customer service agents based on the user's basic information and preset rules (such as business type, time, etc.). This strategy does not consider the specific capabilities and current status of the customer service.

[0029] ④ Dynamic adjustment: In some systems, if it is found that the initially assigned customer service representative cannot effectively handle the request, simple dynamic adjustments may be made, such as reallocation based on the customer service representative’s current idle status.

[0030] ⑤Service processing: After receiving the assigned user request, the customer service agent begins to handle the user's question, providing answers or performing further service operations.

[0031] ⑥Service completion and feedback: After the service is completed, the system may invite users to evaluate the service to collect feedback on the service quality.

[0032] In summary, the customer service seat matching process in the existing customer service system includes static allocation, dynamic allocation, service processing, service feedback and other steps, aiming to improve service efficiency, reduce the burden of manual customer service and ensure that users get a satisfactory service experience.

[0033] However, existing technologies often only focus on a single dimension of data (such as waiting time) and ignore other important factors, such as the agent's skills, experience, and historical performance, resulting in customer requests not always being assigned to the most suitable customer service agent. This one-sided and lack of personalized evaluation mechanism limits the improvement of allocation efficiency and quality.

[0034] In view of the above problems, an embodiment of the present invention proposes a customer service agent matching method, in which the powerful data processing and learning capabilities of a trained large-scale language model are utilized to accurately capture the user's key features; the current indicator values ​​of the customer service agent's key indicators are updated in a timely manner, and the multi-dimensional data of the user's key features and the agent's key indicators are comprehensively evaluated to achieve optimal matching, thereby promoting the intelligent and personalized development of customer service allocation.

[0035] The embodiments of the present invention can be applied to scenarios where customer service agent matching is required. The execution subject of the method can be an electronic device such as a terminal device, a computer, a server, a server cluster, or a specially designed customer service agent matching device, or a customer service agent matching device set in the electronic device, and the customer service agent matching device can be implemented by software, hardware, or a combination of the two.

[0036] Figure 1 It is one of the flow charts of the customer service agent matching method provided by the present invention, such as Figure 1 As shown, the method comprises the following steps: Step 110, obtaining the user's current conversation data; Step 120, analyzing the current conversation data based on the trained large language model to obtain the user's key features; Step 130, based on the key features of the user and the current indicator values ​​of the key indicators of each customer service agent, determine the customer service agent that matches the user, where the current indicator value is updated based on the historical service records of the customer service agent.

[0037] Specifically, the current conversation data may be human-computer conversation data, such as conversation data generated by the interaction between a user and an AI intelligent customer service. For example, a user interacts with a customer service system through a terminal device (mobile phone, computer, etc.), and conversation data may be generated through chat applications, voice calls, video calls, etc. The data collection layer captures this data in real time as the current conversation data. The current conversation data includes but is not limited to text, voice, images, etc. The captured conversation data may also be further preprocessed, such as removing noise, standardizing the format, and text segmentation, for subsequent analysis.

[0038] Here, Large Language Model (LLM) is referred to as large model, which refers to a natural language processing (NLP) model with a large parameter scale. The number of model parameters and / or the complexity of the model structure exceeds the preset threshold. The model processes a large amount of text data during training and has the ability to understand and generate natural language. For example, large language models can include Spark large models, etc.

[0039] Furthermore, the large-scale language model here can be a large-scale language model in a general field, or a large-scale language model in the customer service field obtained by fine-tuning with customer service center field data. For example, the large-scale language model can be obtained by fine-tuning a pre-trained large-scale language model using a large amount of sample conversation data and key feature description information. The embodiments of the present invention do not specifically limit this.

[0040] On this basis, the current conversation data can be analyzed through a large language model to obtain the user's key features. For example, the current conversation data can be included in the input text of the large language model, and the input text can be input into the large language model. The large language model analyzes the conversation content and extracts key features such as the user's emotional tendency (such as positive, negative, neutral), question type (such as technical consultation, complaint, suggestion, etc.), and key needs (such as solving problems, seeking advice, etc.). Here, the dimensions of the user's key features can be multidimensional, which can comprehensively reflect the key information expressed by the user in the current conversation data.

