Method and device for intelligent maintenance of terminal based on large model and electronic equipment

Abstract

The application discloses a terminal intelligent maintenance method and device based on a large model and electronic equipment. Embodiments of the application preliminarily understand the conversation message initiated by the user to the target terminal, determine whether the user initiating the conversation message means that there is a maintenance requirement of the associated equipment associated with the terminal, further acquire the running states of each device, and further determine the maintenance intention of the user initiating the conversation message based on the running states and the correlation of each maintenance task calculated based on the conversation state, so as to determine the maintenance target task, thereby being capable of actively understanding the possible maintenance requirement of the terminal used by the user from the conversation message initiated by the user, and further comprehensively judging by acquiring the running states of the associated equipment, thereby not only being capable of actively maintaining the terminal of the user, but also being capable of determining the specific maintenance target task without the explicit maintenance requirement of the user, thereby greatly improving the maintenance efficiency.

Description

Technical Field

[0001] This application relates to the field of large model technology, and in particular to a terminal intelligent maintenance method, device and electronic device based on a large model. Background Technology

[0002] As enterprises increase their level of informatization, terminal devices have become an important part of enterprise operations. Their management scope has expanded from traditional hardware maintenance to the entire lifecycle management of the hardware and software ecosystem, including terminal devices, operating systems, drivers, application software, and network connections. However, traditional operation and maintenance (O&M) models have many limitations, such as lagging device status monitoring, lack of fault prediction capabilities, chaotic driver-device mapping, and security and compliance risks. Device status monitoring typically relies on user feedback, lacking real-time and automation capabilities; high-frequency faults cannot be predicted through historical data modeling, leading to frequent sudden downtime; fragmented driver version management easily causes incompatibility issues, affecting the normal use of peripherals; and risks such as untimely terminal security patch updates and unauthorized external device access are difficult to dynamically manage. Furthermore, traditional O&M relies on manual operation, which is inefficient and prone to human error, making it difficult to meet the increasingly complex O&M needs of enterprises.

[0003] To address the aforementioned issues, a terminal device maintenance solution based on artificial intelligence, big data analytics, and automation technologies is needed to overcome the shortcomings of existing technologies that rely on manual maintenance of terminal devices. Summary of the Invention

[0004] This application provides a terminal intelligent maintenance method, device, and electronic device based on a large model to solve the shortcomings of the prior art that requires manual maintenance of terminal devices.

[0005] To achieve the above technical objectives, this application proposes a terminal intelligent maintenance method based on a large model, comprising:

[0006] Obtain user session messages for the target terminal;

[0007] The first machine learning model is invoked to perform word segmentation on the session message, and the first correlation between the session message and multiple preset maintenance services is calculated based on the words generated after word segmentation, so as to generate a list of maintenance candidate tasks.

[0008] Based on the correlation between each word contained in the conversation message and the device identifier of each associated device in the preset associated device list of the target terminal, a target device candidate list is determined, wherein the target device candidate list includes devices whose device identifiers are more correlated with the words than a preset threshold.

[0009] Obtain the operating status of each device included in the target device candidate list;

[0010] Based on the operating status of each device and the first correlation, a second machine learning model is invoked to calculate the second correlation between the session message and multiple maintenance services.

[0011] The maintenance candidate tasks in the maintenance candidate task list are sorted according to the second correlation to determine the maintenance target task.

[0012] This application also provides a terminal intelligent maintenance device based on a large model, including:

[0013] The first acquisition module is used to acquire the user's session messages for the target terminal;

[0014] The maintenance candidate task list generation module is used to call the first machine learning model to perform word segmentation on the session message, and calculate the first correlation between the session message and multiple preset maintenance services based on each word generated after word segmentation, so as to generate a maintenance candidate task list.

[0015] The target device candidate list generation module is used to determine the target device candidate list based on the correlation between each word contained in the session message and the device identifier of each associated device in the preset associated device list of the target terminal, wherein the target device candidate list includes devices whose device identifier has a correlation greater than a preset threshold with the word.

[0016] The second acquisition module is used to acquire the operating status of each device included in the target device candidate list;

[0017] The calculation module is used to calculate the second correlation between the session message and multiple maintenance services based on the operating status of each device and the first correlation by calling the second machine learning model;

[0018] The maintenance target task generation module is used to sort each maintenance candidate task in the maintenance candidate task list according to the second correlation, so as to determine the maintenance target task.

[0019] This application also provides an electronic device, including:

[0020] Memory, used to store programs;

[0021] A processor is configured to run the program stored in the memory to execute a terminal intelligent maintenance method based on a large model according to embodiments of this application.

[0022] This application also provides a computer-readable storage medium storing a computer program executable by a processor, wherein the program, when executed by the processor, implements the terminal intelligent maintenance method based on a large model as provided in this application.

