Intelligent Customer Service Methods and Systems

Abstract

This application discloses an intelligent customer service method and system. The method includes: invoking an intelligent agent to engage in a conversation with a client via a first channel, and sending the conversation content to a management terminal via a second channel; responding to a complete takeover command sent by the management terminal via the second channel, pausing the intelligent agent's session processing, sending new user questions from the session to the management terminal via the second channel, and sending a human response from the management terminal via the second channel back to the client via the first channel; responding to an auxiliary modification command sent by the management terminal via the second channel, invoking the intelligent agent to generate an intelligent response to the new user questions from the session, sending the new user questions and the intelligent response to the management terminal via the second channel, and sending a human response determined by the management terminal based on the intelligent response via the second channel back to the client via the first channel. This application solves the technical problem of a relatively disconnect between AI and human services in traditional intelligent customer service, resulting in a poor user experience.

Description

Technical Field

[0001] This application relates to the field of customer service technology, and more specifically, to an intelligent customer service method and system. Background Technology

[0002] Current intelligent customer service systems generally adopt a one-way transfer model of "AI (Artificial Intelligence) first, human backup." When AI cannot accurately respond to user questions, the system usually pops up a "transferring you to a human customer service representative" message, completely handing over the conversation to human handling. However, this model has significant drawbacks: First, users can clearly perceive the switch in service provider, creating a disconnect in interaction, weakening service continuity and user trust in customer service professionalism; second, the transfer process is usually accompanied by queuing and significant response delays, affecting user experience; third, AI completely withdraws after human intervention, unable to provide knowledge assistance or collaborative support, resulting in low resource utilization; fourth, existing systems lack real-time monitoring capabilities for AI response quality, and administrators can only review the conversation afterward, making it difficult to intervene promptly when problems occur. More importantly, existing solutions only support one intervention method: "full takeover," which cannot handle many lightweight scenarios that only require minor adjustments to wording, supplementary details, or enhanced professionalism, leading to a dilemma of either excessively high human costs or insufficient AI response quality. Meanwhile, existing systems generally do not record the type, frequency, and content of intervention behaviors, and lack quantitative analysis methods, which makes the optimization of AI models lack data support.

[0003] There is currently no effective solution to the above problems. Summary of the Invention

[0004] This application provides an intelligent customer service method and system to at least solve the technical problem of the relatively disconnect between AI services and human services in traditional intelligent customer service, resulting in a poor user experience.

[0005] According to one aspect of the embodiments of this application, an intelligent customer service method is provided, comprising: invoking a target intelligent agent in an intelligent customer service engine to conduct a conversation with a client via a first channel, and sending the conversation content to a management terminal via a second channel, wherein the target intelligent agent replies to the client according to a target message format; upon receiving a complete takeover instruction sent by the management terminal via the second channel, pausing the target intelligent agent's processing of the current conversation, sending new user questions in the current conversation to the management terminal via the second channel, and sending a first manual reply content fed back by the management terminal via the second channel to the client via the first channel according to the target message format; upon receiving an auxiliary modification instruction sent by the management terminal via the second channel, invoking the target intelligent agent to generate intelligent reply content for new user questions in the current conversation, sending the new user questions and intelligent reply content to the management terminal via the second channel, and sending a second manual reply content determined by the management terminal based on the intelligent reply content via the second channel to the client via the first channel according to the target message format.

[0006] Optionally, after receiving the intervention release command sent by the management terminal via the second channel, the target agent is re-invoked to conduct a session with the client via the first channel.

[0007] Optionally, during the session, each round of dialogue is stored in a session record database, wherein each round of dialogue includes a user question and the corresponding intelligent response and / or human response; a dialogue tag is added to each round of dialogue, wherein the dialogue tag includes: an intervention identifier indicating whether the dialogue has been intervened by the management terminal, an intervention mode identifier indicating whether the management terminal has completely taken over or assisted in modification, and the identity identifier of the administrator who intervened in the dialogue through the management terminal; a message tag is added to each message in each round of dialogue, wherein the message tag includes: message source and extended metadata, the message source includes one of the following: client, agent, management terminal, and the extended metadata includes at least one of the following: message routing method, response time, and confidence level of agent response content.

[0008] Optionally, the second manual response content determined by the management terminal based on the intelligent response content includes one of the following: the intelligent response content that is used directly as the second manual response content without modification, or the second manual response content obtained after manually modifying the intelligent response content; the message marker corresponding to the intelligent response content also includes: a modification identifier used to indicate whether the intelligent response content has been manually modified.

[0009] Optionally, the management terminal is used to display the latest messages in each active session in the interactive interface. The method further includes: in response to a query command for the target session sent by the management terminal based on the second channel, sending the session records of the target session to the management terminal based on the second channel. The management terminal is used to display the session records in the interactive interface. The session records include: all session messages in the target session and the corresponding dialogue tags and message tags.