[0041] Here, the key indicators of the customer service agents are also multidimensional, reflecting the personal characteristics and advantages of the customer service agents from multiple dimensions. The current indicator value is the value of the key indicator of the agent at the current moment, and the current indicator value is updated with the historical service records of the customer service agent. The historical service records of each customer service agent are obtained from the database, and the current key indicator values ​​of each agent, such as average response time, problem resolution rate, customer satisfaction index, etc., are calculated using big data analysis technology, such as statistical or machine learning algorithms.

[0042] After obtaining the key features and current index values ​​of the user, a matching algorithm (such as multi-objective optimization, distance measurement, etc.) can be used to determine the customer service agent that matches the user. In the matching process, priorities can be set according to business needs, such as giving priority to agents with high satisfaction and low complaint rates; or, based on the type of user's demands, giving priority to customer service agents whose skill levels match the user's demands; or, based on the user's current emotional state, giving priority to customer service agents who are good at soothing the user's emotions and have strong empathy; or, by comprehensively considering the matching degree between the user's key features and the current index values ​​of each customer service agent, calculating a comprehensive matching score, and giving priority to matching the customer service agent with the highest score, which is not specifically limited in the embodiments of the present invention.

[0043] For example, suppose there is a customer service representative named Xiao Zhang, who has handled a large number of product return and exchange issues in his past service records and has received high praise from customers. Through large-scale model analysis, it can be found that Xiao Zhang not only performs well in handling return and exchange issues, but also has a gentle and patient communication style, which can effectively appease dissatisfied customers. Therefore, when the system receives a request for product return, it can give priority to assigning such requests to Xiao Zhang.

[0044] The method provided by the embodiment of the present invention utilizes the powerful data processing and learning capabilities of a trained large-scale language model to accurately capture the user's key features. Compared with simple keyword matching to determine the user's service needs, it can accurately extract the user's key features in multiple dimensions. By timely updating the current indicator values ​​of the customer service agent's key indicators, the multi-dimensional data of the user's key features and the agent's key indicators are comprehensively evaluated to achieve optimal matching, thereby promoting the intelligent and personalized development of customer service allocation.

[0045] The method provided by the embodiment of the present invention firstly understands the actual ability and current status of each customer service more accurately through comprehensive evaluation of multi-dimensional data; secondly, uses a large model to conduct in-depth analysis of user demands and emotions, making the matching process more accurate and personalized; finally, adopts an intelligent matching algorithm to dynamically adjust the seat allocation strategy to adapt to the ever-changing service needs and customer emotions, which significantly improves the quality and efficiency of the service. Compared with simple or static allocation strategies, it can better meet customer needs and enhance user experience.

[0046] In some embodiments, the large language model is obtained through unsupervised pre-training and instruction fine-tuning based on sample conversation data. Multiple rounds of conversation texts in the field are obtained from the Internet, and user privacy information (name, address, phone number, etc.) is removed through keywords and other means. The open source data is cleaned for unsupervised pre-training.

[0047] The entire model structure can adopt a decoder-only architecture, with the tarnsformer structure as the main network. The network module is mainly composed of Multi-head Attention and a fully connected network.

[0048] Training guidelines: The large language model can predict the next word based on the input of the first k words, and the quality of the trained language model can be measured by indicators such as perplexity. Then the instructions are fine-tuned, and a large amount of human-computer dialogue and customer service agent dialogue data are used to regularize the speech, which is convenient for user demand analysis and sentiment analysis, so that the large model can fully learn the user's feature capture, and at the same time, the information describing the key features is input into the large model, so as to provide data that is more in line with the features.

[0049] In some embodiments, the key characteristics of users include at least one of user demands, emotional state, and urgency; the key indicators of agents include at least one of skill level, stress resistance, emotional response ability, and service satisfaction rate.

[0050] Specifically, user demands refer to the main demands expressed by the current dialogue user. Through natural language processing technology, semantic analysis is performed on the user's consultation content to identify the user's specific needs and problems, and divide user requests into different categories, such as product consultation, technical support, account issues, complaint feedback, service improvement suggestions, etc.

