An integrated customer service method and system based on multi-agent

Abstract

The present application relates to an integrated customer service method and system based on multiple agents, wherein the method includes: processing consulting questions raised by active incoming users through a dialogue management agent, obtaining multi-dimensional data information and generating dialogue responses to the consulting questions; based on the feedback of the active incoming users after the dialogue response, obtaining the consulting intention of the active incoming users through analysis by an intention recognition agent; based on the consulting intention and recent activity records of the active incoming users, determining the outgoing call operation to be performed through an outgoing decision agent; and then executing the outgoing call operation through an outgoing execution agent. Through the present application, integrated customer service of multiple agents is realized, and the collaborative division of labor of multiple different agents is achieved, each focusing on its specific business process, thereby providing better services to customers. It is also possible to proactively contact customers by analyzing customer data, track customer issues or conduct proactive marketing, and effectively improve customer satisfaction.

Description

Technical Field

[0001] The present application relates to the field of intelligent customer service technology, and in particular to an integrated customer service method and system based on multiple agents. Background Art

[0002] With the rapid development of information technology, the need for communication between businesses and customers is growing, and customer service systems are playing an increasingly important role in business operations. Early rule-based customer service systems lacked intelligence, handled a single problem, and struggled to meet customer needs.

[0003] However, existing agent-based customer service systems also have limitations in terms of flexibility, contextual understanding, and emotional communication capabilities. Knowledge base updates are often delayed, making it difficult to generate high-quality answers based on specific customer questions. Traditional inbound customer service systems rely primarily on proactive customer contact and lack the ability to proactively serve customers. It is difficult to proactively track issues and engage in marketing in limited customer interactions.

[0004] Currently, no effective solution has been proposed for the problem of how to improve the service capabilities of intelligent customer service in related technologies. Summary of the Invention

[0005] The embodiments of the present application provide an integrated customer service method and system based on multiple agents to at least solve the problem of how to improve the service capabilities of intelligent customer service in related technologies.

[0006] In a first aspect, an embodiment of the present application provides an integrated customer service method based on multiple agents, the method comprising:

[0007] Obtain consultation questions raised by active inbound users;

[0008] The multidimensional data information of the active incoming user is obtained through the dialogue management agent, and a dialogue response to the consultation question is generated based on the multidimensional data information;

[0009] Based on the feedback of the active incoming user after the dialogue response, the consultation intention of the active incoming user is obtained through analysis by the intention recognition agent;

[0010] Based on the consultation intention and recent activity records of the active incoming user, determining the outgoing call operation to be performed through the outgoing call decision agent;

[0011] The outgoing operation is executed by the outgoing execution agent to conduct an outgoing dialogue with the active incoming user.

[0012] In some embodiments, the method includes:

[0013] Analyzing historical conversation records between the conversation management agent and the active incoming user through a knowledge management agent;

[0014] If it is determined that the answer to the question given by the dialogue management agent does not match the consultation question raised by the active incoming user, the knowledge management agent will mark the corresponding data entry in the knowledge base for manual modification, or make suggestions for data supplementation to the knowledge base.

[0015] In some embodiments, the method includes:

[0016] The dialogue management agent, the intention recognition agent, the outbound decision agent, the outbound execution agent and the knowledge management agent all iteratively optimize their respective execution strategies π under the premise of maximizing their respective value functions through reinforcement learning algorithms. i ( a | s ), where i represents the i-th agent, a is the state of the environment perceived by the agent, s Actions performed by the agent.

[0017] In some embodiments, the multi-dimensional data information of the active incoming user obtained through the dialogue management agent processing includes:

[0018] Converting the consultation questions of the active incoming user into streaming data;

[0019] For streaming data that can be processed in real time by a voice activity detection model in a dialogue management agent, the voice activity detection model is input for consumption to detect the start and end time points of valid speech in the consultation question;

[0020] For streaming data that cannot be processed in real time by the voice activity detection model, it is not input into the voice activity detection model, but is consumed quickly through the data packets in the buffer;

[0021] Then, the automatic speech recognition model in the dialogue management agent is used to convert the effective speech within the start and end time points into natural language text, and the multi-dimensional data information of the active incoming user is obtained based on the natural language text.

