Customer service management system applied to Internet of Things platform

Abstract

The invention relates to the technical field of customer service information processing, in particular to a customer service management system applied to an internet of things platform, which comprises the following steps of: accurately identifying a target user establishing a new dialogue, acquiring user file information of the target user, and performing similar processing in combination with a historical theme dialogue to determine analysis information so as to determine a user dialogue establishment purpose; a front-end dialogue module generates a reply text according to a user dialogue establishment purpose and the real-time question information, establishes a front-end dialogue, performs semantic repetition analysis on the content of the front-end dialogue to obtain a real-time dialogue repetition value, and generates a manual service signal based on the real-time dialogue repetition value; the target user is in communication connection with the manual customer service channel, seamless conversion from the AI customer service to the manual customer service is achieved, problems in a conversation can be found in time, manual services are introduced in time, the service process is optimized, and the service quality is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of customer service information processing, and in particular to a customer service management system applied to an Internet of Things platform. Background Art

[0002] In modern business operations, the customer service management system plays a vital role as a bridge connecting enterprises and customers. An efficient and intelligent customer service system can not only improve customer satisfaction, but also significantly reduce customer service costs and improve business operational efficiency.

[0003] The prior art CN117575615A discloses an omni-channel intelligent customer service management system, which includes: collecting conversation data from different channels, extracting context information from the conversation data collected from each channel; encoding the extracted context information; semantically aligning the semantic feature vectors of different channels; merging the aligned semantic feature vectors for omni-channel consistency, and generating conversation text using a conversation generation model.

[0004] However, when handling complex conversations, intelligent customer service lacks the ability to conduct in-depth analysis of conversation content, cannot promptly detect repeated content in the conversation, and has difficulty flexibly switching service modes according to the conversation situation. The user experience is poor. When users repeatedly ask the same questions or the conversation reaches a deadlock, manual service cannot be introduced in a timely manner, affecting the efficiency of solving user problems and satisfaction. Summary of the Invention

[0005] The purpose of the present invention is to solve the problems in the background technology and to propose a customer service management system applied to the Internet of Things platform.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] A customer service management system applied to an Internet of Things platform, comprising:

[0008] An intelligent processing module is configured to detect and process signals, and based on the processed signals, identify a target user for establishing a new conversation, obtain user profile information of the target user, extract basic identity information and account information from the user profile information, perform similarity processing on the target user's basic identity information and account information in combination with historical topic conversations to obtain a comprehensive similarity value, determine designated analysis information based on the comprehensive similarity value, extract key information from the designated analysis information, and determine the purpose of establishing a conversation with the target user;

[0009] The front-end dialogue module is used to obtain the target user's dialogue establishment purpose. Based on the target user's real-time question information, the module combines the real-time question information with the dialogue establishment purpose to obtain a reply text. The AI customer service obtains the reply text and sends it to the target user's dialog box, thereby establishing a front-end dialogue.

[0010] The process monitoring module is used to automatically detect the conversation process of the front-end conversation, and at the same time divide the conversation content in the front-end conversation into the user conversation end and the intelligent conversation end. Then, the semantic repetition of the content in the user conversation end and the intelligent conversation end is analyzed respectively to obtain the user repetition value and the intelligent repetition value. The user repetition value and the intelligent repetition value are then comprehensively processed to obtain the real-time conversation repetition value of the front-end conversation. Then, based on the conversation repetition value, a manual service signal is generated;

[0011] The signal detection module is used to detect manual service signals, connect the target user with the manual customer service channel, and directly convert the target user's AI customer service into manual customer service.

[0012] As a further embodiment of the present invention, a method for determining a purpose of establishing a dialogue includes:

[0013] S1: Identify the user with whom the new conversation is established and mark them as the target user. Search the database for the target user's data information and mark the search results as user profile information. The user profile information includes basic identity information, account information, and historical consultation and service information. Furthermore, basic identity information refers to the target user's age, gender, and location; account information refers to browsing history and the installed devices corresponding to the account; and historical consultation and service information refers to the target user's historical consultation and maintenance records and the operating data of the installed devices.

[0014] Obtain the target user's operating data within a fixed period of time and mark it as real-time operating data. Based on the real-time operating data, determine whether the target user's installed equipment is in an abnormal state. If so, set the target user's conversation establishment purpose to equipment failure maintenance. Otherwise, generate a deep analysis signal if there is no abnormal state.