[0051] Emotional analysis is an integral part of understanding user needs. By analyzing the user's language, tone intensity, etc., the user's emotional state can be inferred. The current emotional state is divided into five levels: calm, slightly dissatisfied, obviously dissatisfied, angry, and desperate.

[0052] Determine the urgency of the issue based on the customer's words and tone, distinguish between urgent and non-urgent issues, and ensure that urgent issues are responded to quickly.

[0053] Considering that in traditional systems, the skills and experience of customer service agents are often not fully utilized. Due to the lack of an effective evaluation and matching mechanism, highly skilled and experienced customer service agents may not be able to fully realize their potential, while novice or underperforming agents may be overused. In this embodiment, the key indicators of agents may include skill levels. Skill level refers to the expertise and problem-solving ability of customer service agents in specific areas (such as technical support, account management, etc.). This includes the degree of understanding of products or services, industry knowledge, and experience in handling specific types of problems. Skill level directly affects whether customer service agents can solve customer problems accurately and quickly, and is one of the important indicators for measuring customer service quality. By evaluating the skills and expertise of customer service agents, it can be ensured that customer requests are assigned to the most appropriate agent, thereby improving the efficiency and quality of problem solving.

[0054] Customer emotions are one of the important factors affecting service quality. However, related technologies have obvious deficiencies in understanding and responding to customer emotions, which may lead to misunderstandings or conflicts in the service process, further affecting the customer experience. In this embodiment, the key indicators of the agent may include emotional responsiveness. Emotional responsiveness refers to the ability to understand and respond to customer emotions. This includes empathy, communication skills, and conflict resolution skills. Emotional responsiveness is essential for building good customer relationships. Effective emotion management can not only improve customer satisfaction, but also help ease tense service scenarios and reduce negative feedback. By analyzing the emotional responsiveness of customer service agents, we can better match customer requests that require a high degree of empathy and patience, thereby improving the overall service quality.

[0055] The real-time status of customer service agents (such as workload, stress level, etc.) has a direct impact on service quality. In this embodiment, the key indicators of agents may include stress tolerance. Stress tolerance refers to the customer service agent's coping strategies and ability to adjust their mentality when facing complaints, emergencies or other high-pressure situations. In customer service, challenging and stressful situations are often encountered. Having a good ability to cope with stress can help customer service agents stay calm, solve problems effectively, and avoid emotional reactions. Assessing the customer service agent's ability to cope with stress helps ensure that complex or urgent customer requests can be handled quickly and properly, ensuring the continuity and stability of the service.

[0056] Based on any of the above embodiments, Figure 2 is a flow chart of step 120 in the customer service agent matching method provided by the present invention, such as Figure 2 As shown, based on the trained large language model, the current conversation data is analyzed to obtain the user's key features, that is, step 120 specifically includes: Step 121, generating analysis input text based on the analysis prompt information and the current conversation data, wherein the analysis prompt information includes instructions for determining key features of the user; Step 122, input the analyzed input text into the large language model to obtain the user key features output by the large language model.

[0057] Specifically, in order to extract key features of the user through the large language model, the input for the large language model can be constructed based on the current conversation data and the prompt of the large language model. In the embodiment of the present invention, the prompt for generating key features of the user for the large language model is recorded as analysis prompt information; the input for the large language model is recorded as analysis input text.

[0058] Here, the analysis prompt information is used to specify the generation requirements of user key features at the input stage of the large language model, that is, the analysis prompt information is used to instruct the large language model to determine the user key features. Therefore, the analysis prompt information may include instructions for instructing the large language model to generate user key features. The instructions here are in the form of natural language text, for example, "Please extract the key features of the user in the following conversation content" or "Assuming that you are a language analysis expert, please analyze the user's main demands based on the following conversation content and extract the user's key features."

[0059] After obtaining the current dialogue data, the analysis input text may be generated based on the analysis prompt information and the current dialogue data. For example, the analysis prompt information and the current dialogue data may be concatenated to obtain the analysis input text; or the analysis prompt information and the current dialogue data may be filled into a pre-set input text template to obtain the analysis input text, which is not specifically limited in the embodiment of the present invention.

[0060] After obtaining the analysis input text, the analysis input text can be input into a large language model, which will understand the conversation content, identify the user's main demands, analyze the user's emotions, and determine the urgency of the problem, thereby extracting the user's key features.