[0022] In some embodiments, obtaining multidimensional data information of the active incoming user through processing by a dialogue management agent, and generating a dialogue response to the consultation question based on the multidimensional data information includes:

[0023] Processing the consultation questions of the active incoming user through the dialogue management agent to obtain user data information, emotion data information and answer data information of the active incoming user;

[0024] Based on the user data information, the emotion data information and the answer data information, a dialogue response to the consultation question is generated by the dialogue management agent.

[0025] In some embodiments, generating a dialogue response to the consultation question by the dialogue management agent based on the user data information, the emotion data information, and the answer data information includes:

[0026] Based on the emotional data information, the dialogue management agent determines whether manual customer service is required, and if so, transfers the call to manual customer service for processing;

[0027] If not, a dialogue response to the consultation question is generated by the dialogue management agent based on the user data information and the answer data information.

[0028] In some embodiments, based on the feedback of the active incoming user after the dialogue response, the consultation intention of the active incoming user obtained through analysis by the intention recognition agent includes:

[0029] Based on the feedback of the active incoming user after the dialogue response, the intention recognition agent determines whether the consultation problem of the active incoming user has been solved;

[0030] If not, continue to process the consultation questions of the active incoming user through the dialogue management agent;

[0031] If so, the current conversation with the active incoming user is ended, and based on the content of the current conversation, the consultation intention of the active incoming user is obtained through intention recognition agent analysis.

[0032] In some embodiments, based on the consultation intention and recent activity records of the actively calling user, determining, by the outbound decision agent, the outbound operation to be performed includes:

[0033] Based on the consultation intention of the active incoming user, the outbound decision agent determines the task type of the outbound operation to be performed, wherein the task type includes coupon issuance, renewal reminder and product promotion;

[0034] Based on the recent activity records of the active incoming user, the execution of the outgoing operation is triggered by the outgoing decision agent.

[0035] In some embodiments, executing the outbound operation by an outbound execution agent to conduct an outbound dialogue with the active incoming user includes:

[0036] In the case where the outbound decision agent triggers an outbound operation, based on the user contact information provided by the outbound decision agent and the task type of the outbound operation, the outbound operation is executed by the outbound execution agent to conduct an outbound dialogue with the active incoming user.

[0037] In a second aspect, an embodiment of the present application provides an integrated customer service system based on multiple agents, the system being used for the method described in the first aspect above, the system comprising a dialogue management module, a knowledge management module, an intent recognition module, an outbound call decision module, and an outbound call execution module;

[0038] The dialogue management module is used to obtain consulting questions raised by active inbound users, obtain multi-dimensional data information of the active inbound users through processing by the dialogue management agent, and generate dialogue responses to the consulting questions based on the multi-dimensional data information;

[0039] The knowledge management module is configured to analyze historical conversation records between the dialogue management agent and the active incoming user through a knowledge management agent; if it is determined that the answer to the question given by the dialogue management agent does not match the consultation question raised by the active incoming user, the knowledge management agent is configured to mark the corresponding data entry in the knowledge base for manual modification or to make suggestions for data supplementation to the knowledge base;

[0040] The intention recognition module is used to obtain the consultation intention of the active incoming user through the intention recognition agent analysis based on the feedback of the active incoming user after the dialogue response;

[0041] The outbound call decision module is used to determine the outbound call operation to be performed through the outbound call decision agent according to the consultation intention and recent activity record of the active incoming user;

[0042] The outgoing call execution module is used to execute the outgoing call operation through the outgoing call execution agent to conduct an outgoing call dialogue with the active incoming call user.