[0015] S2: When a deep analysis signal is detected, obtain the target user's basic identity information and account information, set the account information as the first tag, and set the basic identity information as the second tag;

[0016] Obtain historical topic conversations and topic tags in each topic conversation, perform similarity processing on the topic tag and the first tag to obtain a first similarity value XDi, and then perform similarity processing on the topic tag and the second tag to obtain a second similarity value XEi, where i represents different topic conversations. Furthermore, i∈[1,I], where I represents the total number of existing topic conversations;

[0017] The comprehensive similarity value XZi of topic conversation i is obtained using the formula XZi=XDi×a1+XEi×a2, where a1 and a2 are the weight coefficients of the first label and the second label respectively;

[0018] S3: After all historical topic conversations are processed, all topic comprehensive similarity values XZi are obtained, and the topic comprehensive similarity values XZi are compared to obtain the maximum value of the topic comprehensive similarity values XZi. Then, the topic conversation corresponding to the maximum value is identified and marked as the designated analysis information;

[0019] The key information in the specified analysis information is obtained, and the extracted key information is set as the purpose of establishing a dialogue with the target user.

[0020] As a further solution of the present invention, a method for determining an abnormal state includes:

[0021] Identify the operating parameters of the installed equipment and obtain the normal operating range of the operating parameters, compare the real-time operating data with the normal operating range, if the real-time operating data all belong to the normal operating range, it means that there is no abnormal state of the installed equipment, on the contrary, if there is data in the real-time operating data that does not belong to the normal operating range, the operating state of the installed equipment is marked as abnormal state.

[0022] As a further solution of the present invention, if there is no installed device in the target user's account, a deep analysis signal is directly generated.

[0023] As a further embodiment of the present invention, a method for determining topic conversations and topic tags includes:

[0024] Obtain the system's historical conversation information and mark each complete conversation as a topic conversation. Then identify the user corresponding to each topic conversation, obtain the user's basic identity information, and set the basic identity information as the conversation tag.

[0025] Each topic conversation is obtained again, and the TF-IDF algorithm is used to extract the key information in the topic conversation to obtain the central topic words. The central topic words are integrated with the conversation tags of the corresponding topic conversation, and the integrated information is marked as the topic tag of the corresponding topic conversation.

[0026] As a further solution of the present invention, a method for establishing a front-end dialogue includes:

[0027] Obtaining the target user's conversation establishment purpose, setting the conversation establishment purpose as the central word, then obtaining the target user's real-time question information, and extracting key information from the real-time question information, wherein the key information refers to the information formed by extracting content words from the real-time question information and then reorganizing the extracted content words according to corresponding positions, and content words refer to words with real lexical meaning;

[0028] Combine the central vocabulary with the key information, mark the combined information as real-time retrieval information, and search the real-time retrieval information in the data information database. Use the obtained information as the reply text of the real-time question information, and the AI customer service will send the reply text to the target user's dialog box, so that the AI customer service and the target user can establish a front-end dialogue.

[0029] As a further solution of the present invention, a method for determining a conversation repetition value includes:

[0030] SS1: Acquire real-time front-end conversations and divide them into user conversation end and intelligent conversation end. The user conversation end refers to the information sent by the user in the front-end conversation, and the intelligent conversation end refers to the information content of the AI customer service response.

[0031] Extracting the information content of the user's conversation end, identifying meaningless words in the information content of the user's conversation end, then deleting the identified meaningless words from the corresponding information content, and marking the remaining information content as user key information, wherein meaningless words refer to function words and modal particles;

[0032] Using natural language processing technology, the user's key information is transmitted as input to the language model, and the semantic repetition of the user's key information is analyzed to obtain the user repetition value FY;

[0033] Then, the information content of the intelligent dialogue terminal is obtained and directly used as input information. The information content is transmitted to the language model, and the semantic repetition of the AI customer service is analyzed to obtain the intelligent repetition value FZ;

[0034] SS2: Use the formula FY×b1+FZ×b2=FH to obtain the real-time conversation repetition value FH of the front-end conversation, where b1 and b2 are the user weight coefficient and customer service weight coefficient respectively.