[0010] Optionally, the running status information of each intelligent agent in the intelligent customer service engine is determined, and the running status information is sent to the management terminal via a second channel. The management terminal is used to display the running status information of each intelligent agent in the interactive interface. The running status information includes at least one of the following: number of active sessions, cumulative number of messages processed, and average response time.

[0011] Optionally, for each agent in the intelligent customer service engine, the conversation intervention status information of the agent is periodically statistically analyzed. The conversation intervention status information includes: the first proportion of the number of conversations intervened by the management terminal within the target period to the total number of conversations, and the second proportion of the number of second manual replies involving manual modification in the assisted modification mode to the total number of second manual replies. If the first proportion is greater than the first preset threshold and / or the second proportion is greater than the second preset threshold, the agent is optimized and trained based on all manual replies and the corresponding user questions.

[0012] According to another aspect of the embodiments of this application, an intelligent customer service system is also provided, including: a client, an intelligent customer service engine including at least one intelligent agent, a central processing module, a management terminal, a first channel, and a second channel, wherein the client is used to interact with users; the management terminal is used to interact with administrators; the central processing module is used to call a target intelligent agent in the intelligent customer service engine to conduct a conversation with the client based on the first channel, and send the conversation content to the management terminal based on the second channel, wherein the target intelligent agent replies to the client according to a target message format; after receiving a complete takeover instruction sent by the management terminal based on the second channel, the system suspends the target intelligent agent's processing of the current conversation, sends new user questions in the current conversation to the management terminal based on the second channel, and feeds back the first manual reply content fed back by the management terminal based on the second channel to the client according to the target message format via the first channel; after receiving an auxiliary modification instruction sent by the management terminal based on the second channel, the system calls the target intelligent agent to generate intelligent reply content for new user questions in the current conversation, sends the new user questions and intelligent reply content to the management terminal based on the second channel, and feeds back the second manual reply content determined by the management terminal based on the intelligent reply content via the second channel to the client according to the target message format via the first channel.

[0013] According to another aspect of the embodiments of this application, a computer program product is also provided, the computer program product comprising: a computer program, wherein the computer program, when executed by a processor, implements the above-described intelligent customer service method.

[0014] According to another aspect of the embodiments of this application, an electronic device is also provided, the electronic device including: a memory and a processor, wherein the memory stores a computer program, and the processor is configured to execute the above-described intelligent customer service method through the computer program.

[0015] In this embodiment, by default, the target intelligent agent in the intelligent customer service engine initiates a conversation with the client via a first channel and sends the conversation content to the management terminal via a second channel. The second channel, independent of the first channel, allows the management terminal to monitor the current conversation status in real time and promptly select the appropriate intervention mode to intervene as needed. Upon receiving a complete takeover command from the management terminal via the second channel, the target intelligent agent's processing of the current conversation is paused. New user questions from the current conversation are directly sent to the management terminal via the second channel for manual response. The management terminal then sends the first manual response from the second channel back to the client via the first channel. Upon receiving a command from the management terminal via the second channel... After receiving the modification instruction, the target agent first generates an intelligent response to the new user question in the current session. Then, the user question and the intelligent response are sent to the management terminal via a second channel. The management terminal confirms whether adjustments are needed and then sends the second human response from the management terminal via the second channel back to the client via the first channel. The two intervention modes can be flexibly switched, and both the intelligent agent response and the human response after the management terminal intervenes use the same target message format. This ensures that the client interface is completely consistent, and the user always perceives a continuous and unified customer service. In other words, this application effectively solves the technical problem of the relatively disconnect between AI and human services in traditional intelligent customer service, resulting in a poor user experience. Attached Figure Description

[0016] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0017] Figure 1 This is a schematic diagram of the structure of an optional intelligent customer service system according to an embodiment of this application;

[0018] Figure 2 This is a schematic diagram of a customer service process under an optional full takeover mode according to an embodiment of this application;

[0019] Figure 3This is a schematic diagram of a customer service process under an optional auxiliary modification mode according to an embodiment of this application;

[0020] Figure 4 This is a schematic diagram of an optional management terminal interface according to an embodiment of this application;

[0021] Figure 5 This is a flowchart illustrating an optional intelligent customer service method according to an embodiment of this application;

[0022] Figure 6 This is a schematic diagram of the structure of an optional electronic device according to an embodiment of this application. Detailed Implementation

[0023] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0024] It should be noted that the terms "first," "second," etc., used in the specification, claims, and drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0025] Example 1

[0026] To address the technical problem of a disconnect between AI and human services in traditional intelligent customer service, resulting in a poor user experience, this application first provides a novel intelligent customer service system architecture, such as... Figure 1 As shown, the system includes: a client 11, an intelligent customer service engine 12 including at least one intelligent agent, a central processing module 13, a management terminal 14, a first channel 15, and a second channel 16.