[0061] The method provided by the embodiment of the present invention generates analysis input text as input of a large language model by analyzing prompt information and current conversation data, thereby realizing automatic extraction of key user features by invoking the excellent semantic understanding and analysis capabilities of the large language model.

[0062] Based on any of the above embodiments, the analysis prompt information also includes question information and expected output targeting key features of the user.

[0063] Specifically, the analysis prompt information may include, in addition to instructions for determining user key features to instruct the large language model to generate user key features, question information and expected output for the user key features to facilitate guiding the large language model to generate user key features that meet user expectations.

[0064] Question information refers to questions set for the type of user key features expected to be output, and expected output refers to the expected answer to the question information. For example, if the user key features expected to be output include the type of demand, the question information "What are the main categories of user demands?" can be set, and the corresponding expected output includes "Current demands are divided into five categories: complaint feedback, product consultation, technical support, account issues, and service improvement"; if the user key features expected to be output include the user's emotional state, the question information "What is the user's current emotional level?" can be set, and the corresponding expected output includes "Current emotions are divided into: calm, slightly dissatisfied, obviously dissatisfied, angry, and desperate."

[0065] The large language model understands and analyzes the analysis prompt information and outputs the user's key features that meet the requirements.

[0066] In some embodiments, analyzing the input text may be presented as follows: “You are a language analysis expert, and you know that the user AI customer service conversation is as follows: [User]: Hello, how are you? Why is my phone bill so much higher recently?

AI Customer Service

[0067] [User]: I just want to know why my phone bill is so expensive! I don’t want to talk to a robot, I want to be transferred to a human operator [AI Customer Service]: We are very sorry for the inconvenience. In order to better help you solve the problem, we recommend that you provide some basic information first so that we can find the problem more quickly. If you insist on being transferred to human customer service, I will record it for you and arrange it as soon as possible.

[0068] [User]: I don’t care, I want to be transferred to manual service now! My phone bill is so much higher, I’m going crazy!

AI Customer Service

[0069] Based on the above conversation, please analyze and answer the following questions: 1. What are the main categories of user demands? Currently, demands are divided into five categories: complaint feedback, product consultation, technical support, account issues, and service improvement; 2. What is the user's current emotion level? Current emotions are divided into: calm, slightly dissatisfied, obviously dissatisfied, angry, and desperate; 3. What is the urgency level of the current user's needs? The current urgency level of needs is divided into non-urgent, urgent, and very urgent;".

[0070] The large language model analyzes the key features of users and outputs them as follows: Complaint feedback.

[0071] anger.

[0072] Very urgent.

[0073] That is, based on the current conversation content, it can be known that the user's demand type is complaint feedback, the user's current emotional state is anger, and the current urgency of the user's needs is very urgent.

[0074] Based on any of the above embodiments, Figure 3 is one of the flow charts of step 130 in the customer service agent matching method provided by the present invention, such as Figure 3 As shown, based on the key features of the user and the current indicator values ​​of the key indicators of each customer service agent, a customer service agent matching the user is determined, that is, step 130 specifically includes: Step 131, determining a matching score between the customer and each customer service agent based on the similarity between the user's key features and the agent's key indicators, as well as the current indicator value; Step 132: Determine a customer service agent that matches the user based on the matching score.

[0075] Specifically, the matching between the user and the customer service agent may be determined by comprehensively considering the key features of the user and the current indicator values ​​of the key indicators of each customer service agent.

[0076] First, calculate the similarity between the key features of the user and the key indicators of the agent. Commonly used similarity measurement methods include cosine similarity, Pearson correlation coefficient, Spearman correlation coefficient, etc. Depending on the actual situation, one or more of these methods can be selected for combination.

[0077] For each user key feature, calculate the similarity between it and all agent key indicators. The user key feature and the agent key indicator can be regarded as vectors, and then the cosine value of the angle between these vectors or other similarity indicators are calculated. It can be understood that the higher the similarity, the greater the correlation between the user key feature and the agent key indicator; conversely, the lower the similarity, the smaller the correlation between the user key feature and the agent key indicator. Further, the correlation between the user key feature and the agent key indicator can be established based on the calculated similarity value. The user key feature with a higher similarity value can be associated with the agent key indicator to form an associated pair.