[0035] As a further solution of the present invention, a method for determining a manual service signal includes:

[0036] The conversation repetition value FH is compared with the comprehensive repetition threshold. If the conversation repetition value FH is less than the comprehensive repetition threshold, it means that the AI customer service and the target user are communicating effectively. At this time, an intelligent communication signal is generated, and the AI customer service is continued to be set as the docking customer service of the target user. On the contrary, if the conversation repetition value FH is greater than or equal to the comprehensive repetition threshold, it means that there is an abnormality in the communication process between the AI customer service and the target user. At this time, a manual service signal will be generated, and the process monitoring module will transmit the manual service signal to the signal detection module.

[0037] As a further solution of the present invention, it also includes a dialogue detection module for identifying dialogue information in multiple Internet of Things platforms. When the dialogue detection module detects that a new dialogue has been established, the dialogue detection module generates a processing signal and transmits it to the intelligent processing module.

[0038] Compared with the existing technology, the advantages of the present invention are:

[0039] The present invention can quickly detect and process signals through the intelligent processing module, accurately identify the target user for establishing a new dialogue, obtain their user profile information, and perform similar processing in combination with historical topic dialogues to determine analysis information, thereby clarifying the purpose of establishing the user dialogue. Then, the front-end dialogue module generates a reply text based on the purpose of the user dialogue and real-time question information, realizing the intelligent reply of AI customer service, and can flexibly combine information to provide users with services that better meet their needs, enhance the interactive experience, and improve the accuracy and efficiency of communication;

[0040] The present invention uses a process monitoring module to perform semantic repetition analysis on the front-end conversation content to obtain a real-time conversation repetition value, and generates a manual service signal based on this. After the signal detection module detects the manual service signal, it connects the target user with the manual customer service channel to achieve a seamless transition from AI customer service to manual customer service. It can promptly discover problems in the conversation, introduce manual service at the right time, optimize the service process, and improve service quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 Schematic diagram of the system structure of the present invention. DETAILED DESCRIPTION

[0042] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0043] Reference Figure 1 , a customer service management system applied to the Internet of Things platform, including a dialogue detection module, an intelligent processing module, a front-end dialogue module, a process monitoring module and a signal detection module;

[0044] The conversation detection module is used to identify conversation information in multiple IoT platforms. When the conversation detection module detects that a new conversation has been established, it generates a processing signal and transmits it to the intelligent processing module.

[0045] The intelligent processing module is used to receive the processing signal, and when the intelligent processing module receives the processing signal, it identifies the user of the new conversation established and marks it as the target user, then searches the target user's data information in the database, and marks the search results as user profile information, wherein the user profile information includes basic identity information, account information, and historical consultation and service information. Furthermore, the basic identity information refers to the target user's age, gender, and place of residence, the account information refers to the browsing history and the installed equipment corresponding to the account, and the historical consultation and service information refers to the target user's historical consultation and maintenance records and the operating data of the installed equipment;

[0046] The intelligent processing module then analyzes the target user's user profile information based on the target user's user profile information to determine the target user's purpose for establishing a conversation. Furthermore, the method for determining the target user's purpose for establishing a conversation includes:

[0047] S1: Obtain the target user's operating data within a fixed period and mark it as real-time operating data. Based on the real-time operating data, determine whether the target user's installed equipment is in an abnormal state. If so, set the target user's conversation establishment purpose to equipment failure maintenance. Conversely, if no abnormal state exists, generate a deep analysis signal. The specific duration of the fixed period is set by those skilled in the art based on big data experience.

[0048] Furthermore, the abnormal state determination method includes: identifying operating parameters of the installation equipment, obtaining a normal operating range of the operating parameters, comparing real-time operating data with the normal operating range, and if the real-time operating data all fall within the normal operating range, indicating that the installation equipment is not in an abnormal state; conversely, if the real-time operating data includes data that does not fall within the normal operating range, marking the operating state of the installation equipment as an abnormal state;

[0049] In another embodiment of the present invention, if the target user's account does not have an installed device, a deep analysis signal is directly generated;

[0050] S2: When a deep analysis signal is detected, obtain the target user's basic identity information and account information, set the account information as the first tag, and set the basic identity information as the second tag;