[0043] Compared with related technologies, the embodiments of the present application provide an integrated customer service method and system based on multiple agents, wherein the method obtains consulting questions raised by active incoming users; obtains multi-dimensional data information of active incoming users through dialogue management agents, and generates dialogue responses to consulting questions based on the multi-dimensional data information; obtains the consulting intention of the active incoming users through intention recognition agents based on the feedback of the active incoming users after the dialogue response; determines the outbound operation to be performed through the outbound decision-making agent based on the consulting intention and recent activity records of the active incoming users; and executes the outbound operation through the outbound execution agent to conduct an outbound dialogue with the active incoming users, thereby realizing integrated customer service of multiple agents, collaborative division of labor among multiple different agents, each focusing on its specific business processes, such as context understanding, sentiment analysis, knowledge updating, intention understanding, etc., thereby providing customers with more professional and personalized services, and can also actively contact customers by analyzing customer data, track customer problems or conduct proactive marketing, which can effectively improve customer satisfaction, help to establish more stable customer relationships, and solve the problem of how to improve the service capabilities of intelligent customer service. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] The drawings described herein are used to provide a further understanding of the present application and constitute 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 on the present application. In the drawings:

[0045] Figure 1 is a flowchart of the steps of the multi-agent-based integrated customer service method according to an embodiment of the present application;

[0046] Figure 2 is a structural diagram of an integrated customer service system based on multiple agents according to an embodiment of the present application;

[0047] Figure 3 Schematic diagram of the internal structure of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical solutions and advantages of this application more clearly understood, the present application is described and illustrated below in conjunction with the accompanying drawings and examples. It should be understood that the specific embodiments described herein are merely used to explain this application and are not intended to limit this application. Based on the embodiments provided in this application, all other embodiments obtained by those of ordinary skill in the art without making any creative efforts are within the scope of protection of this application.

[0049] Obviously, the drawings described below are merely examples or embodiments of the present application. Those skilled in the art can, without inventive effort, apply the present application to other similar scenarios based on these drawings. Furthermore, it is also understood that, although the effort involved in such a development process may be complex and lengthy, for those skilled in the art related to the content disclosed in this application, changes in design, manufacturing, or production based on the technical content disclosed in this application are merely conventional technical means and should not be construed as an insufficiency of the content disclosed in this application.

[0050] References to "embodiments" in this application mean that a particular feature, structure, or characteristic described in connection with the embodiment may be included in at least one embodiment of the application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it refer to independent or alternative embodiments that are mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described in this application may be combined with other embodiments unless there is a conflict.

[0051] Unless otherwise defined, technical or scientific terms used herein shall have the ordinary meaning as understood by persons of ordinary skill in the art to which this application belongs. The terms "a," "an," "an," "the," and similar expressions used herein do not denote quantitative limitations and may refer to either the singular or the plural. The terms "comprise," "include," "have," and any variations thereof, used herein, are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or apparatus comprising a series of steps or modules (units) is not limited to the listed steps or units but may also include steps or units not listed, or may include other steps or units inherent to the process, method, product, or apparatus. The terms "connected," "connected," "coupled," and similar expressions used herein are not limited to physical or mechanical connections but may include electrical connections, whether direct or indirect. As used herein, "plurality" means two or more. "And / or" describes an association between associated objects, indicating that three possible relationships exist. For example, "A and / or B" may mean: A exists alone; A and B exist simultaneously; or B exists alone. The character " / " generally indicates that the objects before and after are in an "or" relationship. The terms "first", "second", "third", etc. involved in this application are only used to distinguish similar objects and do not represent a specific order for the objects.

[0052] The embodiment of the present application provides an integrated customer service method based on multiple agents. Figure 1 This is a flowchart of the steps of the integrated customer service method based on multiple agents according to an embodiment of the present application. Figure 1As shown, the method includes:

[0053] Step S102, obtaining the consultation questions raised by the active incoming call user;

[0054] Step S104: The multi-dimensional data information of the active incoming user is obtained through the dialogue management agent, and a dialogue response to the consultation question is generated based on the multi-dimensional data information;

[0055] Step S104 specifically includes the following steps:

[0056] Step S1041, converting the consultation questions of the active incoming user into streaming data;

[0057] Step S1042: For streaming data that can be processed in real time by the voice activity detection model in the dialogue management agent, the streaming data is input into the voice activity detection model for consumption to detect the start and end time points of the valid speech in the consultation question;

[0058] Step S1043: For streaming data that cannot be processed in real time by the voice activity detection model, the streaming data is not input into the voice activity detection model, but is consumed quickly through the data packets in the buffer;

[0059] In step S1044, the effective speech within the start and end time points is converted into natural language text through the automatic speech recognition model in the dialogue management agent, and multi-dimensional data information of the active incoming user is obtained based on the natural language text, wherein the multi-dimensional data information includes user data information, emotion data information and answer data information.