[0027] The client 11 is used to interact with the user and can be a terminal such as a mobile phone, tablet, or computer used by the user. It usually provides a human-computer interaction interface for users to ask questions and display answers to questions.

[0028] The intelligent customer service engine 12 can include multiple intelligent agents for different types of questions, which are used to handle user questions and generate responses.

[0029] The central processing module 13 is the core module of the intelligent customer service system. It is used to maintain the intervention status of each session and determine the routing method of the message flow according to the intervention mode.

[0030] The management terminal 14 is used to interact with the administrator. It can be a terminal such as a mobile phone, tablet, or computer used by the administrator. It usually provides a human-computer interaction interface that allows the administrator to understand the current session status, intervene in the session as needed, and manually respond to user questions in the session.

[0031] Channel 15 is the client communication channel, which can adopt the WebSocket architecture. Its connection endpoint can be set to ws: / / host / ws / chat?session_id={session identifier}. It is mainly responsible for real-time message communication between the client and the backend service, transmitting user messages and system responses.

[0032] The second channel 16 is the management monitoring channel, which is independent of the first channel 15. It can also use the WebSocket architecture, and its connection endpoint can be set to ws: / / host / ws / admin / monitor. It is used by the management end to receive all session event pushes in real time and execute intervention operation commands.

[0033] Optionally, the second channel 16 can push the following event types: 1) new_conversation: New session creation event, containing user information and session ID; 2) new_message: New message event for the user, containing complete message content and session context; 3) bot_reply: AI reply event, containing reply content, agent type and routing decision metadata; 4) draft_reply: AI reply event in assisted modification mode, awaiting administrator approval; 5) session_closed: Session closing event; 6) bot_stats_update: AI running status statistics update event.

[0034] Optionally, the central processing module 13 can perform the following corresponding operations in different scenarios:

[0035] In a scenario where the management terminal 14 does not intervene, the target intelligent agent in the intelligent customer service engine 12 initiates a conversation with the client 11 via the first channel 15 and sends the conversation content to the management terminal 14 via the second channel 16. The target intelligent agent replies to the client 11 according to the target message format.

[0036] After receiving the complete takeover command sent by the management terminal 14 via the second channel 16, the target agent's processing of the current session is suspended. New user questions in the current session are sent to the management terminal 14 via the second channel 16. The first manual reply content fed back by the management terminal 14 via the second channel 16 is fed back to the client 11 via the first channel 15 according to the target message format.

[0037] After receiving the auxiliary modification instruction sent by the management terminal 14 via the second channel 16, the target agent is invoked to generate intelligent response content for the new user question in the current session. The new user question and intelligent response content are sent to the management terminal 14 via the second channel 16. The second manual response content determined by the management terminal 14 based on the intelligent response content is fed back by the management terminal 14 via the second channel 16 and fed back to the client 11 via the first channel 15 according to the target message format.

[0038] After receiving the intervention release command sent by the management terminal 14 via the second channel 16, the target agent is re-invoked to communicate with the client 11 via the first channel 15.

[0039] The intelligent customer service system proposed in this application, by setting up a separate first channel and an independent second channel, allows the management end to understand the current conversation status in real time, enabling intervention as soon as a problem occurs, rather than remediation afterward. It provides two intervention modes: full takeover and assisted modification. Full takeover is suitable for scenarios where the AI's answer is seriously inappropriate or requires complex judgment, while assisted modification is suitable for scenarios where the AI's answer is basically correct but requires fine-tuning of wording or supplementary information. Administrators can flexibly switch between intervention modes according to the actual situation. Moreover, whether it is the intelligent agent's reply or the human reply after the management end intervenes, the same target message format is used. This ensures that the client interface is completely consistent, and users always perceive continuous and unified customer service, avoiding the service break and loss of trust caused by traditional human intervention solutions.

[0040] As an optional implementation, the intelligent customer service system in this application embodiment also includes a session record database, which is used to record intervention behavior metadata to facilitate subsequent querying and auditing of session data, and can also be used to perform quality assessment and continuous optimization of the intelligent agent based on historical session data.

[0041] Optionally, during the conversation, the central processing module can store each round of dialogue in the conversation record database, wherein each round of dialogue includes a user question and the corresponding intelligent response and / or manual response.

[0042] Optionally, the central processing module can add dialogue markers to each round of dialogue. These dialogue markers include: an intervention identifier indicating whether the dialogue has been intervened by the management terminal; an intervention mode identifier indicating whether the management terminal has completely taken over or is assisting in the modification; and the identity identifier of the administrator who intervened in the dialogue through the management terminal.