[0051] Obtain historical topic conversations and topic tags in each topic conversation, perform similarity processing on the topic tag and the first tag to obtain a first similarity value XDi, and then perform similarity processing on the topic tag and the second tag to obtain a second similarity value XEi, where i represents different topic conversations. Furthermore, i∈[1,I], where I represents the total number of existing topic conversations;

[0052] Then, the comprehensive similarity value XZi of the topic conversation i is obtained using the formula XZi=XDi×a1+XEi×a2, where a1 and a2 are the weight coefficients of the first label and the second label respectively. The specific values of a1 and a2 are obtained by those skilled in the art after big data calculation. In this embodiment, the cosine similarity algorithm is selected as the similarity processing method, and the cosine similarity algorithm belongs to the existing technology and will not be described in detail in this embodiment;

[0053] S3: After all historical topic conversations are processed, all topic comprehensive similarity values XZi are obtained, and the topic comprehensive similarity values XZi are compared to obtain the maximum value of the topic comprehensive similarity values XZi. Then, the topic conversation corresponding to the maximum value is identified and marked as the designated analysis information;

[0054] Then, key information in the specified analysis information is obtained, and the extracted key information is set as the purpose of establishing a conversation with the target user;

[0055] In another embodiment of the present invention, a method for determining a topic conversation and a topic tag includes:

[0056] Obtain the system's historical conversation information and mark each complete conversation as a topic conversation. Then identify the user corresponding to each topic conversation, obtain the user's basic identity information, and set the basic identity information as the conversation tag.

[0057] Obtain each topic conversation and use the TF-IDF algorithm to extract key information from the topic conversation to obtain the central topic word. The central topic word is integrated with the conversation tag of the corresponding topic conversation, and the integrated information is marked as the topic tag of the corresponding topic conversation. It should be further explained that the use of the TF-IDF algorithm to extract keywords is a prior art and will not be further described here.

[0058] The intelligent processing module then transmits the target user's conversation establishment purpose to the front-end conversation module;

[0059] The front-end dialogue module is used to obtain the target user's dialogue establishment purpose and, based on the dialogue establishment purpose, use AI customer service to conduct a front-end dialogue with the target user. The specific front-end dialogue establishment method includes:

[0060] Obtaining the target user's conversation establishment purpose, setting the conversation establishment purpose as the central word, then obtaining the target user's real-time question information, and extracting key information from the real-time question information, wherein the key information refers to information formed by extracting content words from the real-time question information and then reorganizing the extracted content words according to corresponding positions. Furthermore, content words refer to words with real lexical meanings, such as subject, predicate, and object;

[0061] Combine the core vocabulary with the key information, mark the combined information as real-time search information, search the real-time search information in the database, and use the obtained information as the reply text to the real-time question information. The AI customer service will send the reply text to the target user's dialog box, thereby establishing a front-end dialogue between the AI customer service and the target user;

[0062] Afterwards, the front-end dialogue module transmits the target user's front-end dialogue to the process monitoring module;

[0063] The process monitoring module is used to automatically detect and analyze the target user's front-end conversation process and determine the target user's conversation progress. Specifically, the method for determining the conversation progress includes:

[0064] SS1: Acquire real-time front-end conversations and divide them into user conversation end and intelligent conversation end. The user conversation end refers to the information sent by the user in the front-end conversation, and the intelligent conversation end refers to the information content of the AI customer service response.

[0065] First, extract the information content of the user's conversation end, identify the meaningless words in the information content of the user's conversation end, then delete the identified meaningless words from the corresponding information content, and mark the remaining information content as user key information. Among them, meaningless words refer to function words and modal particles, such as conjunctions, prepositions, and auxiliary words;

[0066] Then, using natural language processing technology, the user's key information is transmitted as input to the language model, and the semantic repetition of the user's key information is analyzed to obtain the user repetition value FY;

[0067] Then, the information content of the intelligent dialogue terminal is obtained and directly used as input information. The information content is transmitted to the language model, and the semantic repetition of the AI customer service is analyzed to obtain the intelligent repetition value FZ;

[0068] It should be further explained that natural language processing technology belongs to the existing technology, and the specific process of analyzing semantic repetition values using language models will not be described here in detail;

[0069] SS2: Use the formula FY × b1 + FZ × b2 = FH to obtain the real-time conversation repetition value FH of the front-end conversation, where b1 and b2 are the user weight coefficient and customer service weight coefficient, respectively. The specific values of b1 and b2 are obtained by technicians in this field through big data calculations.