[0060] It should be noted that Figure 2 is a structural diagram of an integrated customer service system based on multiple agents according to an embodiment of the present application. Figure 2 As shown in the figure, the dialogue management agent is responsible for handling inquiries and questions raised by customers over the phone. This agent integrates streaming VAD technology, which monitors customer voice input in real time. When a customer asks a question, it first uses ASR (automatic speech recognition) to recognize the audio as natural language. It then quickly finds the most appropriate answer by matching data in the internal knowledge base and database. Finally, it processes the answer using TTS (text-to-speech) technology and plays it back to the customer over the phone.

[0061] It's important to note that the primary function of the VAD (Voice Activity Detection) model is to detect the start and end time points of valid speech in the input audio and feed these detected valid audio segments into the recognition engine for recognition, thereby reducing recognition errors caused by invalid speech. This embodiment uses a VAD model based on the FSMN (Feed-forward Sequential Memory Network) architecture. This model effectively captures long-range sequential dependencies, which are particularly important for the long-term temporal characteristics of speech signals. The model also boasts fast inference speed, making it suitable for determining the end point of customer audio in streaming recognition scenarios.

[0062] When converting customer inquiries—represented by telephone signal packets (RTP, Real-time Transport Protocol)—into streaming VAD model input data, the RTP reception speed is significantly greater than the RTP processing speed. Unprocessed RTP packets are temporarily stored in the receive buffer allocated by the operating system for the socket. However, during streaming VAD model processing, RTP packets in the system buffer interfere with the process link. This is primarily due to the RTP buffer's impact on VAD recognition, resulting in redundant VAD model tail point recognition. Therefore, this embodiment designs a fast RTP buffer consumption solution, whereby packets outside the buffer are input into the VAD model for consumption. Buffered packets are quickly consumed without passing through the VAD model, enabling the link to accurately process customer inquiries.

[0063] Furthermore, it's important to note that the primary function of an ASR (Automatic Speech Recognition) model is to convert customer speech into natural human language for subsequent matching with the model's query answers. This patent utilizes an end-to-end speech recognition framework. Building on the model's existing knowledge, this model is fine-tuned using accumulated customer conversation audio from the business. This results in more accurate recognition, providing a crucial foundation for subsequent processing steps.

[0064] Step S1045: Based on the user data information, emotion data information and answer data information, a dialogue response to the consultation question is generated through the dialogue management agent.

[0065] Specifically, in step S1045, based on the emotional data information, the dialogue management agent determines whether manual customer service is needed. If so, the call is transferred to manual customer service for processing.

[0066] If not, a dialogue response to the consultation question is generated by the dialogue management agent based on the user data information and the answer data information.

[0067] It should be noted that the dialogue management agent also has emotion recognition capabilities and can analyze the emotional fluctuations in customer questions. If it detects that the customer is emotionally agitated or the problem is complex and requires human intervention, the agent will transfer the matter to manual customer service to ensure that the customer receives more personalized service.

[0068] Step S106, based on the feedback of the active incoming user after the dialogue response, the consultation intention of the active incoming user is obtained through the intention recognition agent analysis;

[0069] Specifically, step S106 determines whether the consultation problem of the active incoming user is solved based on the feedback of the active incoming user after the dialogue response through the intention recognition agent; if not, the consultation problem of the active incoming user is continued to be processed through the dialogue management agent; if so, the current conversation with the active incoming user is ended, and based on the content of the current conversation, the consultation intention of the active incoming user is obtained through the intention recognition agent analysis.

