Intelligent customer service method, device and system

Abstract

The present invention discloses a method, device and system for intelligent customer service. The method comprises: step S1: a service center accesses a first message; step S2: the service center cleans the acquired first message; step S3: performing semantic analysis; step S4: when the first message is an after-sales message, the service center extracts the device identification and operation intention; step S5: execution feedback. The present invention can uniformly access multimodal first messages, use a large cloud AI model to realize automatic parsing of multi-language mixed instructions and recognition of device operation intentions, combine with automated API workflows to send to a global device management platform to perform remote operations, and generate intelligent replies based on execution feedback, solving the problems of insufficient cross-language parsing capabilities, low device operation efficiency and low semantic recognition accuracy of traditional systems, and has the advantages of improving the parsing capabilities of multi-language mixed instructions, realizing efficient integration of customer service and device management platforms, and optimizing semantic parsing accuracy and automated closed-loop feedback mechanisms.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent customer service robots, and in particular to a method, device and system for intelligent customer service. Background Art

[0002] There are several significant technical flaws in the existing intelligent customer service system.

[0003] First, existing intelligent customer service robots lack the ability to parse mixed multilingual commands, typically relying on a single language or proprietary database for semantic processing. This limitation often forces cross-lingual users to rely on manual translation when requesting service, which not only increases communication costs but also easily leads to semantic deviations due to translation errors, impacting service quality and user experience. Second, existing customer service systems lack effective integration with device management platforms. Existing customer service systems only support text or voice interaction and are unable to directly translate user commands into remote device operations, such as fault diagnosis or firmware upgrades. This fragmented design requires manual mediation of user commands, which is not only inefficient but also prone to errors due to human error. Furthermore, existing semantic parsing models lack accuracy, typically relying on static knowledge bases or single-modal input (such as plain text) for parsing. They fail to dynamically optimize by integrating historical device data with multimodal information (such as text and images), resulting in low recognition accuracy for complex commands. Finally, existing systems lack an automated closed-loop feedback mechanism, requiring manual confirmation of execution results and lacking real-time feedback and execution logging. This not only reduces the user experience but also hinders the continuous improvement of the system's intelligence level.

[0004] In summary, when users initiate multilingual service requests through fragmented communication platforms, the traditional model also faces two major pain points. First, the decentralized nature of cross-platform customer service systems requires human customer service representatives to switch between multiple software programs to respond. This operation not only increases response time but also easily leads to incomplete information transmission due to communication barriers. Second, the interpretation of multilingual requests relies on manual translation, which not only increases communication costs but also easily leads to semantic deviations due to translation errors, further affecting service quality and user experience.

[0005] In view of the above problems, the existing technology is in urgent need of improvement. Summary of the Invention

[0006] In order to solve at least one of the problems existing in the prior art described above, a first aspect of the present invention provides an intelligent customer service method based on AI semantic understanding, which includes the following steps:

[0007] Step S1: The service center receives a first message; wherein the first message is sent by a user through a communication software, and the first message includes at least one of: an inquiry message or an after-sales message, and the communication software sends the message to the service center;

[0008] Step S2: the service center performs data cleaning on the acquired first message;

[0009] Step S3: Perform semantic analysis to identify mixed language instructions and extract device identification and operation intention;

[0010] Step S4: When the first message is an after-sales message, the service center extracts the device identification and operation intention based on the cloud AI big model and sends the information to the global device management platform. The service center then generates and sends device operation instructions to the global device management platform, thereby directly performing after-sales / repair operations on the device to be after-sold.

[0011] Step S5: Execution feedback; wherein the global device management platform executes the device operation instruction and returns execution feedback information of the device to the service center.

[0012] In the intelligent customer service method as described above, optionally, in step S1, the following steps are also included: the communication software determines whether the received message is the user's first message. If so, the message is sent to the service center; otherwise, the execution is terminated.

[0013] In the aforementioned intelligent customer service method, optionally, the cloud-based AI large model is a multimodal semantic parsing model, and the first message includes at least one modal data of text, voice, or image;

[0014] In step S2, data from different channels are cleaned separately, and the corpus of user messages for each channel is transmitted to the cloud-based AI model for training, so as to understand the communication habits, speaking tone, and style of users in different channels, analyze the device identification and operation intention more accurately, and provide the corpus for large model training according to the different speaking habits of users.