[0023] According to embodiments of this application, a terminal intelligent maintenance method, apparatus, and electronic device based on a large model are used to generate a maintenance candidate task list by: firstly, segmenting the conversation message into words based on the correlation between each word in the conversation message sent to the target terminal and the terminal maintenance services; secondly, calculating a first correlation between the conversation message and multiple preset maintenance services based on the words generated after segmentation; thirdly, determining a target device candidate list based on the correlation between each word in the conversation message and the device identifiers of each associated device in a preset associated device list of the target terminal; and finally, calculating a second correlation between the conversation message and multiple maintenance services based on the operating status of each device and the first correlation; and finally, sorting the maintenance candidate tasks in the maintenance candidate task list according to the second correlation to determine the maintenance target task. Therefore, the terminal intelligent maintenance method of this application embodiment can make a preliminary understanding based on the conversation message initiated by the user to the target terminal, to determine whether the user initiating the conversation message means that there is a maintenance need for associated devices related to the terminal. Then, it can further obtain the operating status of each device, and further determine the maintenance intention of the user initiating the conversation message based on the operating status and the correlation of each maintenance task calculated based on the conversation status, thereby determining the maintenance target task. Thus, the terminal intelligent maintenance method of this application embodiment can proactively understand the possible maintenance needs of the terminal used by the user from the conversation message initiated by the user, and make a further comprehensive judgment by obtaining the operating status of associated devices. Therefore, it can not only achieve proactive maintenance of the user terminal, but also determine the specific maintenance target task without the user explicitly specifying the maintenance needs, thereby greatly improving maintenance efficiency.

[0024] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0025] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the scope of this application. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0026] Figure 1 A flowchart of an embodiment of the terminal intelligent maintenance method based on a large model provided in this application;

[0027] Figure 2 A schematic diagram of the structure of an embodiment of the terminal intelligent maintenance device based on a large model provided in this application;

[0028] Figure 3 A schematic diagram of the structure of an embodiment of the electronic device provided in this application. Detailed Implementation

[0029] Exemplary embodiments of the present application will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present application are shown in the drawings, it should be understood that the present application may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this application will be thorough and complete, and will fully convey the scope of the present application to those skilled in the art.

[0030] Example 1

[0031] With the rapid development of information technology and the continuous improvement of enterprise informatization, terminal devices have become an indispensable part of daily enterprise operations. Enterprise terminal systems not only include traditional desktop and laptop computers, but also peripherals such as printers, card readers, and handwriting displays, as well as a complex hardware and software ecosystem encompassing operating systems, drivers, and application software running on these devices. At the same time, the network connectivity environment of enterprises is becoming increasingly complex, requiring terminal devices to seamlessly interface with cloud services, internal servers, and external networks. This complexity places higher demands on the operation and maintenance management of terminal systems.

[0032] In traditional desktop operations and maintenance (O&M) models, work primarily focuses on hardware-level maintenance, such as repairing physical faults in equipment. However, as enterprises become increasingly reliant on IT, the scope of terminal system O&M has expanded from simple hardware maintenance to the entire lifecycle management of the software and hardware ecosystem. This shift has exposed the limitations of traditional O&M models. Current O&M systems exhibit significant lag in device status monitoring. Issues such as device connectivity (e.g., printer offline), driver compatibility anomalies, and application crashes often require proactive user feedback to be detected. This passive response model not only prolongs problem-solving time but may also lead to a wider impact of problems. Furthermore, traditional monitoring methods lack real-time and multi-dimensional automated monitoring capabilities, making it difficult to comprehensively grasp the operational status of terminal devices.

[0033] Meanwhile, traditional operation and maintenance (O&M) models are primarily reactive, lacking proactive fault prediction capabilities. During long-term operation, common fault types for enterprise terminal devices include hard drive lifespan issues, memory leaks, and CPU overheating. However, existing O&M systems lack the ability to model and analyze historical data, making it impossible to predict potential problems using data mining and machine learning techniques. This weakness leads to frequent sudden downtime events, severely impacting business continuity.

[0034] Furthermore, enterprises often have a wide variety of device models, and different devices often require different versions of drivers to ensure proper operation. However, due to the fragmented nature of driver version management, driver version mismatches or conflicts are prone to occur during manual maintenance. For example, peripherals such as printers, card readers, or handwriting displays may not function properly if the drivers are incompatible, directly impacting user experience and work efficiency. The lack of a unified management mechanism for the mapping relationship between drivers and devices further exacerbates the complexity of version management.

[0035] As enterprises become increasingly reliant on terminal devices, terminal security issues are becoming more prominent. For example, untimely security patch updates for terminal devices may lead to malicious exploitation of system vulnerabilities, and unauthorized access to external devices may cause data leaks and other security risks. However, traditional security auditing methods are inefficient and struggle to achieve dynamic monitoring and real-time control of terminal devices. This security compliance risk not only threatens the security of enterprise data assets but may also expose enterprises to legal and regulatory risks.