[0070] The conversation repetition value FH is then compared with the comprehensive repetition threshold. If the conversation repetition value FH is less than the comprehensive repetition threshold, it indicates that the AI customer service and the target user are communicating effectively. At this time, an intelligent communication signal is generated, and the AI customer service is continued to be set as the target user's docking customer service. Conversely, if the conversation repetition value FH is greater than or equal to the comprehensive repetition threshold, it indicates that there is an abnormality in the communication process between the AI customer service and the target user. At this time, a manual service signal is generated, and the process monitoring module transmits the manual service signal to the signal detection module. The specific value of the comprehensive repetition threshold is obtained by those skilled in the art after big data calculation.

[0071] The signal detection module is used to detect manual service signals. When the signal detection module detects a manual service signal, it immediately obtains the manual customer service channel and connects the target user to the manual customer service channel, directly converting the target user's AI customer service into manual customer service, further improving the target user's consulting experience.

[0072] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solution and inventive concept of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. A customer service management system applied to an Internet of Things platform, characterized in that: include: An intelligent processing module is configured to detect and process signals, and based on the processed signals, identify a target user for establishing a new conversation, obtain user profile information of the target user, extract basic identity information and account information from the user profile information, perform similarity processing on the target user's basic identity information and account information in combination with historical topic conversations to obtain a comprehensive similarity value, determine designated analysis information based on the comprehensive similarity value, extract key information from the designated analysis information, and determine the purpose of establishing a conversation with the target user; The front-end dialogue module is used to obtain the target user's dialogue establishment purpose. Based on the target user's real-time question information, the module combines the real-time question information with the dialogue establishment purpose to obtain a reply text. The AI customer service obtains the reply text and sends it to the target user's dialog box, thereby establishing a front-end dialogue. The process monitoring module is used to automatically detect the conversation process of the front-end conversation, and at the same time divide the conversation content in the front-end conversation into the user conversation end and the intelligent conversation end. Then, the semantic repetition of the content in the user conversation end and the intelligent conversation end is analyzed respectively to obtain the user repetition value and the intelligent repetition value. The user repetition value and the intelligent repetition value are then comprehensively processed to obtain the real-time conversation repetition value of the front-end conversation. Then, based on the conversation repetition value, a manual service signal is generated; The signal detection module is used to detect manual service signals, connect the target user with the manual customer service channel, and directly convert the target user's AI customer service into manual customer service.

2. A customer service management system applied to an Internet of Things platform according to claim 1, characterized in that: Methods for determining the purpose of establishing a dialogue include: S1: Identify the user with whom the new conversation is established and mark them as the target user. Search the database for the target user's data information and mark the search results as user profile information. The user profile information includes basic identity information, account information, and historical consultation and service information. Furthermore, basic identity information refers to the target user's age, gender, and location; account information refers to browsing history and the installed devices corresponding to the account; and historical consultation and service information refers to the target user's historical consultation and maintenance records and the operating data of the installed devices. Obtain the target user's operating data within a fixed period of time and mark it as real-time operating data. Based on the real-time operating data, determine whether the target user's installed equipment is in an abnormal state. If so, set the target user's conversation establishment purpose to equipment failure maintenance. Otherwise, generate a deep analysis signal if there is no abnormal state. S2: When a deep analysis signal is detected, obtain the target user's basic identity information and account information, set the account information as the first tag, and set the basic identity information as the second tag; Obtain historical topic conversations and topic tags in each topic conversation, perform similarity processing on the topic tag and the first tag to obtain a first similarity value XDi, and then perform similarity processing on the topic tag and the second tag to obtain a second similarity value XEi, where i represents different topic conversations. Furthermore, i∈[1,I], where I represents the total number of existing topic conversations; The comprehensive similarity value XZi of topic conversation i is obtained using the formula XZi=XDi×a1+XEi×a2, where a1 and a2 are the weight coefficients of the first label and the second label respectively; S3: After all historical topic conversations are processed, all topic comprehensive similarity values XZi are obtained, and the topic comprehensive similarity values XZi are compared to obtain the maximum value of the topic comprehensive similarity values XZi. Then, the topic conversation corresponding to the maximum value is identified and marked as the designated analysis information; The key information in the specified analysis information is obtained, and the extracted key information is set as the purpose of establishing a dialogue with the target user.