[0078] Then, based on the current index value and similarity, the matching score between the customer and each customer service agent is calculated. The similarity and the current index value are weighted and summed to obtain the matching score. Based on the matching score, the customer service agent with the highest score is selected as the agent that matches the user.

[0079] If there are multiple agents with very close matching scores, you can consider other factors (such as the agent's experience, skills, etc.) for further screening.

[0080] The method provided by the embodiment of the present invention determines the matching score between the customer and each customer service agent based on the similarity between the user's key features and the agent's key indicators, as well as the current indicator value, and determines the customer service agent that matches the user based on the matching score, so as to optimally assign the user request to the most suitable customer service agent.

[0081] Based on any of the above embodiments, Figure 4 This is the second flow chart of step 130 in the customer service agent matching method provided by the present invention. Figure 4 As shown, based on the similarity between the key features of the user and the key indicators of the agent, as well as the current indicator value, the matching score between the customer and each customer service agent is determined, that is, step 131 includes: Step 131-1, based on the weight value, similarity and current index value of each user's key feature, determine the initial matching score between the customer and each customer service agent, where the weight value represents the importance of the user's key feature for agent matching; Step 131 - 2 , determining a matching score based on the current indicator value of the service satisfaction rate indicator in the agent key indicators and the initial matching score.

[0082] Specifically, the weight value of each user key feature can be pre-set to reflect the importance of each user key feature for seat matching. The higher the importance, the greater the weight value; conversely, the lower the importance, the smaller the weight value. The weight values ​​of user key features can be displayed as shown in Table 1.

[0083] Table 1

[0084] For each customer service agent, the weight value of the user's key features, the similarity between the key indicators of the agent, and the current indicator value are multiplied and weighted to obtain the initial matching score. Then, based on the current indicator value of the service satisfaction rate indicator in the key indicators of the agent and the initial matching score, the matching score is obtained.

[0085] In some embodiments, assume that U represents a set of user demand types, C represents a set of customer service capabilities, and for each user demand And each customer service , define a matching function It can be expressed as: in is the weight value of the Kth user's key feature, To measure user demands and customer service The similarity on the Kth feature, T is the current index value.

[0086] Ultimately, the system will choose The best customer service Assign to user .

[0087] For example, the weight values ​​of key user features are shown in Table 2 below: Table 2

[0088] The current coordinate values ​​of the customer service agent are shown in Table 3 below: Table 3

[0089] Use the above formula to calculate the matching score between the current customer and each agent: The matching score between the customer and customer service X is: (1*90 + 0.8*65 + 0.8*70)*8.85 = 198*8.85 = 175.23 The matching score between the customer and customer service Y is: (1*80 + 0.8*75 + 0.8*90)*0.912 = 212*0.912=193.344 According to the calculation results, we can see that customer A and customer service Y have the highest matching score, so the customer can be assigned to customer service Y.

[0090] The method provided by the embodiment of the present invention assigns a weight value to each user key feature according to the importance of the user key feature to agent matching, further optimizes the matching algorithm, and improves the accuracy of customer service agent allocation.

[0091] Based on any of the above embodiments, after determining the customer service agent matching the user, the method further includes: Receive user reviews of customer service agents; Based on the service evaluation, the current indicator values ​​of the key indicators of the customer service agents are updated.

[0092] Specifically, considering the real-time status changes of customer service seats, this embodiment also includes a dynamic scoring mechanism for service satisfaction based on user service evaluations to ensure that the best allocation decision can be made at any given moment.

[0093] After the customer service agent finishes the call with the user, the user can evaluate the overall quality of the current customer service, and can give a score or classify it as satisfied or dissatisfied. After receiving the user's service evaluation of the customer service agent, the service satisfaction rate of the customer service agent can be updated in real time = (number of satisfied ratings / total number of ratings). The service satisfaction rate indicator will be used as an important indicator for customer service scoring and will be used in the next round of user and customer service agent matching calculations.