[0036] In terms of operational efficiency, traditional operation and maintenance (O&M) models typically rely on manual operations, involving complex and time-consuming processes such as problem diagnosis, fault repair, and equipment configuration. This model not only demands high levels of professional skills from O&M personnel but is also prone to further complicating issues due to human error. Furthermore, as enterprises expand, the increase in the number of terminal devices further exacerbates the complexity and workload of O&M work.

[0037] Therefore, an intelligent solution is needed that can perform user terminal maintenance without relying on manual troubleshooting. For example, such as... Figure 1 As shown, Figure 1 This is a flowchart illustrating one embodiment of a terminal intelligent maintenance method based on a large model according to an embodiment of this application. The terminal intelligent maintenance method according to an embodiment of this application may include:

[0038] S101, Obtain the user's session messages for the target terminal.

[0039] In step S101, a session message input by the user targeting the target terminal can be obtained. In this embodiment, the session message input by the user can be received, for example, through an input device or input module in the terminal currently being used by the user. Specifically, in this embodiment, when the user senses a potential problem while handling daily business, they can input the session message in step S101 through a window of a dedicated application on the terminal. Alternatively, in step S101, the application or module of the terminal intelligent maintenance method according to this embodiment can detect each message input by the user during the use of the target terminal, and can use the message containing keywords or semantics targeting the target terminal as the session message in step S101.

[0040] In addition, in step S101, historical service execution information related to the target terminal can be obtained from the service server; the probability that the target terminal is currently in an abnormal operation is calculated based on the historical service execution information; when the probability is greater than a preset probability threshold, the user is prompted to enter a session message.

[0041] Furthermore, in this embodiment, the association of a new associated device with the target terminal can also be detected. That is, when a new device connection is detected, the user can be prompted to enter session information to confirm the reason for the current new associated device connection or whether the new connected device needs maintenance.

[0042] S102, the first machine learning model is called to perform word segmentation on the session message, and the first correlation between the session message and multiple preset maintenance services is calculated based on the words generated after word segmentation, so as to generate a list of maintenance candidate tasks.

[0043] In step S102, the correlation between each word in the conversation message and the terminal maintenance service can be detected based on the conversation message obtained in step S101. For example, a pre-trained first machine learning model can be called to perform word segmentation processing on the conversation message obtained in step S101. Furthermore, after word segmentation in step S102, in this embodiment, word embedding (WE) can be used to convert each word in the conversation message into word vectors, enabling further processing in subsequent steps. Specifically, in this embodiment, sentiment-specific word embedding (SSWE) can be further used to convert the text data of the conversation message obtained in step S101 into word vectors, thereby obtaining word vectors with sentiment information. This better reflects the sentiment of the words in the text data, allowing for further determination of the user's maintenance intent in the conversation message received in step S101 based on the sentiment.

[0044] In step S102, after word segmentation, the first machine learning model can be used to further calculate the correlation between the session message represented by each segment and the preset maintenance service. For example, a user can send a session message such as "signature not recognized" or "signature failed" through their terminal, or the session message can be the result message returned when the user performs an operation such as signing using an associated device connected to the target terminal, such as a handwriting tablet. When the session message is obtained in step S101, the first machine learning model can be used in step S102 to calculate the correlation between the segment and the word in the session message. For example, it can be calculated that "signature" in the session message is related to the hardware maintenance of the handwriting tablet or the software maintenance of the image text recognition function, thus it can be calculated that the session message has a high correlation with the handwriting tablet hardware maintenance task. Of course, it can also be calculated that the session message has a high correlation with the software maintenance task of the text recognition function, or it also has a certain correlation with the computing resource maintenance task of the target terminal.

[0045] In addition, in step S102, at least one historical maintenance task associated with the target terminal can be obtained; and the third correlation between the historical maintenance task and each word can be calculated; using the third correlation as the weight of each word, the first correlation between the session message and multiple preset maintenance services can be calculated to generate a list of maintenance candidate tasks.

[0046] S103, determine the target device candidate list based on the correlation between each word contained in the session message and the device identifier of each associated device in the preset associated device list of the target terminal.

[0047] In step S103, the target device candidate list can be determined simultaneously with or after step S102 based on the relevance between each word in the session message obtained in step S101 and the associated devices of the target terminal. For example, in this embodiment, the relevance can be determined based on the textual or semantic overlap between the words in the user's session message obtained in step S101 and the device name, thus identifying the target devices involved in the user's session message as the target device candidate list. This means the target device candidate list includes devices whose device identifier has a relevance greater than a preset threshold with respect to the words. For example, if the user mentions "signature" in the session message, although this word is different from the device name of the handwriting tablet, its semantic relevance to the handwriting tablet can be calculated to be high, therefore the handwriting tablet can be identified as a target device candidate.