[0015] In the intelligent customer service method as described above, optionally, the method further includes step S6: customer service response; wherein the service center intelligently generates a reply message based on the execution feedback information, and sends the reply message to the user through the communication software interface.

[0016] In the intelligent customer service method as described above, optionally, the service center provides an external API for receiving messages from various channels, and the API of the global device management platform is called by the service center to operate the devices in the hands of channel users.

[0017] In the intelligent customer service method as described above, optionally, in step S2, the semantic analysis includes the following steps: determining the language, performing semantic conversion of a preset language, and then analyzing the device identifier and the user's current intention to perform operations on the device based on the semantics of the preset language.

[0018] In the intelligent customer service method as described above, optionally, the method also includes the service center determining whether it is necessary to call the global device management platform for operation: when the service center receives the device identification and operation intention extracted by the cloud-based AI big model, it is determined that the first message is an after-sales message and it is necessary to call the global device management platform for operation; conversely, when the cloud-based AI big model does not extract the device identification and operation intention, it is determined that the first message is consulting information and there is no need to call the global device management platform to operate the device.

[0019] In the intelligent customer service method as described above, optionally, in step S6, the service center automatically generates a corresponding reply message to the communication software based on the execution feedback information of the global device management platform; wherein the reply message is determined by the operation intention.

[0020] To achieve the above objectives, a second aspect of the present invention provides an AI model training method, wherein the method for implementing the intelligent customer service as described in any one of the first aspects includes:

[0021] Step T1: Construct a multilingual training corpus, which includes at least one of the following: user conversations from various channels, mixed-language instruction texts, device operation logs, and corresponding intent labels;

[0022] Step T2: Initialize the AI ​​parsing model and jointly train the language translation task and the device operation intention recognition task through multi-task learning. The language translation task is used to align the multilingual semantic space, and the intention recognition task is used to extract device identifiers and operation instructions.

[0023] Step T3: Introduce the device operation knowledge base, embed the device's historical operation and maintenance data as prior knowledge into the model training, and dynamically weight the device-related features through the attention mechanism; the historical operation and maintenance data includes at least one of the following: fault type, after-sales operation record, and firmware version information;

[0024] Step T4: Regularly optimize the model based on reinforcement learning to make the analysis results increasingly accurate;

[0025] Step T5: Deploy an incremental learning mechanism to regularly input new multilingual command data and device operation logs into the model for fine-tuning to adapt to semantic changes in global scenarios.

[0026] To achieve the above-mentioned object, a third aspect of the present invention provides an intelligent customer service device, wherein the intelligent customer service method described in any one of the first aspects includes:

[0027] The communication interface module is configured to receive a first message from the communication software, wherein the first message includes at least one of: an inquiry message or an after-sales message;

[0028] a data preprocessing module, connected to the communication interface module, and removing dirty data from the received first message;

[0029] An AI semantic parsing module deploys a large cloud AI model and a multi-language processing unit, connects to the data pre-processing module via a high-speed network interface, performs semantic parsing on the cleaned first message, automatically identifies mixed-language instructions, and extracts device identification and operation intent;

[0030] An instruction conversion control module, connected to the AI ​​semantic parsing module, includes a protocol adapter and an API workflow controller. Based on the device identification and operation intention extracted by the cloud-based AI large model, the module sends the device operation instruction to the global device management platform through an automated API workflow.

[0031] a device operation execution module, communicatively connected to the instruction conversion control module, comprising a distributed control node of the global device management platform and an execution feedback sensor, wherein the distributed control node remotely operates the device according to the device operation instruction, and the execution feedback sensor collects device status data in real time; wherein the global device management platform executes the device operation instruction and returns execution feedback information to the response generation module;

[0032] The response generation module is connected to the device operation execution module and the communication interface module respectively, and includes an intelligent reply generation chip and a communication interface driving unit. The intelligent reply generation chip intelligently generates a multilingual reply message based on the execution feedback information, and returns it to the communication software of the user terminal through the communication interface module.

[0033] In order to achieve the above-mentioned purpose, the fourth aspect of the present invention provides a system for intelligent customer service, wherein the method of using intelligent customer service as described in any one of the above-mentioned first aspects includes: a user terminal equipped with communication software, a service center, a cloud-based AI big model service, and a global device management platform; wherein the service center is connected to the user terminal, the cloud-based AI big model service and the global device management platform in a two-way communication manner.