[0043] The intervention identifier can be denoted as is_taken_over. When is_taken_over=1, it means that the conversation has been intervened by the management terminal. When is_taken_over=0, it means that the conversation has not been intervened by the management terminal. The intervention mode identifier can be denoted as intervention_mode. When intervention_mode=null, it means that the conversation has not been intervened by the management terminal. When intervention_mode=takeover, it means that the conversation has been completely taken over by the management terminal. When intervention_mode=assist, it means that the conversation has been modified with the assistance of the management terminal. The administrator's identity identifier can be denoted as taken_over_by. The specific value is the account identifier of the administrator who logged into the management terminal and intervened in the conversation.

[0044] Optionally, the central processing module can also add a message tag to each message in each round of dialogue. The message tag includes at least the message source and extended metadata. The message source includes one of the following: client, agent, management end. The extended metadata includes at least one of the following: message routing method, response time, and confidence level of the agent's reply content.

[0045] The message source can be denoted as sender_type. When sender_type=user, it means the message comes from the client. When sender_type=bot, it means the message comes from the agent. When sender_type=admin, it means the message comes from the management end. When sender_type=system, it means the message is an internal system message. Extended metadata can be denoted as metadata, and the specific values ​​are usually in JSON format.

[0046] As an optional implementation, in the assisted modification mode, the intelligent response of the target agent to the user's question can be sent to the management terminal as a draft. After the administrator evaluates the draft, if he thinks that no modification is needed, he can directly confirm that the draft is used as the second human response. If he thinks that the draft has flaws and needs to be modified, he can manually enter the modified content as the second human response.

[0047] To further differentiate between the two types of responses and to facilitate targeted optimization of the agent in the future, the message tags for intelligent responses may also include a modification identifier to indicate whether the intelligent response has been manually modified.

[0048] The modification flag can be denoted as is_human_edited. When is_human_edited=0, it means that the intelligent reply content has not been manually modified, and when is_human_edited=1, it means that the intelligent reply content has been manually modified.

[0049] The following description, in conjunction with the accompanying diagrams and specific implementation scenarios, illustrates the interaction between the various modules of the system under the two intervention modes.

[0050] Figure 2 This is the customer service process under the complete takeover model, specifically including the following steps S1-S8:

[0051] S1, the administrator clicks the "Full Takeover" button in the interactive interface of the management terminal;

[0052] S2, the agent pauses processing the current session;

[0053] S3, the user enters a new question in the client's interactive interface;

[0054] S4, the central processing module sends the new question to the management terminal via the second channel and stores the new question in the session record database. It adds a dialogue marker to this round of dialogue: is_taken_over=1, intervention_mode=takeover, taken_over_by=administrator ID, and adds a message marker to the new question: sender_type=user;

[0055] S5, the administrator enters a manual reply in the interactive interface of the management terminal;

[0056] S6, the central processing module sends the manual reply to the client via the first channel and stores the manual reply in the session record database, adding a message tag to the manual reply: sender_type=admin;

[0057] S7, Does the administrator click the "Intervention and Release" button in the management interface? If not clicked, repeat steps S3-S6; if clicked, proceed to step S8;

[0058] S8, the agent continues to process the current session.

[0059] The following is a specific application scenario interaction flow:

[0060] A consumer inquired, "I have diabetes, can I take health product A?" The AI ​​agent generated a standard response about the nutritional components of health product A based on product information in its knowledge base, but it did not fully consider the special dietary precautions for diabetic patients, nor did it provide the necessary disclaimer that "it is recommended to take it under the guidance of a doctor."

[0061] The customer service manager saw this reply and its routing decision information (tech agent, confidence level 0.87) in the real-time monitoring panel on the management side. Determining that the issue involved health and safety and required a more cautious response, the manager immediately clicked the "Full Takeover" button. The agent stopped processing the session.

[0062] When the user followed up with questions like, "How much should I take each day? What precautions should I take?", the message was no longer routed through the AI ​​agent but was instead pushed directly to the customer service manager via a second channel. The customer service manager, relying on professional experience, manually replied: "Health product A is a nutritional supplement. We recommend consulting your doctor before consuming it. If your doctor permits, you can start with a lower dosage (1 tablet once daily) and monitor your blood sugar levels." Throughout the process, the user only saw the "product health assistant" as the sender and was unaware that the response had switched from AI to a human customer service representative.

[0063] After the customer service supervisor finishes handling the issue, they click the "Intervention and Release" button, and the AI ​​agent resumes processing subsequent messages.