[0048] S104, obtain the operating status of each device included in the target device candidate list.

[0049] S105, based on the operating status of each device and the first correlation, calls the second machine learning model to calculate the second correlation between the session message and multiple maintenance services.

[0050] In step S104, the operating status of each device can be obtained based on the target device candidates determined in step S103. For example, in the terminal intelligent maintenance scheme according to the embodiment of this application, the operating status can be requested from each device periodically, or each device can be set to periodically report its operating status to the server. Thus, after the target device candidate list is determined in step S103, the terminal can directly obtain the operating status of the candidate devices in the most recently reported device list from the server. Of course, in the embodiment of this application, in step S104, in response to the determination of the target device candidate list in step S103, the terminal can also request the current operating status from each device in the list. In step S105, based on the operating status of the candidate devices obtained in step S104, the correlation between the session messages and maintenance tasks calculated in step S102 can be further combined, and a second machine learning model can be used to calculate a second correlation between the session messages and maintenance services. For example, in step S105, the operating status of devices that are highly correlated with the session message can further characterize the current maintenance needs of each device. Therefore, by combining the semantic correlation between the session message and the maintenance task calculated in step S102, the device that the user may actually want to maintain in the session message can be determined. For example, when the user initiates a session message "signature cannot be recognized", in step S102 it is calculated that the session message is related to maintenance tasks such as handwriting tablet failure and abnormal operation of recognition software. In step S104, the operating status of the handwriting tablet and the terminal can be obtained, and for example, a status acquisition request can be sent to the handwriting tablet, but the feedback information is no response. In step S105, the correlation between the session message and the maintenance task can be further calculated by combining the operating status and the correlation calculated in step S102. Since the device operating status information is further introduced in step S105, a more accurate reflection of the user's maintenance intention in the session message can be obtained.

[0051] Furthermore, in this embodiment of the application, in step S105, a second machine learning model can be used to calculate the first opinion bias feature of each word in the conversation message and the second opinion bias feature of each operating device in the conversation message based on the word vectors of each word. Specifically, the first opinion bias feature and the second opinion bias feature can respectively represent the opinion bias regarding whether the target terminal is operating normally. For example, if a user expresses a complaint about a signature failure or other problems in the conversation message, the negative emotion can be identified in step S105 by calculating the opinion bias of each word, thereby further determining that the words targeted by the negative emotion are likely words representing maintenance needs. Therefore, feature fusion calculation can be performed on the first opinion bias feature and the second opinion bias feature to generate a fused feature; the fused feature is then used as an attention feature to calculate the second relevance of the conversation message to multiple maintenance services.

[0052] S106, Sort each maintenance candidate task in the maintenance candidate task list according to the second correlation to determine the maintenance target task.

[0053] In step S106, candidate tasks can be reordered based on the correlation between the session messages and maintenance tasks calculated in step S105, and the top-ranked maintenance task with the highest correlation can be identified as the target task. Alternatively, all maintenance tasks with a correlation greater than a preset threshold calculated in step S105 can be executed as target tasks. Or, all maintenance tasks with a correlation greater than a preset threshold can be displayed to the user through a display component, allowing the user to further confirm the final target task.

[0054] In this embodiment, after the target maintenance task is determined in step S106, corresponding processing can be performed to eliminate the fault mentioned in the session message. For example, in this embodiment, when the target maintenance task is a device connection task, the driver information corresponding to the target device can be obtained based on the device identifier of the target device corresponding to the target maintenance task. Then, the installed driver corresponding to the target device in the target terminal can be updated based on the driver information, and a first test task can be generated based on the updated driver and the session message. After the first test task is executed, the test result returned by the target device after executing the first test task can be obtained, and the maintenance task status of the target device can be updated based on the test result.

[0055] Furthermore, in this embodiment, historical business data and historical maintenance task data can be acquired periodically or in real time, and the first machine learning model used in step S102 and the second machine learning model used in step S105 can be trained based on the historical business data and historical maintenance task data. This training can be performed after each calculation or periodically.