[0034] In order to achieve the above-mentioned purpose, the fifth aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions or a computer program, and when the computer-executable instructions or computer program are processed and executed, the method of intelligent customer service as described in any one of the above-mentioned first aspects is implemented.

[0035] The present invention provides a method, device and system for intelligent customer service that can uniformly access multimodal first messages, use a large cloud-based AI model to automatically parse multi-language mixed instructions and identify device operation intentions, establish a dual association mechanism between operation intentions and execution feedback, combine an automated API workflow to send to a global device management platform to directly perform remote operations on the device, and generate intelligent replies based on the execution feedback on the device's operation completion status. This solves the problems of insufficient cross-language parsing capabilities, low device operation efficiency, and low semantic recognition accuracy in traditional systems, and has the advantages of improving the ability to parse multi-language mixed instructions, achieving efficient integration of customer service and device management platforms, and optimizing semantic parsing accuracy and an automated closed-loop feedback mechanism. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments or descriptions of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0037] Figure 1 This is a flow chart of an embodiment of an intelligent customer service method of the present invention;

[0038] Figure 2 yes Figure 1 Detailed flow chart of the method of intelligent customer service;

[0039] Figure 3 This is a message flow diagram of an embodiment of an intelligent customer service and device operation integration system of the present invention. DETAILED DESCRIPTION

[0040] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0041] Terms such as “comprise” and “include” indicate that in addition to the components directly and explicitly stated in the description and claims, the technical solution of the present invention does not exclude the situation where it has other components that are not directly or explicitly stated.

[0042] like Figures 1 to 2 As shown, the method of intelligent customer service based on AI semantic understanding of the present invention may include the following steps:

[0043] Step S1: The service center receives the first message. The service center here is a global or multi-device after-sales service intelligent platform, a central hub platform that centrally processes user service requests and device operation instructions. Specifically, it can be implemented using a microservices architecture and containerized deployment technology to achieve unified access and management of cross-regional device services.

[0044] In step S1, the first message sent by the user via the communication software is a multi-channel message, specifically categorized as either an inquiry message or an after-sales message. The communication software sends the message to the service center via a standardized interface. A standardized interface refers to a standardized data exchange channel that adheres to an open communication protocol. Specifically, it can be implemented using the RESTful API or WebSocket protocol to ensure data transmission compatibility between the communication software and the service center. For example, the service center receives these messages via a RESTful API and stores them in a message queue. The service center provides external APIs for receiving messages from various channels. Furthermore, the global device management platform provides APIs for the service center to call, enabling remote control of user devices. These APIs may include functions such as device status query, remote control, and firmware updates. After receiving the action intent analyzed by AI, the service center calls the corresponding device management API. For example, if the action intent is to "reboot the device," the service center calls the device management platform's "reboot" API, passing in the corresponding device identifier and authentication information.

[0045] However, in this process, messages from non-user sources may be mistakenly connected to the service center, resulting in resource waste, data processing redundancy and potential system interference problems.

[0046] In this regard, step S1 may further include the following steps: the communication software determines whether the received message is the user's first message, and if so, sends the message to the service center; otherwise, terminates the execution.

[0047] In this embodiment, the communication software can be a customer service system within a website or app. Information received through these websites or app customer service systems is collectively referred to as user channel information. Another type of communication software can be third-party communication software. This application provides this type of software with a separate account, serving as the unified input and output for communicating with the user. Information received from this account is considered user channel information, as this account is used solely for communication with the user.

[0048] As a result, redundant calculations for data cleaning and semantic analysis are reduced, and the overall processing efficiency of the system is improved. Furthermore, the solution prevents malicious or irrelevant messages from interfering with the device operation process, ensures that the service center only handles legitimate user needs, and enhances the security and reliability of the system. Specifically, by screening messages at the early stage of message reception, the waste of system resources is greatly reduced, and potential system interference problems are reduced. For example, system error messages, test messages or other non-user requests are effectively intercepted and will not enter the processing flow of the service center. This not only optimizes the allocation of system resources, but also improves the response speed of the service center to real user requests. Therefore, the technical solution of the present application not only improves system efficiency but also enhances the quality of user service.

[0049] Step S2: The service center performs data cleaning on the first message. Data cleaning refers to the process of filtering out noise and converting the format of the original message. Specifically, it can be achieved by using regular expression matching and outlier detection algorithms to eliminate the interference of unstructured data on semantic parsing.