[0094] Based on any of the above embodiments, Figure 5 FIG. 2 is a flow chart of the customer service agent matching method provided by the present invention. Figure 5 As shown, the method includes: S1, collects information such as the skills, experience, and historical performance of customer service agents, collects historical interaction records of users, and determines the current indicator values ​​of key indicators of each customer service agent. Select training samples, clean them, and annotate them, and fine-tune the large language model.

[0095] S2, obtaining the user's current conversation data; based on the trained large-scale language model, analyzing the current conversation data to obtain the user's key features, the user's key features including at least one of the user's demands, emotional state, and urgency.

[0096] S3, based on the similarity between the key features of the user and the key indicators of the agent, and the current indicator value, determine the matching score between the customer and each customer service agent; and determine the customer service agent that matches the user based on the matching score. The key indicator of the agent includes at least one of skill level, stress resistance, emotional response ability, and service satisfaction rate.

[0097] S4, receiving the user's service evaluation of the customer service agent; based on the service evaluation, updating the current indicator value of the key indicator of the customer service agent.

[0098] The embodiments of the present invention utilize the powerful data processing and learning capabilities of the big model to accurately capture user demands and emotional changes, and combine the actual capabilities and status of the customer service agents to achieve precise matching. Through in-depth analysis of user demands and emotions, more personalized service solutions can be provided, thereby significantly improving user satisfaction. The intelligent matching algorithm reduces unnecessary transfers and waiting time, improves processing speed and overall service efficiency. The matching method of the present invention can be adjusted according to different scenarios and needs, and has strong flexibility and scalability. The big model technology is used to deeply analyze user needs and the real-time status of customer service to achieve highly accurate agent matching. Customized services are provided based on user emotions and demands to enhance user satisfaction. Intelligent algorithms reduce manual intervention, speed up response speed, and improve service process efficiency.

[0099] The embodiment of the present invention comprehensively considers the skill level, stress tolerance, emotional response ability and other dimensions of customer service agents to ensure that the ability of each agent can be fully evaluated. The analysis is combined with the user's demands and emotional characteristics to more accurately understand user needs and provide more considerate services. Using advanced matching algorithms, based on the results of the comprehensive evaluation and user characteristics, user requests are optimally assigned to the most suitable customer service agent.

[0100] The customer service seat matching device provided by the present invention is described below. The customer service seat matching device described below and the customer service seat matching method described above can be referenced to each other.

[0101] Based on any of the above embodiments, Figure 6 is a structural schematic diagram of the customer service seat matching device provided by the present invention, such as Figure 6 As shown, the device comprises: The data acquisition unit 610 is used to acquire the current conversation data of the user; A data analysis unit 620, configured to analyze the current conversation data based on a trained large language model to obtain key user features of the user; The agent matching unit 630 is used to determine a customer service agent that matches the user based on the user key features and the current indicator value of the agent key indicator of each customer service agent, wherein the current indicator value is updated based on the historical service record of the customer service agent.

[0102] Based on any of the above embodiments, the data analysis unit is specifically used for: generating an analysis input text based on the analysis prompt information and the current conversation data, wherein the analysis prompt information includes instructions for determining key features of the user; The analyzed input text is input into the large language model to obtain the user key features output by the large language model.

[0103] Based on any of the above embodiments, the analysis prompt information also includes question information and expected output for the key features of the user.

[0104] Based on any of the above embodiments, the seat matching unit is specifically used for: Based on the similarity between the key features of the user and the key indicators of the agent, and the current indicator value, determining a matching score between the customer and each of the customer service agents; A customer service agent matching the user is determined based on the matching score.

[0105] Based on any of the above embodiments, the seat matching unit is specifically used for: Based on the weight value of each user's key feature, the similarity and the current index value, determine the initial matching score between the customer and each customer service agent, wherein the weight value represents the importance of the user's key feature for agent matching; The matching score is determined based on the current indicator value of the service satisfaction rate indicator in the agent key indicators and the initial matching score.

[0106] Based on any of the above embodiments, it also includes an indicator updating unit, which is used to: Receiving a service evaluation from the user regarding the customer service agent; Based on the service evaluation, the current indicator value of the key indicator of the customer service agent is updated.