[0056] According to the large-model-based intelligent terminal maintenance method of this application, the method performs word segmentation on the conversation message based on the correlation between each word contained in the conversation message sent to the target terminal and the terminal maintenance service, and calculates the first correlation between the conversation message and a number of preset maintenance services based on each word generated after word segmentation to generate a maintenance candidate task list; determines a target device candidate list based on the correlation between each word contained in the conversation message and the device identifier of each associated device in the preset associated device list of the target terminal; then, based on the operating status of each device and the first correlation, the method calls a second machine learning model to calculate the second correlation between the conversation message and the number of maintenance services; and sorts the maintenance candidate tasks in the maintenance candidate task list according to the second correlation to determine the maintenance target task. Therefore, the terminal intelligent maintenance method of this application embodiment can make a preliminary understanding based on the conversation message initiated by the user to the target terminal, to determine whether the user initiating the conversation message means that there is a maintenance need for associated devices related to the terminal. Then, it can further obtain the operating status of each device, and further determine the maintenance intention of the user initiating the conversation message based on the operating status and the correlation of each maintenance task calculated based on the conversation status, thereby determining the maintenance target task. Thus, the terminal intelligent maintenance method of this application embodiment can proactively understand the possible maintenance needs of the terminal used by the user from the conversation message initiated by the user, and make a further comprehensive judgment by obtaining the operating status of associated devices. Therefore, it can not only achieve proactive maintenance of the user terminal, but also determine the specific maintenance target task without the user explicitly specifying the maintenance needs, thereby greatly improving maintenance efficiency.

[0057] Example 2

[0058] Figure 2 This is a schematic diagram of the structure of one embodiment of the terminal intelligent maintenance device based on a large model provided in this application. Figure 2 As shown in the figure, the terminal maintenance device provided in this application embodiment may include: a first acquisition module 21, a maintenance candidate task list generation module 22, a target device candidate list generation module 23, a second acquisition module 24, a calculation module 25, and a maintenance target task generation module 26.

[0059] The first acquisition module 21 can be used to acquire user session messages for the target terminal.

[0060] The first acquisition module 21 can acquire session messages input by the user for the target terminal. In this embodiment, the first acquisition module 21 can be an input device or input module in the terminal currently used by the user, and is used to receive session messages input by the user. Specifically, in this embodiment, when the user senses a potential problem while handling daily business, they can input the session message through a window of a dedicated application in the terminal. Alternatively, the application or module of the terminal intelligent maintenance method according to this embodiment can act as the first acquisition module 21 to detect each message input by the user during the use of the target terminal, and can treat the message as a session message when it is detected that it contains keywords or semantics related to the target terminal.

[0061] In addition, the first acquisition module 21 can obtain historical service execution information related to the target terminal from the service server; calculate the probability that the target terminal is currently in an abnormal operation based on the historical service execution information; and prompt the user to enter a session message when the probability is greater than a preset probability threshold.

[0062] In addition, in this embodiment of the application, the first acquisition module 21 can also detect the association of the new associated device with the target terminal. That is, when the connection of the new device is detected, the first acquisition module 21 can prompt the user to enter session information to confirm the reason for the connection of the new associated device or whether the new connected device needs to be maintained.

[0063] The maintenance candidate task list generation module 22 can be used to call the first machine learning model to perform word segmentation on the session message, and calculate the first correlation between the session message and multiple preset maintenance services based on the words generated after word segmentation, so as to generate a maintenance candidate task list.

[0064] The maintenance candidate task list generation module 22 can detect the correlation between each word in the conversation message and the terminal maintenance business based on the conversation message obtained by the first acquisition module 21. For example, it can call a pre-trained first machine learning model to perform word segmentation processing on the conversation message obtained by the first acquisition module 21. In addition, after the maintenance candidate task list generation module 22 performs word segmentation processing, in this embodiment, word embedding (WE) can be used to convert each word contained in the conversation message into word vectors, so that it can be further processed in subsequent steps. In particular, in this embodiment, sentiment-specific word embedding (SSWE) can be further used to convert the text data of the conversation message obtained by the first acquisition module 21 into word vectors, so as to obtain word vectors with sentiment information. This can better reflect the sentiment of words in the text data, so as to further determine the maintenance intention in the user's conversation message received by the first acquisition module 21 based on the sentiment.

[0065] After performing word segmentation, the maintenance candidate task list generation module 22 can use a first machine learning model to further calculate the correlation between each segmented word representing the session message and the preset maintenance service. For example, a user can send a session message such as "signature not recognized" or "signature failed" through their terminal, or the session message can be the result message returned when the user performs an operation such as signing using an associated device connected to the target terminal, such as a handwriting tablet. When the first acquisition module 21 acquires the session message, the maintenance candidate task list generation module 22 can use the first machine learning model to calculate the correlation between the segmented words in the session message. For example, it can calculate that "signature" in the session message is related to the hardware maintenance of the handwriting tablet or the software maintenance of the image text recognition function, thus calculating that the session message has a high correlation with the handwriting tablet hardware maintenance task. Of course, it can also calculate that the session message has a high correlation with the software maintenance task of the text recognition function, or it also has a certain correlation with the target terminal's computing resource maintenance task.

