Intelligent customer service implementation method and device facing DevOps platform, and medium

Abstract

The invention discloses an intelligent customer service implementation method and device for a DevOps platform and a medium, and the method comprises the steps: collecting real-time operation data of the DevOps platform, inputting the real-time operation data into a preset large model, and analyzing the real-time operation data through the large model to obtain a platform operation state corresponding to the DevOps platform; according to the historical operation record of the user and the current task parameter of the DevOps platform, filling the user question to generate a corresponding question and answer statement; inputting the question and answer statement into the large model, analyzing the running log of the DevOps platform and the question and answer statement through the large model, and outputting a plurality of candidate abnormal descriptions corresponding to the question and answer statement; performing mode matching on the running log through a preset knowledge base to output a corresponding target exception description; and under the condition that the candidate exception description is matched with the target exception description, acquiring an exception repair strategy corresponding to the candidate exception description based on the knowledge base, and pushing the exception repair strategy to the user.

Description

Technical Field

[0001] The present application relates to the field of large model technology, and specifically to a method, device and medium for implementing intelligent customer service for a DevOps platform. Background Art

[0002] DevOps is a software development method that combines development and operation and maintenance, aiming to speed up software delivery, improve software quality and reduce development costs. The DevOps platform integrates multiple links such as development, testing, deployment and operation and maintenance, and achieves efficient management of the software life cycle through automation and continuous integration.

[0003] The DevOps platform not only involves a wide range of technology stacks (such as code management, continuous integration, automated testing, deployment, monitoring, etc.), but also involves a large amount of operation logs, system status data, and process information. In the DevOps platform, the development and operation and maintenance teams need to frequently handle a large number of tasks such as automated deployment, code management, CI / CD pipelines, etc. However, due to the complexity of the work content of the DevOps platform and the need for rapid response, the traditional manual customer service system is difficult to help users quickly find solutions in complex operations, and has poor real-time and accuracy, and relatively low processing efficiency. Summary of the invention

[0004] In order to solve the above problems, this application proposes an intelligent customer service implementation method for a DevOps platform, including:

[0005] Collecting real-time operation data of the DevOps platform, inputting the real-time operation data into a preset big model, parsing the real-time operation data through the big model, and obtaining the platform operation status corresponding to the DevOps platform;

[0006] When there is an abnormality in the operation status of the platform, receiving a user question sent by a user, and filling in the user question according to the user's historical operation record and the current task parameters of the DevOps platform to generate a corresponding question and answer statement;

[0007] Input the question and answer statement into the big model, analyze the operation log of the DevOps platform and the question and answer statement through the big model, and output several candidate exception descriptions corresponding to the question and answer statement;

[0008] Performing pattern matching on the operation log through a preset knowledge base to output a corresponding target exception description;

[0009] In the case where the candidate exception description matches the target exception description, an exception repair strategy corresponding to the candidate exception description is acquired based on the knowledge base, and the exception repair strategy is pushed to the user.

[0010] In one implementation of the present application, the user question is filled in according to the historical operation record of the user and the current task parameters of the DevOps platform to generate a corresponding question and answer statement, specifically including:

[0011] Analyze the historical operation records of the user to determine the operation preference of the user; wherein the operation preference includes at least one or more of the following: fault handling strategy, operation mode, and operation field;

[0012] Determine the task type corresponding to the DevOps platform according to the current task parameters of the DevOps platform;

[0013] Associating the task type with the user's operation preference to filter out the question words of interest corresponding to the task type according to the operation preference;

[0014] The user question is filled in according to the associated words of interest to generate a corresponding question-and-answer sentence.

[0015] In one implementation of the present application, before pushing the abnormality repair strategy to the user, the method further includes:

[0016] Determine, according to the operation preference, a target path corresponding to the anomaly repair strategy; wherein the target path refers to an operation sequence consisting of a plurality of operations for implementing the anomaly repair strategy;

[0017] According to the target path, the abnormality repair strategy is pushed to the user.