[0050] Step S3: Perform semantic analysis based on the cloud-based AI big model, automatically identify mixed language instructions, and extract the device identification and operation intention to be sold.

[0051] After removing dirty data from the received first message, the service center uses an API to call a cloud-based AI model for semantic parsing, automatically identifying mixed-language commands and extracting device identification and operational intent. Calling the cloud-based AI model with an API refers to remotely accessing artificial intelligence computing resources through an application programming interface (API). This can be achieved using HTTP requests and the OAuth 2.0 authentication mechanism, invoking a multimodal semantic parsing model deployed on a cloud server. The cloud-based AI model in this embodiment is a multimodal semantic parsing model, and the first message can include data in multiple modalities, such as text, voice, or images.

[0052] Since user messages from different channels may contain at least one modal data including text, voice or images, and different channels have different communication habits, speaking tone and style, if differentiated processing is not performed based on multimodal data and channel characteristics, the semantic parsing model will not be able to accurately analyze device identification and operation intentions, which will affect the accuracy of subsequent command conversion.

[0053] Furthermore, in step S3, the data from different channels are cleaned separately, and the corpus of different user messages from each channel is transmitted to the cloud-based AI large model for training, so as to understand the communication habits, speaking tones, and styles of different channels, and to more accurately analyze the device identification and operation intentions. For example, for the same problem of poor network, the user's feedback may be "device number + poor network" or "device number + slow usage" or directly "device number + switch operator". Through this embodiment, the corpus can be provided to the large model for training according to the different speaking habits of users. Make the analysis of each channel more and more accurate. The device number is the unique identifier of the global device management platform for the operation to be performed on the device. The user's intention to operate the device may include, for example, restarting the device, switching networks, querying the WIFI name and password, etc.

[0054] Specifically, the multimodal semantic parsing model achieves cross-modal feature extraction by integrating a convolutional neural network with a sequence modeling module. For example, voice data can be converted to mel-spectrograms before being fed into the speech recognition sub-model, while image data can be processed using an object detection algorithm to extract device interface elements. Data from different channels is cleaned using separate data pipelines. For example, messages from social media channels use regular expressions to filter out emojis, while messages from email channels use keyword matching to remove signature information. A dynamic learning rate strategy is employed for training corpora for each channel. For example, a higher learning rate is set for social media channel corpora to accommodate the rapid iteration of spoken language, while a lower learning rate is set for email channel corpora to maintain the stability of written descriptions. Differentiated training of user speaking habits is achieved by constructing channel-specific word vectors. For example, for user groups that habitually use abbreviations, a "device ID → device identifier" mapping rule is established in the social media channel corpus.

[0055] During the analysis process, the large model trains on the received corpus, determines the language, performs semantic conversion to the preset language, and then uses the semantics of the preset language to analyze the device number and the user's current intended operation on the device. For example, customized training will be used for channel messages in different national languages ​​to align with the channel's communication language and communication habits. For example, if the channel uses English as the communication language, the AI ​​semantic parsing model will be set as an English expert. Combined with customized corpus training, it will analyze the channel issue and extract the device number and the channel's intended operation on the device. This step is automatic recognition, identifying channels in different languages ​​and taking different branches.

[0056] Semantic parsing may include the following steps: determining the language, performing semantic conversion of a preset language, and then analyzing the device identifier and the user's current intention to perform operations on the device based on the semantics of the preset language.

[0057] Among them, in the language judgment step, a natural language processing model is used to identify the grammatical structure and lexical features of the original message to determine whether it is a preset language or a mixed language containing non-preset language characters. Assuming that in the home country, the preset language is Chinese, the Chinese semantic conversion step can adopt a cross-language conversion model based on a neural network to map non-Chinese words or sentence patterns to the corresponding Chinese semantic framework. For example, the English device model "Model-X" is converted into the Chinese standard model "Model X", or the Japanese grammatical structure "操作を実行" is converted into the Chinese verb-object structure "Execute the operation". In the device identifier and operation intention analysis step, based on the converted Chinese semantics, a named entity recognition model is used to extract the device identifier, and the operation verb and target parameters are parsed in combination with the context dependency relationship. For example, the device ID "Device123" and the operation parameter "Temperature setting 25°C" are extracted from "Please adjust the temperature of Device123 to 25°C".