[0066] In addition, the maintenance candidate task list generation module 22 can further obtain at least one historical maintenance task associated with the target terminal; and calculate the third correlation between the historical maintenance task and each word; using the third correlation as the weight of each word, calculate the first correlation between the session message and multiple preset maintenance services to generate a maintenance candidate task list.

[0067] The target device candidate list generation module 23 can be used to determine the target device candidate list based on the correlation between each word contained in the session message and the device identifier of each associated device in the preset associated device list of the target terminal.

[0068] The target device candidate list generation module 23 can determine the target device candidate list simultaneously with or after the processing of the maintenance candidate task list generation module 22, based on the relevance between each word in the session message obtained by the first acquisition module 21 and each associated device associated with the target terminal. For example, in this embodiment, the relevance can be determined by the textual or semantic overlap between the words in the user's session message obtained by the first acquisition module 21 and the device name, thereby identifying the target devices involved in the user's session message as the target device candidate list. Thus, the target device candidate list includes devices whose device identifiers have a relevance greater than a preset threshold to the words. For example, if the user mentions "signature" in the session message, although this word is different from the device name of the handwriting tablet, its semantic relevance to the handwriting tablet can be calculated to be high, so the handwriting tablet can be identified as a target device candidate.

[0069] The second acquisition module 24 can be used to acquire the operating status of each device included in the target device candidate list.

[0070] The calculation module 25 can be used to calculate the second correlation between session messages and multiple maintenance services based on the operating status of each device and the first correlation, by calling the second machine learning model.

[0071] The second acquisition module 24 can acquire the operating status of each device based on the target device candidates determined by the target device candidate list generation module 23. For example, in the terminal intelligent maintenance scheme according to the embodiment of this application, the terminal can periodically request the operating status of each device, or each device can be set to periodically report its operating status to the server. Thus, when the target device candidate list generation module 23 determines the target device candidate list, the terminal can directly obtain the operating status of the candidate devices in the most recently reported device list from the server. Of course, in the embodiment of this application, the second acquisition module 24 can also respond to the determination of the target device candidate list by the target device candidate list generation module 23, and the terminal can request the current operating status of each device in the list. The calculation module 25 can further combine the correlation between the session messages and maintenance tasks calculated by the maintenance candidate task list generation module 22 based on the operating status of the candidate devices acquired by the second acquisition module 24, and use a second machine learning model to calculate the second correlation between the session messages and maintenance services. For example, the operating status of devices that are highly correlated with the session messages can further characterize the current maintenance needs of each device. Therefore, by combining the semantic correlation between the session messages calculated by the maintenance candidate task list generation module 22 and the maintenance tasks, the device that the user may actually want to maintain in the session message can be determined. For example, when the user initiates a session message "signature cannot be recognized," the maintenance candidate task list generation module 22 calculates that the session message is related to maintenance tasks such as handwriting tablet failure and abnormal operation of recognition software. The second acquisition module 24 can further acquire the operating status of the handwriting tablet and the terminal, and for example, it can send a status acquisition request to the handwriting tablet, but the feedback information is no response. Then, the calculation module 25 can further combine the operating status and the correlation calculated by the maintenance candidate task list generation module 22 to further calculate the correlation between the session message and the maintenance tasks. Since the maintenance candidate task list generation module 22 further introduces the operating status information of the devices, it can obtain a more accurate reflection of the user's maintenance intention in the session message.

[0072] Furthermore, in this embodiment, the calculation module 25 can use a second machine learning model to calculate the first opinion bias feature of each word in the conversation message and the second opinion bias feature of each device's operating device in the conversation message based on the word vectors of each word in the conversation message. Specifically, the first opinion bias feature and the second opinion bias feature can respectively represent the opinion bias regarding whether the target terminal is operating normally. For example, a user might express a complaint about a signature failure or other problems in the conversation message. The calculation module 25 can identify this negative emotion by calculating the opinion bias of each word, thereby further determining that the words targeted by this negative emotion are likely words representing maintenance needs. Therefore, feature fusion calculation can be performed on the first opinion bias feature and the second opinion bias feature to generate a fused feature; the fused feature is then used as an attention feature to calculate the second relevance of the conversation message to multiple maintenance services.

[0073] The maintenance target task generation module 26 can be used to sort each maintenance candidate task in the maintenance candidate task list according to the second correlation in order to determine the maintenance target task.

[0074] The maintenance target task generation module 26 can reorder candidate tasks based on the correlation between the session messages and maintenance tasks calculated by the calculation module 25, and determine the top-ranked maintenance task with the highest correlation as the target task. Alternatively, it can execute all maintenance tasks with a correlation greater than a preset threshold calculated by the calculation module 25 as target tasks, or display all maintenance tasks with a correlation greater than a preset threshold to the user through a display component, allowing the user to further confirm the final target task.