3. A customer service management system applied to an Internet of Things platform according to claim 2, characterized in that: Methods for determining abnormal conditions include: Identify the operating parameters of the installed equipment and obtain the normal operating range of the operating parameters, compare the real-time operating data with the normal operating range, if the real-time operating data all belong to the normal operating range, it means that there is no abnormal state of the installed equipment, on the contrary, if there is data in the real-time operating data that does not belong to the normal operating range, the operating state of the installed equipment is marked as abnormal state.

4. A customer service management system applied to an Internet of Things platform according to claim 2, characterized in that: If the target user's account does not have an installed device, a deep analysis signal is directly generated.

5. The customer service management system applied to the Internet of Things platform according to claim 2, characterized in that: Methods for determining topic conversations and topic tags include: Obtain the system's historical conversation information and mark each complete conversation as a topic conversation. Then identify the user corresponding to each topic conversation, obtain the user's basic identity information, and set the basic identity information as the conversation tag. Each topic conversation is obtained again, and the TF-IDF algorithm is used to extract the key information in the topic conversation to obtain the central topic words. The central topic words are integrated with the conversation tags of the corresponding topic conversation, and the integrated information is marked as the topic tag of the corresponding topic conversation.

6. A customer service management system applied to an Internet of Things platform according to claim 1, characterized in that: The methods for establishing front-end dialogue include: Obtaining the target user's conversation establishment purpose, setting the conversation establishment purpose as the central word, then obtaining the target user's real-time question information, and extracting key information from the real-time question information, wherein the key information refers to the information formed by extracting content words from the real-time question information and then reorganizing the extracted content words according to corresponding positions, and content words refer to words with real lexical meaning; Combine the central vocabulary with the key information, mark the combined information as real-time retrieval information, and search the real-time retrieval information in the data information database. Use the obtained information as the reply text of the real-time question information, and the AI customer service will send the reply text to the target user's dialog box, so that the AI customer service and the target user can establish a front-end dialogue.

7. The customer service management system applied to the Internet of Things platform according to claim 1, characterized in that: The method for determining the conversation repetition value includes: SS1: Acquire real-time front-end conversations and divide them into user conversation end and intelligent conversation end. The user conversation end refers to the information sent by the user in the front-end conversation, and the intelligent conversation end refers to the information content of the AI customer service response. Extracting the information content of the user's conversation end, identifying meaningless words in the information content of the user's conversation end, then deleting the identified meaningless words from the corresponding information content, and marking the remaining information content as user key information, wherein meaningless words refer to function words and modal particles; Using natural language processing technology, the user's key information is transmitted as input to the language model, and the semantic repetition of the user's key information is analyzed to obtain the user repetition value FY; Then, the information content of the intelligent dialogue terminal is obtained and directly used as input information. The information content is transmitted to the language model, and the semantic repetition of the AI customer service is analyzed to obtain the intelligent repetition value FZ; SS2: Use the formula FY×b1+FZ×b2=FH to obtain the real-time conversation repetition value FH of the front-end conversation, where b1 and b2 are the user weight coefficient and customer service weight coefficient respectively.

8. A customer service management system applied to an Internet of Things platform according to claim 7, characterized in that: Methods for determining manual service signals include: The conversation repetition value FH is compared with the comprehensive repetition threshold. If the conversation repetition value FH is less than the comprehensive repetition threshold, it means that the AI customer service and the target user are communicating effectively. At this time, an intelligent communication signal is generated, and the AI customer service is continued to be set as the docking customer service of the target user. On the contrary, if the conversation repetition value FH is greater than or equal to the comprehensive repetition threshold, it means that there is an abnormality in the communication process between the AI customer service and the target user. At this time, a manual service signal will be generated, and the process monitoring module will transmit the manual service signal to the signal detection module.

9. The customer service management system applied to the Internet of Things platform according to claim 1, characterized in that: It also includes a dialogue detection module for identifying dialogue information in multiple IoT platforms. When the dialogue detection module detects that a new dialogue has been established, the dialogue detection module generates a processing signal and transmits it to the intelligent processing module.

Images