[0107] Figure 7 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 7 As shown, the electronic device may include: a processor 710, a communications interface 720, a memory 730 and a communication bus 740, wherein the processor 710, the communications interface 720 and the memory 730 communicate with each other through the communication bus 740. The processor 710 may call the logic instructions in the memory 730 to execute the customer service agent matching method, which includes: obtaining the current conversation data of the user; analyzing the current conversation data based on the trained large language model to obtain the user key features of the user; determining the customer service agent matching the user based on the user key features and the current indicator value of the agent key indicator of each customer service agent, wherein the current indicator value is updated based on the historical service record of the customer service agent.

[0108] In addition, the logic instructions in the above-mentioned memory 730 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on this understanding, the technical solution of the present invention can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0109] On the other hand, the present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the customer service agent matching method provided by the above methods, which method includes: obtaining the user's current conversation data; based on a trained large language model, analyzing the current conversation data to obtain the user's key features; based on the user's key features and the current indicator values ​​of the key indicators of each customer service agent, determining the customer service agent that matches the user, the current indicator value is updated based on the historical service record of the customer service agent.

[0110] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the customer service agent matching method provided by the above-mentioned methods, the method comprising: obtaining the user's current conversation data; analyzing the current conversation data based on a trained large language model to obtain the user's key features; determining the customer service agent matching the user based on the user's key features and the current indicator values ​​of the key indicators of each customer service agent, wherein the current indicator values ​​are updated based on the historical service records of the customer service agent.

[0111] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0112] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0113] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A customer service agent matching method, characterized in that: include: Get the user's current conversation data; Analyze the current conversation data based on the trained large language model to obtain key user features of the user; Based on the key features of the user and the current indicator values ​​of the key indicators of each customer service agent, a customer service agent matching the user is determined, wherein the current indicator values ​​are updated based on the historical service records of the customer service agent.

2. The customer service agent matching method according to claim 1, characterized in that: The current conversation data is analyzed based on the trained large language model to obtain the key user features of the user, including: generating an analysis input text based on the analysis prompt information and the current conversation data, wherein the analysis prompt information includes instructions for determining key features of the user; The analyzed input text is input into the large language model to obtain the user key features output by the large language model.

3. The customer service agent matching method according to claim 2, characterized in that: The analysis prompt information also includes question information and expected output targeting the key features of the user.

4. The customer service agent matching method according to claim 1, characterized in that: The determining of a customer service agent matching the user based on the key features of the user and the current indicator values ​​of the key indicators of each customer service agent includes: Based on the similarity between the key features of the user and the key indicators of the agent, and the current indicator value, determining a matching score between the customer and each of the customer service agents; A customer service agent matching the user is determined based on the matching score.

5. The customer service agent matching method according to claim 4, characterized in that: The determining of the matching score between the customer and each of the customer service agents based on the similarity between the key features of the user and the key indicators of the agent, and the current indicator value, includes: Based on the weight value of each user's key feature, the similarity and the current index value, determine the initial matching score between the customer and each customer service agent, wherein the weight value represents the importance of the user's key feature for agent matching; The matching score is determined based on the current indicator value of the service satisfaction rate indicator in the agent key indicators and the initial matching score.

6. The customer service agent matching method according to any one of claims 1 to 5, characterized in that: After determining the customer service agent matching the user, the method further includes: Receiving a service evaluation from the user regarding the customer service agent; Based on the service evaluation, the current indicator value of the key indicator of the customer service agent is updated.

7. The customer service agent matching method according to any one of claims 1 to 5, characterized in that: The key user characteristics include at least one of user demands, emotional state, and urgency; the key agent indicators include at least one of skill level, stress resistance, emotional response ability, and service satisfaction rate.

8. A customer service seat matching device, characterized in that: include: A data acquisition unit, used to acquire the user's current conversation data; A data analysis unit, configured to analyze the current conversation data based on a trained large language model to obtain key user features of the user; The agent matching unit is used to determine a customer service agent that matches the user based on the user key features and the current indicator value of the agent key indicator of each customer service agent, wherein the current indicator value is updated based on the historical service record of the customer service agent.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the customer service agent matching method as described in any one of claims 1 to 7 is implemented.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the customer service agent matching method as described in any one of claims 1 to 7 is implemented.

Images