[0075] In this embodiment, after the maintenance target task generation module 26 determines the target maintenance task, it can perform corresponding processing to eliminate the fault mentioned in the session message. For example, in this embodiment, when the maintenance target task is a device connection task, the driver information corresponding to the target device can be obtained based on the device identifier of the target device corresponding to the maintenance target task. Then, the installed driver corresponding to the target device in the target terminal can be updated based on the driver information, and a first test task can be generated based on the updated driver and the session message. After executing the first test task, the test result returned by the target device after executing the first test task can be obtained, and the maintenance task status of the target device can be updated based on the test result.

[0076] Furthermore, in this embodiment, historical business data and historical maintenance task data can be acquired periodically or in real time. Based on this historical business data and historical maintenance task data, the first machine learning model used by the maintenance candidate task list generation module 22 and the second machine learning model used by the calculation module 25 are trained. This training can be performed after each calculation or periodically.

[0077] According to an embodiment of this application, a terminal intelligent maintenance device based on a large model performs word segmentation on the conversation message based on the correlation between each word contained in the conversation message sent to the target terminal and the terminal maintenance service. Based on the words generated after word segmentation, a first correlation between the conversation message and multiple preset maintenance services is calculated to generate a maintenance candidate task list. A target device candidate list is determined based on the correlation between each word contained in the conversation message and the device identifier of each associated device in a preset associated device list of the target terminal. Then, based on the operating status of each device and the first correlation, a second machine learning model is called to calculate a second correlation between the conversation message and multiple maintenance services. Finally, the maintenance candidate tasks in the maintenance candidate task list are sorted according to the second correlation to determine the maintenance target task. Therefore, the terminal intelligent maintenance method of this application embodiment can make a preliminary understanding based on the conversation message initiated by the user to the target terminal, to determine whether the user initiating the conversation message means that there is a maintenance need for associated devices related to the terminal. Then, it can further obtain the operating status of each device, and further determine the maintenance intention of the user initiating the conversation message based on the operating status and the correlation of each maintenance task calculated based on the conversation status, thereby determining the maintenance target task. Thus, the terminal intelligent maintenance method of this application embodiment can proactively understand the possible maintenance needs of the terminal used by the user from the conversation message initiated by the user, and make a further comprehensive judgment by obtaining the operating status of associated devices. Therefore, it can not only achieve proactive maintenance of the user terminal, but also determine the specific maintenance target task without the user explicitly specifying the maintenance needs, thereby greatly improving maintenance efficiency.

[0078] Example 3

[0079] The above describes the internal functions and structure of the terminal intelligent maintenance device, which can be implemented as an electronic device. Figure 3 A schematic diagram illustrating the structure of an embodiment of the electronic device provided in this application. Figure 3 As shown, the electronic device includes a memory 31 and a processor 32.

[0080] Memory 31 is used to store programs. In addition to the programs described above, memory 31 can also be configured to store various other data to support operation on the electronic device. Examples of this data include instructions for any application or method used to operate on the electronic device, contact data, phonebook data, messages, pictures, videos, etc.

[0081] The memory 31 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0082] The processor 32 is not limited to a central processing unit (CPU), but may also be a graphics processing unit (GPU), a field-programmable gate array (FPGA), an embedded neural network processor (NPU), or an artificial intelligence (AI) chip. The processor 32 is coupled to the memory 31 and executes the program stored in the memory 31. When this program runs, it executes the terminal intelligent maintenance method described in Embodiment 1 above.

[0083] Furthermore, such as Figure 3 As shown, the electronic device may also include other components such as a communication component 33, a power supply component 34, an audio component 35, and a display 36. Figure 3 The diagram only shows some components and does not mean that the electronic device includes only these components. Figure 3 The components shown.

[0084] Communication component 33 is configured to facilitate wired or wireless communication between electronic devices and other devices. The electronic devices can access wireless networks based on communication standards, such as WiFi, 3G, 4G, or 5G, or combinations thereof. In one exemplary embodiment, communication component 33 receives broadcast signals or broadcast-related information from an external broadcast management system via a broadcast channel. In one exemplary embodiment, communication component 33 also includes a near-field communication (NFC) module to facilitate short-range communication. For example, the NFC module may be implemented based on radio frequency identification (RFID) technology, Infrared Data Association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology, and other technologies.

[0085] Power supply component 34 provides power to various components of the electronic device. Power supply component 34 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to the electronic device.

[0086] Audio component 35 is configured to output and / or input audio signals. For example, audio component 35 includes a microphone (MIC) configured to receive external audio signals when the electronic device is in an operating mode, such as call mode, recording mode, and voice recognition mode. The received audio signals may be further stored in memory 31 or transmitted via communication component 33. In some embodiments, audio component 35 also includes a speaker for outputting audio signals.