[0058] Specifically, when a mixed-language instruction is received, first, the language components in the message are identified by the language judgment module. If non-Chinese content is detected, the semantic conversion process is triggered. For example, the instruction "Help me check the status of device A" containing Chinese and English is converted into the pure Chinese instruction "Help me check the status of device A". The converted Chinese text is input into the semantic analysis model, and through matching with Chinese grammar rules and the device database, the device identifier "device A" and the operation intention "status query" are accurately extracted. In this process, Chinese serves as the unified analysis benchmark, avoiding parsing errors caused by grammar differences when multiple languages are mixed. For example, after the German verb-postposed structure "Gerät B einschalten" is converted into Chinese "Turn on device B", the adjustment of the verb position makes the intention recognition more accurate. Through the dual processing of language standardization and semantic mapping, it is ensured that instructions expressed in different languages are parsed under a unified logical framework, improving the accuracy of device operation instruction conversion.

[0059] In an optional embodiment, the present application further proposes that the service center determines whether it is necessary to call the global device management platform for operation: in the semantic analysis stage, when the service center receives the device identifier and operation intention extracted by the cloud AI large model, it determines that the first message is an after-sales message, and triggering the device operation requires calling the global device management platform for operation; otherwise, when the cloud AI large model fails to extract the device number and operation intention, it determines that the first message is a consultation message, and there is no need to call the global device management platform to operate the device.

[0060] Among them, the core of the judgment logic lies in the dual-element extraction status of device identification and operation intention. The device identification may include structured data such as the device serial number and product model, and the operation intention can be parsed into action types such as firmware upgrade instructions and remote restart commands. For example, when the message contains "Device SN12345 needs to be restarted", the AI ​​model will extract the device identification SN12345 and the operation intention "restart" to trigger the device operation process; if the message only contains "How to set the device parameters", then because the device identification and operation intention are not extracted, only a customer service reply is generated. The judgment process is directly related to the semantic parsing results. Specifically, the parsing results are transmitted to the service center through the API interface, and the classification decision is executed by the built-in logical control unit in the middle platform.

[0061] Specifically, during the message processing process, after receiving the analysis results from the cloud-based AI model, the service center first verifies the integrity of the device ID and the action intent. If both are present, an action instruction containing the device ID is sent to the device management platform, for example, generating a JSON-formatted instruction package {"device_id":"SN12345", "action":"restart"}. If either element is missing, the instruction conversion step is skipped and the customer service response module is directly called to generate a text reply. During execution, the device management platform's API call frequency is dynamically matched to the message type. For example, the trigger rate for after-sales messages is 100%, while the trigger rate for consultation messages is 0%, thereby reducing invalid interface calls by over 30%. This mechanism synergizes with the multilingual parsing module: when the cleaned multimodal data is input into the AI ​​model, the device ID can be extracted through QR code recognition in the image or serial number announcement in the voice. The action intent may be analyzed by combining the text instruction and the device status in the image. For example, if a user sends a message containing a photo of a device failure and the text "unable to power on," the model extracts the device ID through image recognition and, based on text analysis, determines the action intent as "fault diagnosis," triggering a call to the device management platform.

[0062] For messages identified as consultations, the service center does not trigger device operation procedures. Instead, it forwards the message to a dedicated consultation processing module. This module can include functions such as knowledge base query and intelligent question-and-answer functions to generate responses to user inquiries.

[0063] Furthermore, to ensure system security, the service center performs authorization verification before each call to the global device management platform. This verification includes the legitimacy of the message source, the compliance of the operation instructions, and the user's permission to operate the target device. Only after all verification items are completed is the actual device operation allowed. The service center also implements an operation logging function. Every device operation request, regardless of success or failure, is recorded in detail, including timestamp, device ID, operation type, and execution result. These logs are used for subsequent audits, troubleshooting, and system optimization.

[0064] Step S4: instruction conversion.

[0065] In step S4, if the first message is a post-sales message, the service center sends the device identification and operation intent extracted by the cloud-based AI model to the global device management platform through an automated API workflow, sending device operation instructions to the global device management platform. As previously mentioned, if the cloud-based AI model fails to extract the device identification and operation intent, the first message is determined to be a consultation message. There is no need to call the global device management platform to operate the device, and the user can simply reply according to the preset reply template.

[0066] Among them, the automated API workflow refers to the preset device operation instruction generation and transmission logic chain, which can be implemented using Apache Airflow or AWS Step Functions workflow engines to ensure the real-time and accuracy of instruction triggering on the device management platform.