[0070] It should be noted that if Figure 2 As shown in the figure, the intent recognition agent plays a vital role both during and after the customer's incoming call conversation. During the conversation, by analyzing the customer's input, it determines whether the customer's inquiry has been fully answered, and thus decides whether to end the current call. After the call ends, the entire call content is identified and analyzed, and the customer's inquiry intention is classified and identified, such as whether to purchase a new package, renew an existing service, inquire about available discounts, and issues that require manual assistance. The intent classification identified by the incoming call intent recognition agent provides decision support for subsequent collaborative outbound call services, ensuring that customer needs are fully met.

[0071] Step S108, based on the consultation intention and recent activity records of the active incoming user, the outbound decision agent determines the outbound operation to be performed;

[0072] Specifically, step S108 determines the task type of the outgoing operation that needs to be performed based on the consultation intention of the active incoming user through the outgoing decision-making agent, wherein the task types include coupon issuance, renewal reminders and product promotion; based on the recent activity records of the active incoming user, the execution of the outgoing operation is triggered through the outgoing decision-making agent.

[0073] It should be noted that if Figure 2As shown, the Outbound Decision Agent is responsible for analyzing the customer's incoming call intent, their recent activity history, and comparing it with pre-set task types (such as issuing coupons, reminding them to renew their subscription, or promoting product packages) to determine whether an outbound call is necessary. For example, in an inbound call, if a customer inquires about product renewal, the Dialogue Management Agent will provide immediate renewal guidance. If the customer's renewal operation has not been completed within the pre-set business day, the Outbound Decision Agent will capture this behavioral data and determine whether an outbound call is necessary. Once it is determined that an outbound call is necessary, the Outbound Execution Agent will proactively contact the customer according to the Outbound Decision Agent's instructions to remind them to renew their subscription or provide further assistance to ensure service continuity and improve customer satisfaction.

[0074] Step S110, executing an outgoing operation through an outgoing execution agent to conduct an outgoing dialogue with the active incoming user.

[0075] Specifically, in step S110, when the outgoing decision agent triggers an outgoing operation, the outgoing operation is executed by the outgoing execution agent based on the user contact information provided by the outgoing decision agent and the task type of the outgoing operation to conduct an outgoing dialogue with the active incoming user.

[0076] It should be noted that if Figure 2 As shown in the figure, the Outbound Call Execution Agent is responsible for executing specific outbound call operations based on the customer contact information and pre-set tasks provided by the Outbound Call Decision Agent. Pre-set tasks include issuing coupons, reminding customers to renew their subscriptions, promoting packages, or assisting customers with their transactions. The Outbound Call Execution Agent leverages streaming VAD technology and conversational intent recognition to execute the corresponding actions assigned by the Outbound Call Decision Agent, ensuring that customers can successfully complete their intended transactions, thereby achieving efficient customer interaction.

[0077] Through the above steps in the embodiment of the present application, the integrated processing of incoming and outgoing calls is achieved to improve the customer experience, efficiency and quality of intelligent services. Not only can it timely resolve customer emotions and effectively solve customer problems, but it can also continue to pay attention to and respond to subsequent customer feedback, thereby enhancing customer satisfaction, reducing the risk of customer churn, and improving the market competitiveness of products. By integrating the results of different intelligent agent modules in the form of data streams, and using reinforcement learning technology to enhance the learning and comprehension capabilities of the intelligent agent, it can more accurately identify and meet customer needs and provide customized solutions. In the process of providing solutions, the intelligent agent will make outgoing calls in a timely manner according to the customer's response and interaction, ensuring the continuity and initiative of the service, thereby achieving effective communication and problem solving with customers. In addition, the method will also continuously optimize the question-and-answer knowledge base based on customer feedback, improve the problem-solving rate, and improve service quality through continuous learning and adaptation to adapt to the ever-changing customer needs and market environment.