[0087] Display 36 includes a screen, which may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen can be implemented as a touchscreen to receive input signals from a user. The touch panel includes one or more touch sensors to sense touches, swipes, and gestures on the touch panel. The touch sensors can detect not only the boundaries of the touch or swipe action, but also the duration and pressure associated with the touch or swipe operation.

[0088] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0089] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

Claims

1. A terminal intelligent maintenance method based on a large model, characterized in that, include: Obtain user session messages for the target terminal; The first machine learning model is invoked to perform word segmentation on the session message, and the first correlation between the session message and multiple preset maintenance services is calculated based on the words generated after word segmentation, so as to generate a list of maintenance candidate tasks. Based on the correlation between each word contained in the conversation message and the device identifier of each associated device in the preset associated device list of the target terminal, a target device candidate list is determined, wherein the target device candidate list includes devices whose device identifiers are more correlated with the words than a preset threshold. Obtain the operating status of each device included in the target device candidate list; Based on the operating status of each device and the first correlation, a second machine learning model is invoked to calculate the second correlation between the session message and multiple maintenance services. The maintenance candidate tasks in the maintenance candidate task list are sorted according to the second correlation to determine the maintenance target task.

2. The terminal intelligent maintenance method based on a large model according to claim 1, characterized in that, Further includes: When the maintenance target task is a device connection task, the driver information corresponding to the target device is obtained according to the device identifier of the target device corresponding to the maintenance target task; Update the installed driver program corresponding to the target device in the target terminal based on the driver information; The first test task is generated based on the updated driver and the session messages; Obtain the test results returned by the target device after executing the first test task; Update the maintenance task status of the target device based on the test results.

3. The terminal intelligent maintenance method based on a large model according to claim 1, characterized in that, The step of calculating the first relevance between the session message and multiple preset maintenance services based on each word generated after word segmentation, in order to generate a list of maintenance candidate tasks, includes: Obtain at least one historical maintenance task associated with the target terminal; Calculate the third correlation between the historical maintenance task and each word; Using the third correlation as the weight of each word, the first correlation between the session message and multiple preset maintenance services is calculated to generate a list of maintenance candidate tasks.

4. The terminal intelligent maintenance method based on a large model according to claim 1, characterized in that, The step of calculating the second correlation between the session message and multiple maintenance services based on the operating status of each device and the first correlation includes: The second machine learning model is used to calculate the first opinion bias feature of each word in the conversation message and the second opinion bias feature of each device's operating device in the conversation message based on the word vectors of each word in the conversation message. The first opinion bias feature and the second opinion bias feature respectively represent the opinion bias regarding whether the target terminal is operating normally. The first opinion tendency feature and the second opinion tendency feature are subjected to feature fusion calculation to generate fused features; The fusion feature is used as an attention feature to calculate the second relevance of the session message to multiple maintenance services.

5. The terminal intelligent maintenance method based on a large model according to claim 1, characterized in that, The acquisition of user session messages for the target terminal includes: Obtain historical service execution information related to the target terminal from the service server; Calculate the probability that the target terminal is currently experiencing an operational anomaly based on the historical service execution information; When the probability is greater than a preset probability threshold, the user is prompted to enter the session message.

6. The terminal intelligent maintenance method based on a large model according to claim 1, characterized in that, Before obtaining the user's session messages for the target terminal, the method further includes: Detect the association of the new associated device with the target terminal; In response to the associated detection, the user is prompted to enter the session message.

7. The terminal intelligent maintenance method based on a large model according to claim 1, characterized in that, The method further includes: Regularly or in real-time acquire historical business data and historical maintenance task data; The first machine learning model and the second machine learning model are trained based on the historical business data and the historical maintenance task data.

8. A terminal intelligent maintenance device based on a large model, characterized in that, include: The first acquisition module is used to acquire the user's session messages for the target terminal; The maintenance candidate task list generation module is used to call the first machine learning model to perform word segmentation on the session message, and calculate the first correlation between the session message and multiple preset maintenance services based on each word generated after word segmentation, so as to generate a maintenance candidate task list. The target device candidate list generation module is used to determine the target device candidate list based on the correlation between each word contained in the session message and the device identifier of each associated device in the preset associated device list of the target terminal, wherein the target device candidate list includes devices whose device identifier has a correlation greater than a preset threshold with the word. The second acquisition module is used to acquire the operating status of each device included in the target device candidate list; The calculation module is used to calculate the second correlation between the session message and multiple maintenance services based on the operating status of each device and the first correlation by calling the second machine learning model; The maintenance target task generation module is used to sort each maintenance candidate task in the maintenance candidate task list according to the second correlation, so as to determine the maintenance target task.

9. An electronic device, characterized in that, include: Memory, used to store programs; A processor is configured to run the program stored in the memory to perform the terminal intelligent maintenance method based on a large model as described in any one of claims 1 to 7.

Images