[0018] In one implementation of the present application, determining the target path corresponding to the abnormality repair strategy according to the operation preference specifically includes:

[0019] Determine, according to the fault handling strategy in the operation preference, a strategy organization form corresponding to the abnormality repair strategy;

[0020] Determine whether the abnormality repair strategy exists in the fault handling strategy, and if so, update the priority corresponding to the abnormality repair strategy in the knowledge base so that the knowledge base determines the candidate order of the abnormality repair strategy according to the priority;

[0021] According to the operation mode, determining the operation habit corresponding to the user; wherein the operation habit is used to determine the display format corresponding to the abnormality repair strategy;

[0022] According to the strategy organization form, the operation habits and the candidate order, a target path corresponding to the abnormal repair strategy is determined.

[0023] In one implementation of the present application, according to the fault handling strategy in the operation preference, a strategy organization form corresponding to the abnormal repair strategy is determined, specifically including:

[0024] Determining, according to the fault handling strategy in the operation preference, the operation frequency corresponding to the candidate abnormality description by the user;

[0025] Determining, according to the operation frequency, the adaptability of the user to the abnormality repair strategy;

[0026] The fitness is compared with a preset fitness, and a policy organization form corresponding to the abnormal repair strategy is determined according to a size relationship between the fitness and the preset fitness; wherein the policy organization form includes policy details and policy prompts.

[0027] In an implementation of the present application, determining the user's adaptability to the abnormality repair strategy according to the operation frequency specifically includes:

[0028] Determine, according to the historical operation records, an average operation frequency corresponding to the abnormal repair strategy and a strategy matching coefficient of the user for the abnormal repair strategy; wherein the strategy matching coefficient is obtained according to an average repair time corresponding to the abnormal repair strategy;

[0029] A ratio between the operating frequency and the average operating frequency is calculated, and a product of the ratio and the strategy matching coefficient is used as the adaptability of the user to the abnormality repair strategy.

[0030] In one implementation of the present application, before collecting the real-time operation data of the DevOps platform, the method further includes:

[0031] Collecting development and operation data corresponding to the DevOps platform, clustering the development and operation data, and constructing a training data set according to the clustering results obtained after clustering;

[0032] The training data set is enhanced, and based on the enhanced training data set, a large model for intelligent customer service question and answering on the DevOps platform is trained.

[0033] In one implementation of the present application, data enhancement is performed on the training data set, specifically including:

[0034] Performing sentence type conversion on the training data set to convert declarative sentences in the training data set into question sentences;

[0035] For the converted training data set, reconstructing the context of the training data set to obtain the reconstructed training data set;

[0036] Obtain negative example sentences that are unrelated to the semantics in the training data set, and expand the data in the reconstructed training data set according to the negative example sentences.

[0037] The embodiment of the present application provides an intelligent customer service implementation device for a DevOps platform, the device comprising:

[0038] at least one processor;

[0039] and, a memory communicatively coupled to the at least one processor;

[0040] Among them, the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute an intelligent customer service implementation method for a DevOps platform as described in any of the above items.

[0041] The embodiment of the present application provides a non-volatile computer storage medium storing computer executable instructions, wherein the computer executable instructions are configured as follows:

[0042] A method for implementing intelligent customer service for a DevOps platform as described in any of the above items.

[0043] The intelligent customer service implementation method for the DevOps platform proposed in this application can bring the following beneficial effects:

[0044] Combining the user's historical operation records and current task parameters, the user's questions are filled in and accurate question and answer statements are generated, which improves the accuracy and completeness of the problem description. The operation logs and question and answer statements are analyzed through a large model to output multiple candidate exception descriptions, which greatly shortens the problem diagnosis time. At the same time, the knowledge base is used for pattern matching to ensure the accuracy of the target exception description, further enhancing the pertinence of the solution. The exception repair strategy that matches the candidate exception description is provided to the user, achieving immediate response and efficient resolution of platform problems, significantly improving the operation and maintenance efficiency and user experience of the DevOps platform. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0046] Figure 1 A flowchart of a method for implementing intelligent customer service on a DevOps platform provided in an embodiment of the present application;

[0047] Figure 2 A schematic diagram of the structure of an intelligent customer service implementation device for a DevOps platform provided in an embodiment of the present application. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical solution and advantages of the present application clearer, the technical solution of the present application will be clearly and completely described below in combination with the specific embodiments of the present application and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application.