[0067] In this embodiment, the service center can convert the structured data output by the AI ​​model (including fields such as device ID and action type) into JSON format. Using a predefined API mapping table, the JSON data is converted into a command format recognizable by the device management platform. For example, "Check the temperature sensor of device A" is converted into {"device_id": "A", "action": "check_sensor", "sensor_type": "temperature"}.

[0068] Step S5: Execute feedback.

[0069] In step S5, the global device management platform executes the device operation command and returns execution feedback information to the service center. This execution feedback information is a collection of status data on the device operation results. Specifically, it can be implemented by encapsulating the device response code and operation log in JSON format, providing structured input for intelligent response generation. This has been described in detail above and will not be repeated here.

[0070] Step S6: Customer service responds.

[0071] In step S6, the service center intelligently generates a reply message based on the execution feedback information and sends it to the user via the communication software interface. Intelligently generating a reply message refers to automatically constructing a user response based on the execution results. This can be achieved by using a template engine and natural language generation technology, combining the device's operating status to generate multilingual feedback information.

[0072] In an optional embodiment, the service center automatically generates a corresponding reply message to the communication software based on the execution feedback information of the global device management platform; wherein the reply message is determined by the operation intention.

[0073] The reply message generation process may include a dynamic template matching mechanism based on the operation intention.

[0074] Optionally, the specific implementation method may include establishing a mapping relationship library between operation intention types and standard response templates. For example, when the operation intention is to restart the device, it is mapped to a template containing execution status confirmation and estimated completion time. The parsing unit of the execution feedback information is configured to identify the status code or log data returned by the device management platform, such as combining the "instruction received" status code with the operation intention, triggering the template filling engine to generate a complete response. A correlation verification module between the operation intention and the execution feedback is deployed to detect whether the feedback information has a logical conflict with the original intention. If a conflict is detected, the exception handling process is triggered.

[0075] After a device operation command completes, the global device management platform returns execution feedback information, including a status code, to the service center. The service center uses a parsing module to extract key fields from the status code, such as the command reception status and the execution progress code. Simultaneously, the operation intent database is accessed to obtain the original operation intent type corresponding to the current session. The response generation engine inputs the operation intent and execution feedback information into a multi-dimensional matching algorithm. For example, if the operation intent is a firmware upgrade and the feedback contains "Upgrade package verification failed," the response template "Upgrade file anomaly detected. Please re-upload the installation package" is automatically selected. Dynamic parameters in the response template are populated based on the specific values ​​returned by the device management platform. For example, in a device reboot scenario, the response "Expected completion in 2 minutes and 30 seconds" is generated based on the average reboot time in historical data for the device model. This generates a response message that not only accurately reflects the actual execution status of the device operation but also, by integrating the semantics of the original operation intent, creates a natural language expression that aligns with user cognition, enabling a closed-loop response without manual intervention.

[0076] This embodiment establishes a dual association mechanism between operational intent and execution feedback, allowing responses to accurately reflect the device's operational results while dynamically adapting to the context of the user's original request. This approach effectively bridges the semantic gap between device operational results and user understanding, improving the reliability of the automated closed-loop. Furthermore, because responses are more tailored to actual user needs, the need for manual secondary verification is reduced, thereby enhancing the efficiency and accuracy of the intelligent customer service system's responses.

[0077] To achieve the above objectives, the present invention further provides an AI model training method to implement the intelligent customer service method as described in any of the aforementioned embodiments, comprising:

[0078] Step T1: Construct a multilingual training corpus, which includes at least one of the following: user conversations from various channels, mixed-language instruction texts, device operation logs, and corresponding intent labels;

[0079] Step T2: Initialize the AI ​​parsing model and jointly train the language translation task and the device operation intention recognition task through multi-task learning. The language translation task is used to align the multilingual semantic space, and the intention recognition task is used to extract device identifiers and operation instructions.

[0080] Step T3: Introduce the device operation knowledge base, embed the device's historical operation and maintenance data as prior knowledge into the model training, and dynamically weight the device-related features through the attention mechanism; the historical operation and maintenance data includes at least one of the following: fault type, after-sales operation record, and firmware version information;

[0081] Step T4: Regularly optimize the model based on reinforcement learning to make the analysis results increasingly accurate;

[0082] Step T5: Deploy an incremental learning mechanism to regularly input new multilingual command data and device operation logs into the model for fine-tuning to adapt to semantic changes in global scenarios.