[0078] In some embodiments, such as Figure 2 As shown, the multi-agent-based integrated customer service method also includes a knowledge management agent;

[0079] The knowledge management agent analyzes the historical conversation records between the dialogue management agent and the active incoming user; if it is determined that the answer to the question given by the dialogue management agent does not match the consulting question raised by the active incoming user, the knowledge management agent marks the corresponding data entry in the knowledge base for manual modification, or makes suggestions for data supplementation to the knowledge base.

[0080] It's important to note that the knowledge management agent undertakes the important task of meticulously evaluating the quality and comprehensiveness of answers in past conversations. By analyzing historical conversations between the inbound conversation management agent and customers, it meticulously checks whether the answers in the knowledge base match the questions posed by the customer. When incorrect or incomplete answers are found in the knowledge base, the agent flags these entries for manual review and modification, ensuring that the information provided to customers is accurate and reliable. If the knowledge base lacks answers to certain questions, the agent will provide supplementary suggestions and submit them for manual review to ensure that the knowledge base fully covers potential customer questions and avoid "hallucination questions," where large models generate erroneous or fabricated answers without sufficient supporting information. Through these operations, the agent can promptly make necessary additions and modifications to the knowledge base, improving its accuracy and coverage, and ensuring that higher-quality, more comprehensive answers can be provided in subsequent conversations.

[0081] In some embodiments, during the operation of the dialogue management agent, intent recognition agent, outbound decision agent, outbound execution agent, and knowledge management agent in the above embodiments, each agent continuously performs self-adaptation and self-learning to improve its own processing capabilities. Specifically:

[0082] The dialogue management agent, intent recognition agent, outbound decision agent, outbound execution agent, and knowledge management agent all use reinforcement learning algorithms to iteratively optimize their respective execution strategies π while maximizing their respective value functions. i ( a | s ), where i represents the i-th agent, a is the state of the environment perceived by the agent, s Actions performed by the agent.

[0083] It should be noted that the improvement of the capabilities of multiple agents is optimized through the idea of reinforcement learning. In a multi-agent environment, each agent has its own strategy π i and the value function Q i(s, a), the goal of the agent is to maximize its cumulative reward, corresponding to improving indicators such as customer satisfaction and conversion rate. For each agent i, update its value function:

[0084]

[0085] Where s represents the state of the environment perceived by the agent; a represents the actions that the agent can perform; and y represents the discount factor for future rewards. The value function is then used to improve the strategy to obtain the maximum value for each agent: .

[0086] Through the above-mentioned multi-agent collaborative reinforcement learning, the capabilities of the entire multi-agent system are enhanced: ① The agents learn how to coordinate incoming and outgoing tasks more effectively, thereby providing more efficient intelligent scheduling solutions; ② The agents are able to adapt to the strategy changes of other agents; ③ The system remains stable in the face of environmental changes or unexpected behaviors of other agents.

[0087] It should be noted that the steps shown in the above process or the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0088] The embodiment of the present application provides an integrated customer service system based on multiple agents, which includes a dialogue management module, a knowledge management module, an intention recognition module, an outbound decision module, and an outbound execution module;

[0089] The dialogue management module is used to obtain consulting questions raised by active inbound users, process the multidimensional data information of active inbound users through the dialogue management agent, and generate dialogue responses to the consulting questions based on the multidimensional data information;

[0090] The knowledge management module is used to analyze historical conversation records between the dialogue management agent and the active incoming user through the knowledge management agent; if it is determined that the answer to the question given by the dialogue management agent does not match the consultation question posed by the active incoming user, the knowledge management agent will mark the corresponding data entry in the knowledge base for manual modification or make suggestions for data supplementation to the knowledge base;

[0091] The intent recognition module is used to obtain the consultation intention of the active incoming user based on the feedback from the active incoming user after the dialogue response through the intent recognition agent analysis;

[0092] The outbound decision module is used to determine the outbound operation that needs to be performed through the outbound decision agent based on the consultation intention and recent activity records of the active inbound user;

[0093] The outgoing call execution module is used to execute the outgoing call operation through the outgoing call execution agent to conduct an outgoing call dialogue with the active incoming call user.