[0064] Figure 3 This is the customer service process in the auxiliary modification mode, specifically including the following steps S1-S8:

[0065] S1, the administrator clicks the "Assisted Modification" button in the interactive interface of the management terminal;

[0066] S2, the user enters a new question in the client's interactive interface;

[0067] S3, the central processing module routes the new question to the agent and adds dialogue markers to this round of dialogue: is_taken_over=1, intervention_mode=assist, taken_over_by=administrator ID, and adds message markers to the new question: sender_type=user;

[0068] S4, the agent generates a smart response (draft).

[0069] S5, the central processing module sends new questions and intelligent replies to the management terminal via the second channel, and stores the intelligent replies in the session record database, adding a message tag to the intelligent replies: sender_type=bot;

[0070] S61, the administrator confirmed that the smart reply does not require modification;

[0071] S71, the central processing module sends the intelligent reply to the client via the first channel and adds a message flag to the intelligent reply: is_human_edited=0;

[0072] S62, the administrator modifies the smart reply to get a human reply;

[0073] S72, the central processing module sends the manual reply to the client via the first channel, adds a message tag to the intelligent reply: is_human_edited=1, stores the manual reply in the session record database, and adds a message tag to the manual reply: sender_type=admin;

[0074] S8, Does the administrator click the "Intervention Release" button in the management interface? If not clicked, repeat steps S2-S7; if clicked, proceed to step S9;

[0075] S9, the agent continues to handle the current session independently.

[0076] The following is a specific application scenario interaction flow:

[0077] A distributor customer inquired about discount strategies for large group purchase orders.

[0078] After searching the company's strategy knowledge base, the intelligent agent generated a draft reply: "We have group-buying discounts; you can get a discount if you buy more." While the direction of this reply is correct, it lacks specific discount tiers and professional business language, making it unsuitable for dealing with distributor customers.

[0079] The customer service manager observed the conversation in the real-time monitoring panel on the management side and identified it as an important distributor customer (the customer background showed that their cumulative order amount in the past six months reached 300,000 yuan). They then clicked the "Assisted Modification" button to enter the assisted modification mode.

[0080] The agent's draft reply was sent to the customer service manager for review via a second channel. The manager revised the draft to: "Thank you for your continued trust! For group purchase orders, we offer the following tiered discounts: 5% off for orders of 10 boxes or more, 10% off for orders of 50 boxes or more, and a special discount program available for orders of 100 boxes or more. Based on your order history, you have reached the Silver Distributor level and are eligible for an additional annual rebate program. For detailed information, I can arrange a dedicated account manager to assist you." After confirmation, the final reply was sent to the distributor via the first channel. Throughout the process, the distributor only saw the "Order Inquiry Assistant" as the sender and was unaware of the changes in the respondent's identity.

[0081] After the customer service supervisor finishes handling the issue, they can click the "Intervention Release" button, and the AI ​​agent will resume handling subsequent messages independently.

[0082] As an optional implementation, in order to allow administrators to intuitively and in real-time obtain the status of all current sessions, the management terminal can display the latest messages in each active session in the interactive interface.

[0083] The central processing module can respond to a query command sent by the management terminal via the second channel for a target session, retrieve relevant session records for the target session from the session record database, and send the session records of the target session to the management terminal via the second channel. The management terminal can then display the session records in the interactive interface. The session records include all session messages in the target session, along with their corresponding dialogue tags and message tags.

[0084] Optionally, the central processing module can also determine the operational status information of each intelligent agent in the intelligent customer service engine and send the operational status information to the management terminal via a second channel. The management terminal can display the operational status information of each intelligent agent in the interactive interface. The operational status information includes at least one of the following: number of active sessions, cumulative number of messages processed, and average response time.

[0085] Figure 4 An optional management interface is shown. The left side is the conventional management interface, which displays the current running status information of each agent and the latest messages of multiple active sessions. The right side is the specific session interface after clicking on session A, which includes all relevant records of session A and provides "Complete Takeover", "Assisted Modification", and "Intervention Release" buttons.

[0086] As an optional implementation, in order to improve the response quality of the intelligent agent and minimize the frequency of administrator intervention in the conversation, for each intelligent agent in the intelligent customer service engine, the central processing module can periodically collect the conversation intervention status information of the intelligent agent. The conversation intervention status information includes: the first proportion of the number of conversations intervened by the management terminal within the target period to the total number of conversations, and the second proportion of the number of second manual responses involving manual modification in the assisted modification mode to the total number of second manual responses. If the first proportion is greater than the first preset threshold and / or the second proportion is greater than the second preset threshold, it indicates that the number of times the administrator intervenes in the conversation is relatively high, that is, the response quality of the corresponding intelligent agent needs to be improved. At this time, the intelligent agent can be optimized and trained based on all manual responses and corresponding user questions.