[0083] In order to achieve the above-mentioned objectives, the present invention also provides an intelligent customer service device, wherein the intelligent customer service method as described in any of the aforementioned embodiments includes: a communication interface module, a data preprocessing module, an AI semantic parsing module, an instruction conversion control module, a device operation execution module and a response generation module.

[0084] Specifically, the communication interface module is used to receive a first message from the communication software, and the first message may include an advisory message or an after-sales message; the data preprocessing module is connected to the communication interface module to remove dirty data from the received first message; the AI ​​semantic parsing module deploys a cloud-based AI large model and a multi-language processing unit, and is connected to the data preprocessing module through a high-speed network interface to perform semantic parsing on the cleaned first message, automatically identify mixed-language instructions, and extract device identification and operation intentions; the instruction conversion control module is connected to the AI ​​semantic parsing module, and includes a protocol adapter and an API workflow controller, which sends the device identification and operation intentions extracted by the cloud-based AI large model to the global device management platform through an automated API workflow. , sending device operation instructions to the global device management platform; the device operation execution module is communicated with the instruction conversion control module, including the distributed control node and execution feedback sensor of the global device management platform, the distributed control node remotely operates the device according to the device operation instruction, and the execution feedback sensor collects device status data in real time; wherein, the global device management platform executes the device operation instruction and returns the execution feedback information to the response generation module; the response generation module is respectively connected to the device operation execution module and the communication interface module, and may include an intelligent reply generation chip and a communication interface driver unit. The intelligent reply generation chip intelligently generates multi-language reply messages based on the execution feedback information, and returns them to the communication software of the user terminal through the communication interface module.

[0085] In order to achieve the above purpose, Figure 3 As shown, the present invention also provides an intelligent customer service system, wherein the intelligent customer service method described in any of the aforementioned embodiments is used, including: a user terminal equipped with communication software, a service center, a cloud-based AI large model service, and a global device management platform; wherein the service center is bidirectionally connected to the user terminal, the cloud-based AI large model service, and the global device management platform. The functions and specific communication transmission methods have been described in detail above and will not be further elaborated here.

[0086] In order to achieve the above-mentioned objectives, the present invention also provides a computer-readable storage medium, which stores executable instructions or programs. When the executable instructions or programs are processed and executed, the method of intelligent customer service as described in any of the previous embodiments is implemented.

[0087] The readable storage medium is, for example, a memory. The memory may be a volatile memory or a non-volatile memory, or the memory may include both volatile and non-volatile memories. The non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of RAM are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0088] If the integrated units in the above embodiments are implemented as software functional units and sold or used as independent products, they can be stored in the above-mentioned readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion 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 software product is stored in a storage medium and includes several instructions for causing one or more devices (such as personal terminals, user terminals, or network devices) to execute all or part of the steps of the methods described in various embodiments of the present invention.

[0089] The above describes in detail the preferred specific embodiments of the present invention, which only express several implementation methods of the present invention, but it cannot be understood as limiting the scope of the patent. The technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the various technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification. It should be understood that ordinary technology in this field can make many modifications and changes based on the concept of the present invention without creative work. Therefore, without departing from the concept of the present invention, all technical solutions that can be obtained by technicians in this technical field through logical analysis, reasoning or limited experiments based on the concept of the present invention on the basis of existing technology should be within the scope of protection determined by the claims.

Claims

1. A method of intelligent customer service, characterized in that: The following steps are involved: Step S1: The service center obtains a first message; wherein the first message is sent by a user through a communication software, and the first message includes at least one of: an inquiry message or an after-sales message, and the communication software sends the message to the service center; Step S2: the service center performs data cleaning on the acquired first message; Step S3: Perform semantic analysis based on the cloud AI big model, automatically identify mixed language instructions, and extract the device identification and operation intention to be sold; Step S4: The service center determines whether it is necessary to call the global device management platform for operation: when the service center receives the device identification and operation intention extracted by the cloud-based AI large model, it determines that the first message is an after-sales message and needs to call the global device management platform for operation; conversely, when the cloud-based AI large model does not extract the device identification and operation intention, it determines that the first message is an advisory message and does not need to call the global device management platform to operate the device; when the first message is an after-sales message, the service center sends the device identification and operation intention to be after-sold to the global device management platform, generates and sends device operation instructions to the global device management platform, and directly performs after-sales / repair operations on the after-sales device; Step S5: Execution feedback; wherein the global device management platform executes the device operation instruction and returns execution feedback information of the device to the service center.