[0094] Through the dialogue management module, intent recognition module, outbound decision module and outbound execution module in the embodiments of the present application, integrated customer service of multiple intelligent agents is realized, and the collaborative division of labor of multiple different intelligent agents is achieved, each focusing on its specific business process, such as context understanding, sentiment analysis, knowledge updating, intent understanding, etc., thereby providing customers with more professional and personalized services. It can also actively contact customers by analyzing customer data, track customer issues or conduct proactive marketing, which can effectively improve customer satisfaction, help build more stable customer relationships, and solve the problem of how to improve the service capabilities of intelligent customer service.

[0095] It should be noted that the above modules can be functional modules or program modules, and can be implemented through software or hardware. For modules implemented through hardware, the above modules can be located in the same processor; or the above modules can be located in different processors in any combination.

[0096] This embodiment further provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0097] Optionally, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0098] It should be noted that the specific examples in this embodiment can refer to the examples described in the above embodiments and optional implementation modes, and this embodiment will not be repeated here.

[0099] In addition, in conjunction with the multi-agent-based integrated customer service method in the above embodiments, embodiments of the present application may provide a storage medium for implementation. The storage medium stores a computer program; when the computer program is executed by a processor, it implements any of the multi-agent-based integrated customer service methods in the above embodiments.

[0100] In one embodiment, a computer device is provided, which may be a terminal. The computer device includes a processor, memory, a network interface, a display screen, and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal via a network connection. When executed by the processor, the computer program implements an integrated customer service method based on multiple agents. The display screen of the computer device may be a liquid crystal display or an electronic ink display screen. The input device of the computer device may be a touch layer covering the display screen, or may be buttons, a trackball, or a touchpad provided on the computer device housing, or may be an external keyboard, touchpad, or mouse.

[0101] In one embodiment, Figure 3 is a schematic diagram of the internal structure of an electronic device according to an embodiment of the present application, such as Figure 3 As shown, an electronic device is provided, which may be a server, and its internal structure diagram may be as shown in FIG. Figure 3 As shown. The electronic device includes a processor, a network interface, an internal memory, and a non-volatile memory connected via an internal bus, wherein the non-volatile memory stores an operating system, a computer program, and a database. The processor is used to provide computing and control capabilities, the network interface is used to communicate with external terminals via a network connection, the internal memory is used to provide an environment for the operation of the operating system and computer program, and when the computer program is executed by the processor, it implements an integrated customer service method based on multiple agents. The database is used to store data.

[0102] Those skilled in the art will understand that Figure 3 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the electronic device to which the solution of the present application is applied. The specific electronic device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0103] Those skilled in the art will understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0104] Those skilled in the art should understand that the various technical features of the above-described embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the various technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0105] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art could make various modifications and improvements without departing from the spirit of the present application, all of which fall within the scope of protection of the present application. Therefore, the scope of protection of the present patent application shall be determined by the appended claims.

Claims

1. An integrated customer service method based on multi-agent, characterized in that: The method comprises: Obtain consultation questions raised by active inbound users; The multidimensional data information of the active incoming user is obtained through the dialogue management agent, and a dialogue response to the consultation question is generated based on the multidimensional data information; Based on the feedback of the active incoming user after the dialogue response, the consultation intention of the active incoming user is obtained through analysis by the intention recognition agent; Based on the consultation intention of the active incoming user, the outbound decision agent determines the task type of the outbound operation to be performed, wherein the task type includes coupon issuance, renewal reminder, and product promotion; based on the recent activity record of the active incoming user, the outbound decision agent triggers the execution of the outbound operation; Executing the outgoing operation through an outgoing execution agent to conduct an outgoing dialogue with the active incoming user; The historical conversation records between the dialogue management agent and the active incoming user are analyzed by the knowledge management agent; if it is determined that the answer to the question given by the dialogue management agent does not match the consultation question raised by the active incoming user, the corresponding data entry is marked in the knowledge base by the knowledge management agent for manual modification, or suggestions for data supplementation to the knowledge base are made.