[0087] The intelligent customer service system proposed in this application, by setting up a separate first channel and an independent second channel, allows the management end to understand the current conversation status in real time, enabling intervention as soon as a problem occurs, rather than remediation afterward. It provides two intervention modes: full takeover and assisted modification. Full takeover is suitable for scenarios where the AI's answer is seriously inappropriate or requires complex judgment, while assisted modification is suitable for scenarios where the AI's answer is basically correct but requires fine-tuning of wording or supplementary information. Administrators can flexibly switch between intervention modes according to the actual situation. Moreover, whether it is the intelligent agent's reply or the human reply after the management end intervenes, the same target message format is used. This ensures that the client interface is completely consistent, and users always perceive continuous and unified customer service, avoiding the service break and loss of trust caused by traditional human intervention solutions.

[0088] Example 2

[0089] Based on the intelligent customer service system provided in Embodiment 1, this application embodiment also provides an intelligent customer service method implemented by a central processing module. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0090] Figure 5 This is a flowchart illustrating an intelligent customer service method according to an embodiment of this application, such as... Figure 5 As shown, the method includes the following steps S102-S106:

[0091] Step S502: The target intelligent agent in the intelligent customer service engine is invoked to conduct a conversation with the client based on the first channel, and the conversation content is sent to the management terminal based on the second channel. The target intelligent agent replies to the client according to the target message format.

[0092] Step S504: After receiving the complete takeover instruction sent by the management terminal via the second channel, the target agent's processing of the current session is paused, new user questions in the current session are sent to the management terminal via the second channel, and the first manual reply content fed back by the management terminal via the second channel is fed back to the client via the first channel according to the target message format.

[0093] Step S506: After receiving the auxiliary modification instruction sent by the management terminal based on the second channel, the target intelligent agent is invoked to generate intelligent response content for the new user question in the current session. The new user question and intelligent response content are sent to the management terminal based on the second channel. The second manual response content determined by the management terminal based on the intelligent response content is fed back to the client based on the first channel in the target message format.

[0094] Optionally, after receiving the intervention release command sent by the management terminal via the second channel, the target agent can be re-invoked to communicate with the client via the first channel.

[0095] As an optional implementation, in order to facilitate subsequent querying and auditing of session data, as well as to conduct quality assessment and continuous optimization of the agent based on historical session data, each round of dialogue can be stored in a session record database during the session. Each round of dialogue includes a user question and the corresponding intelligent response and / or human response.

[0096] Optionally, dialogue markers can be added to each round of dialogue, wherein the dialogue markers include: an intervention identifier to indicate whether the dialogue has been intervened by the management terminal, an intervention mode identifier to indicate whether the management terminal has completely taken over or assisted in modification, and the identity identifier of the administrator who intervened in the dialogue through the management terminal.

[0097] Optionally, a message tag can be added to each message in each round of dialogue. The message tag includes: message source and extended metadata. The message source includes one of the following: client, agent, management end. The extended metadata includes at least one of the following: message routing method, response time, and confidence level of the agent's reply content.

[0098] Optionally, in the assisted modification mode, the second manual response content determined by the management terminal based on the intelligent response content includes one of the following: the intelligent response content that is used directly as the second manual response content without modification, or the second manual response content obtained after manually modifying the intelligent response content.

[0099] To differentiate between the two types of responses mentioned above, and to facilitate targeted optimization of the agent in the future, the message tags for intelligent responses may also include a modification identifier to indicate whether the intelligent response has been manually modified.

[0100] As an optional implementation, in order to allow administrators to intuitively and in real-time obtain the status of all current sessions, the management terminal can display the latest messages in each active session in the interactive interface.

[0101] The central processing module can respond to a query command sent by the management terminal via the second channel for a target session, retrieve relevant session records for the target session from the session record database, and send the session records of the target session to the management terminal via the second channel. The management terminal can then display the session records in the interactive interface. The session records include all session messages in the target session, along with their corresponding dialogue tags and message tags.

[0102] Optionally, the central processing module can also determine the operational status information of each intelligent agent in the intelligent customer service engine and send the operational status information to the management terminal via a second channel. The management terminal can display the operational status information of each intelligent agent in the interactive interface. The operational status information includes at least one of the following: number of active sessions, cumulative number of messages processed, and average response time.

[0103] As an optional implementation, in order to improve the response quality of the intelligent agent and minimize the frequency of administrator intervention in the conversation, for each intelligent agent in the intelligent customer service engine, the central processing module can periodically collect the conversation intervention status information of the intelligent agent. The conversation intervention status information includes: the first proportion of the number of conversations intervened by the management terminal within the target period to the total number of conversations, and the second proportion of the number of second manual responses involving manual modification in the assisted modification mode to the total number of second manual responses. If the first proportion is greater than the first preset threshold and / or the second proportion is greater than the second preset threshold, it indicates that the number of times the administrator intervenes in the conversation is relatively high, that is, the response quality of the corresponding intelligent agent needs to be improved. At this time, the intelligent agent can be optimized and trained based on all manual responses and corresponding user questions.