2. The intelligent customer service method according to claim 1, characterized in that: In the step S1, the following steps are also included: the communication software determines whether the received message is the user's first message, and if so, sends the message to the service center; otherwise, the execution ends.

3. The intelligent customer service method according to claim 1, characterized in that: The cloud-based AI model is a multimodal semantic parsing model, and the first message includes at least one modal data of text, voice, or image; In step S2, data from different channels are cleaned separately, and the corpus of user messages for each channel is transmitted to the cloud-based AI model for training, so as to understand the communication habits, speaking tone, and style of users in different channels, analyze the device identification and operation intention more accurately, and provide the corpus for large model training according to the different speaking habits of users.

4. The intelligent customer service method according to claim 1, characterized in that: The method further includes step S6: customer service response; wherein the service center intelligently generates a reply message based on the execution feedback information, and sends the reply message to the user through a communication software interface.

5. The intelligent customer service method according to claim 1, characterized in that: In step S2, the semantic analysis includes the following steps: determining the language, performing semantic conversion of a preset language, and then analyzing the device identifier and the user's current operation intention for the device based on the semantics of the preset language.

6. The intelligent customer service method according to claim 4, characterized in that: In step S6, the service center automatically generates a corresponding reply message to the communication software based on the execution feedback information of the global device management platform; wherein the reply message is determined by the operation intention.

7. An AI model training method, characterized in that: The method for implementing the intelligent customer service according to any one of claims 1 to 6 comprises: Step T1: Construct a multilingual training corpus, which includes at least one of the following: user conversations from various channels, mixed-language instruction texts, device operation logs, and corresponding intent labels; Step T2: Initialize the AI ​​model and jointly train the language translation task and the device operation intention recognition task through multi-task learning. The language translation task is used to align the multilingual semantic space, and the intention recognition task is used to extract device identifiers and operation instructions. Step T3: Introduce the device operation knowledge base, embed the device's historical operation and maintenance data as prior knowledge into the model training, and dynamically weight the device-related features through the attention mechanism; the historical operation and maintenance data includes at least one of the following: fault type, after-sales operation record, and firmware version information; Step T4: Regularly optimize the model based on reinforcement learning to make the analysis results increasingly accurate; Step T5: Deploy an incremental learning mechanism to regularly input new multilingual command data and device operation logs into the model for fine-tuning to adapt to semantic changes in global scenarios.

8. An intelligent customer service device, characterized in that: The method for using the intelligent customer service according to any one of claims 1 to 6 comprises: The communication interface module is configured to receive a first message from the communication software, wherein the first message includes at least one of: an inquiry message or an after-sales message; a data preprocessing module, connected to the communication interface module, and removing dirty data from the received first message; An AI semantic parsing module deploys a large cloud AI model and a multi-language processing unit, connects to the data pre-processing module via a high-speed network interface, performs semantic parsing on the cleaned first message, automatically identifies mixed-language instructions, and extracts device identification and operation intent; An instruction conversion control module, connected to the AI ​​semantic parsing module, includes a protocol adapter and an API workflow controller. Based on the device identification and operation intention extracted by the cloud-based AI large model, the module sends the device operation instruction to the global device management platform through an automated API workflow. a device operation execution module, communicatively connected to the instruction conversion control module, comprising a distributed control node of the global device management platform and an execution feedback sensor, wherein the distributed control node remotely operates the device according to the device operation instruction, and the execution feedback sensor collects device status data in real time; wherein the global device management platform executes the device operation instruction and returns execution feedback information to the response generation module; The response generation module is connected to the device operation execution module and the communication interface module respectively, and includes an intelligent reply generation chip and a communication interface driving unit. The intelligent reply generation chip intelligently generates a multilingual reply message based on the execution feedback information, and returns it to the communication software of the user terminal through the communication interface module.

9. An intelligent customer service system, characterized in that: The method of using the intelligent customer service as described in any one of claims 1 to 6 includes: a user terminal equipped with communication software, a service center, a cloud-based AI big model service, and a global device management platform; wherein the service center is connected to the user terminal, the cloud-based AI big model service, and the global device management platform in a two-way communication manner.

Images