2. The method according to claim 1, characterized in that The method comprises: The dialogue management agent, the intention recognition agent, the outbound decision agent, the outbound execution agent and the knowledge management agent all iteratively optimize their respective execution strategies π under the premise of maximizing their respective value functions through reinforcement learning algorithms. i ( a | s ), where i represents the i-th agent, a is the state of the environment perceived by the agent, s Actions performed by the agent.

3. The method according to claim 1, characterized in that The multi-dimensional data information of the active incoming user obtained through the dialogue management agent processing includes: Converting the consultation questions of the active incoming user into streaming data; For streaming data that can be processed in real time by a voice activity detection model in a dialogue management agent, the voice activity detection model is input for consumption to detect the start and end time points of valid speech in the consultation question; For streaming data that cannot be processed in real time by the voice activity detection model, it is not input into the voice activity detection model, but is consumed quickly through the data packets in the buffer; Then, the automatic speech recognition model in the dialogue management agent is used to convert the effective speech within the start and end time points into natural language text, and the multi-dimensional data information of the active incoming user is obtained based on the natural language text.

4. The method according to claim 1, wherein The multidimensional data information of the active incoming user is obtained by processing the dialogue management agent, and the dialogue response to the consultation question is generated based on the multidimensional data information, including: Processing the consultation questions of the active incoming user through the dialogue management agent to obtain user data information, emotion data information and answer data information of the active incoming user; Based on the user data information, the emotion data information and the answer data information, a dialogue response to the consultation question is generated by the dialogue management agent.

5. The method according to claim 4, characterized in that Generating a dialogue response to the consultation question by the dialogue management agent based on the user data information, the emotion data information, and the answer data information includes: Based on the emotional data information, the dialogue management agent determines whether manual customer service is required, and if so, transfers the call to manual customer service for processing; If not, a dialogue response to the consultation question is generated by the dialogue management agent based on the user data information and the answer data information.

6. The method according to claim 1, characterized in that Based on the feedback of the active incoming user after the dialogue response, the consultation intention of the active incoming user obtained through the intention recognition agent analysis includes: Based on the feedback of the active incoming user after the dialogue response, the intention recognition agent determines whether the consultation problem of the active incoming user has been solved; If not, continue to process the consultation questions of the active incoming user through the dialogue management agent; If so, the current conversation with the active incoming user is ended, and based on the content of the current conversation, the consultation intention of the active incoming user is obtained through intention recognition agent analysis.

7. The method according to claim 1, characterized in that Executing the outgoing operation by the outgoing execution agent to conduct an outgoing dialogue with the active incoming user includes: In the case where the outbound decision agent triggers an outbound operation, based on the user contact information provided by the outbound decision agent and the task type of the outbound operation, the outbound operation is executed by the outbound execution agent to conduct an outbound dialogue with the active incoming user.

8. An integrated customer service system based on multi-agent, characterized by: The system is used to execute the method according to any one of claims 1 to 7, and the system includes a dialogue management module, a knowledge management module, an intention recognition module, an outbound decision module, and an outbound execution module; The dialogue management module is used to obtain consulting questions raised by active inbound users, obtain multi-dimensional data information of the active inbound users through processing by the dialogue management agent, and generate dialogue responses to the consulting questions based on the multi-dimensional data information; The knowledge management module is configured to analyze historical conversation records between the dialogue management agent and the active incoming user through a knowledge management agent; if it is determined that the answer to the question given by the dialogue management agent does not match the consultation question raised by the active incoming user, the knowledge management agent is configured to mark the corresponding data entry in the knowledge base for manual modification or to make suggestions for data supplementation to the knowledge base; The intention recognition module is used to obtain the consultation intention of the active incoming user through the intention recognition agent analysis based on the feedback of the active incoming user after the dialogue response; The outbound call decision module is used to determine the outbound call operation to be performed through the outbound call decision agent according to the consultation intention and recent activity record of the active incoming user; The outgoing call execution module is used to execute the outgoing call operation through the outgoing call execution agent to conduct an outgoing call dialogue with the active incoming call user.

Images