[0104] It should be noted that each module in the intelligent customer service system in this application embodiment corresponds one-to-one with each implementation step of the intelligent customer service method in embodiment 1. Since embodiment 1 has been described in detail, some details not shown in this embodiment can be referred to embodiment 1, and will not be elaborated further here.

[0105] Example 3

[0106] According to an embodiment of this application, a computer program product is also provided, which includes a computer program, wherein when the computer program is executed by a processor, it implements the intelligent customer service method in embodiment 1.

[0107] According to an embodiment of this application, a non-volatile storage medium is also provided, which includes a stored computer program, wherein the device containing the non-volatile storage medium executes the intelligent customer service method in Embodiment 1 by running the computer program.

[0108] According to an embodiment of this application, a processor is also provided for running a computer program, wherein the computer program executes the intelligent customer service method in Embodiment 1 during runtime.

[0109] According to an embodiment of this application, an electronic device is also provided, comprising: a memory and a processor, wherein the memory stores a computer program, and the processor is configured to execute the intelligent customer service method of Embodiment 1 through the computer program.

[0110] Specifically, the computer program executes the following steps during runtime: It invokes the target intelligent agent in the intelligent customer service engine to initiate a conversation with the client via a first channel, and sends the conversation content to the management terminal via a second channel, wherein the target intelligent agent replies to the client according to the target message format; upon receiving a complete takeover instruction from the management terminal via the second channel, it suspends the target intelligent agent's processing of the current conversation, sends new user questions in the current conversation to the management terminal via the second channel, and sends the first human response content fed back by the management terminal via the second channel to the client via the first channel according to the target message format; upon receiving an auxiliary modification instruction from the management terminal via the second channel, it invokes the target intelligent agent to generate intelligent response content for new user questions in the current conversation, sends the new user questions and intelligent response content to the management terminal via the second channel, and sends the second human response content determined by the management terminal based on the intelligent response content via the second channel to the client via the first channel according to the target message format.

[0111] As an alternative implementation, the above-mentioned electronic device may exist in the form of a mobile terminal, a computer terminal, or a similar computing device. Figure 6 A hardware structure block diagram of an electronic device for implementing an intelligent customer service method is shown. Figure 6 As shown, the electronic device 60 may include one or more (shown as 602a, 602b, ..., 602n) processors 602 (processors 602 may include, but are not limited to, processing devices such as microprocessors or programmable logic devices), a memory 604 for storing data, and a transmission device 606 for communication functions. In addition, it may also include: a display, an input / output interface, a universal serial bus port (which may be included as one of the ports of a BUS bus), a network interface, a power supply, and / or a camera. Those skilled in the art will understand that... Figure 6 The structure shown is for illustrative purposes only and does not limit the structure of the aforementioned electronic device. For example, electronic device 60 may also include components that are more... Figure 6 The more or fewer components shown, or having the same Figure 6 The different configurations shown.

[0112] It should be noted that the aforementioned one or more processors 602 and / or other data processing circuits are generally referred to herein as "data processing circuits". These data processing circuits may be embodied, in whole or in part, in software, hardware, firmware, or any other combination thereof. Furthermore, the data processing circuits may be a single, independent processing module, or may be integrated, in whole or in part, into any other element of the electronic device 60. As involved in the embodiments of this application, the data processing circuits serve as a processor control mechanism (e.g., selection of a variable resistor termination path connected to an interface).

[0113] The memory 604 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the intelligent customer service method in this embodiment. The processor 602 executes various functional applications and data processing by running the software programs and modules stored in the memory 604, thereby realizing the aforementioned intelligent customer service method. The memory 604 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 604 may further include memory remotely located relative to the processor 602, and these remote memories can be connected to the electronic device 60 via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0114] The transmission device 606 is used to receive or send data via a network. Specific examples of the network described above may include a wireless network provided by the communication provider of the electronic device 60. In one example, the transmission device 606 includes a network adapter that can connect to other network devices via a base station to communicate with the Internet. In another example, the transmission device 606 may be a radio frequency module used for wireless communication with the Internet.

[0115] The display may be, for example, a touchscreen LCD display, which allows the user to interact with the user interface of the electronic device 60.

[0116] The sequence numbers of the above embodiments are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0117] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0118] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some interfaces; indirect couplings or communication connections between units or modules may be electrical or other forms.

[0119] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0120] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0121] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, read-only memory, random access memory, portable hard drive, magnetic disk, or optical disk.

[0122] The above are merely preferred embodiments of this application. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. An intelligent customer service method, characterized in that, include: The target intelligent agent in the intelligent customer service engine initiates a conversation with the client through a first channel and sends the conversation content to the management terminal through a second channel. The target intelligent agent replies to the client according to the target message format. Upon receiving the complete takeover instruction sent by the management terminal via the second channel, the target agent's processing of the current session is suspended, new user questions in the current session are sent to the management terminal via the second channel, and the first manual response content fed back by the management terminal via the second channel is fed back to the client via the first channel according to the target message format. Upon receiving the auxiliary modification instruction sent by the management terminal via the second channel, the target agent is invoked to generate intelligent response content for the new user question in the current session. The new user question and the intelligent response content are then sent to the management terminal via the second channel. The second manual response content determined by the management terminal based on the intelligent response content and fed back by the management terminal via the second channel is fed back to the client via the first channel according to the target message format.

2. The method according to claim 1, characterized in that, The method further includes: After receiving the intervention release command sent by the management terminal based on the second channel, the target agent is re-invoked to conduct a session with the client based on the first channel.

3. The method according to claim 1, characterized in that, The method further includes: During the conversation, each round of dialogue is stored in the conversation record database. Each round of dialogue includes a user question and the corresponding intelligent response and / or human response. Add dialogue markers to each round of dialogue, wherein the dialogue markers include: an intervention identifier for indicating whether the dialogue has been intervened by the management terminal, an intervention mode identifier for indicating whether the management terminal has completely taken over or assisted in modification, and an identity identifier of the administrator who intervened in the dialogue through the management terminal. Add a message tag to each message in each round of dialogue, wherein the message tag includes: message source and extended metadata. The message source includes one of the following: client, agent, management end. The extended metadata includes at least one of the following: message routing method, response time, and confidence level of agent's reply content.

4. The method according to claim 3, characterized in that, The second manual response content determined by the management terminal based on the intelligent response content includes one of the following: the intelligent response content that is used directly as the second manual response content without modification, or the second manual response content obtained after manually modifying the intelligent response content. The message markers corresponding to the smart reply content also include a modification identifier to indicate whether the smart reply content has been manually modified.

5. The method according to claim 3, characterized in that, The management terminal is used to display the latest messages in each active session in the interactive interface, and the method further includes: In response to a query command for a target session sent by the management terminal via the second channel, the session records of the target session are sent to the management terminal via the second channel. The management terminal is used to display the session records in an interactive interface. The session records include all session messages in the target session and corresponding dialogue tags and message tags.

6. The method according to claim 1, characterized in that, The method further includes: The operating status information of each intelligent agent in the intelligent customer service engine is determined, and the operating status information is sent to the management terminal via the second channel. The management terminal is used to display the operating status information of each intelligent agent in the interactive interface. The operating status information includes at least one of the following: number of active sessions, number of cumulative processed messages, and average response time.

7. The method according to claim 4, characterized in that, The method further includes: For each agent in the intelligent customer service engine, the conversation intervention status information of the agent is periodically counted. The conversation intervention status information includes: the first proportion of the number of conversations intervened by the management terminal within the target period to the total number of conversations, and the second proportion of the number of second manual replies involving manual modification in the assisted modification mode to the total number of second manual replies. If the first proportion is greater than the first preset threshold and / or the second proportion is greater than the second preset threshold, the agent is optimized and trained based on all manual responses and corresponding user questions.

8. An intelligent customer service system, characterized in that, include: The client includes an intelligent customer service engine comprising at least one intelligent agent, a central processing module, a management terminal, a first channel, and a second channel, wherein... The client is used to interact with the user; The management terminal is used for interaction with the administrator; The central processing module is used to invoke the target intelligent agent in the intelligent customer service engine to conduct a conversation with the client via the first channel, and send the conversation content to the management terminal via the second channel. The target intelligent agent replies to the client according to a target message format. Upon receiving a complete takeover instruction from the management terminal via the second channel, the module suspends the target intelligent agent's processing of the current conversation, sends new user questions from the current conversation to the management terminal via the second channel, and sends the first human response from the management terminal via the second channel back to the client via the first channel according to the target message format. Upon receiving an auxiliary modification instruction from the management terminal via the second channel, the module invokes the target intelligent agent to generate intelligent response content for new user questions in the current conversation, sends the new user questions and the intelligent response content to the management terminal via the second channel, and sends the second human response content determined by the management terminal based on the intelligent response content via the second channel back to the client via the first channel according to the target message format.

9. A computer program product, characterized in that, include: A computer program, wherein when executed by a processor, the computer program implements the intelligent customer service method according to any one of claims 1 to 7.

10. An electronic device, characterized in that, include: A memory and a processor, wherein the memory stores a computer program, and the processor is configured to execute the intelligent customer service method according to any one of claims 1 to 7 through the computer program.

Images