[0049] The embodiment of the present application provides a method for implementing intelligent customer service for a DevOps platform, which is applied to the DevOps platform. By analyzing the real-time status of the DevOps platform, the corresponding exception repair strategy is provided to the user, thereby helping the user to solve the exceptions of the DevOps platform in development, operation and maintenance, testing, etc. Compared with the traditional manual customer service system, the processing efficiency is significantly improved.

[0050] It should be noted that the intelligent customer service implementation method provided in the embodiment of the present application is mainly used in the DevOps platform. If this method is applied to the development of other systems, although the intelligent analysis and recommendation mechanism of the large model can be used to a certain extent, the development process of other systems may not have the complex, dynamic interactive real-time data flow and high-frequency fault diagnosis requirements of DevOps. In addition, the uniqueness of the DevOps platform lies in that it includes complex collaboration across multiple stages and multiple tools. It emphasizes the seamless integration between development, testing, deployment and operation and maintenance, and a large amount of real-time data is generated in each link, which requires rapid processing and feedback. Taking the above factors into consideration, when facing the daily development and operation and maintenance of the DevOps platform, the customer service system needs to have higher domain knowledge and powerful real-time data processing capabilities, which may not be so necessary for other types of development systems. Therefore, although this intelligent customer service technology can play a role in other fields, due to the high-frequency technology updates, huge tool chains and complex operation and maintenance environment of the DevOps platform, it is preferably applied to the DevOps platform.

[0051] The technical solutions provided by various embodiments of the present application are described in detail below in conjunction with the accompanying drawings.

[0052] like Figure 1As shown, an intelligent customer service implementation method for a DevOps platform provided in an embodiment of the present application includes:

[0053] S101: Collect the real-time operation data of the DevOps platform, input the real-time operation data into a preset big model, analyze the real-time operation data through the big model, and obtain the platform operation status corresponding to the DevOps platform.

[0054] The real-time operation data of the DevOps platform includes resource data, task layer data, and log data. The real-time operation data of the DevOps platform can be collected in real time through the real-time monitoring system. After collecting the real-time operation data, the real-time operation data is pre-processed, such as formatting, timestamp alignment, data cleaning, etc., and then the real-time operation data is classified into three types: resource data, CPU, GPU, task layer data (pipeline execution time, task status, etc.), and log data (error type, stack data, etc.). By parsing the classified real-time operation data through the large model and matching the real-time operation data with the operation data under normal operation, it can be determined whether the platform operation status corresponding to the current DevOps platform is normal or abnormal.

[0055] It should be noted that the big model can be a ChatGPT model or other big models with reasoning capabilities, which need to be trained in advance. The training process of the big model includes two steps: data collection and model training. First, during the data collection process, the development and operation data of the DevOps platform are collected and transmitted to the big model through timed API calls, real-time monitoring tools, user chat records and feedback storage. The development and operation data mainly include CI / CD pipeline parameters and logs, such as parameter information and log information in the source code, compilation, construction, testing, deployment and other stages; errors and abnormal log information during operation; resource monitoring information of system operation data such as CPU, memory, disk, network, etc.; configuration information such as environment configuration, version control records, dependencies, etc.; and user interaction history information of questions raised by users and their feedback. After that, the collected data is deduplicated, invalid data is eliminated, and abnormal data is sorted, and the development and operation data is clustered. According to the clustering results obtained after clustering, the development and operation data is divided into multiple categories of data such as logs, monitoring indicators, and error reports. At the same time, the data is uniformly formatted to construct a training data set.

[0056] In order to enable the big model to understand and process DevOps-related content during the training process, it is necessary to reduce training time and resources through transfer learning based on the pre-trained big model, and fine-tune it to focus on understanding DevOps-related content on the basis of general language understanding ability. Therefore, after constructing the training data set, it is also necessary to enhance the training data set. In this way, by training the big model with the enhanced training data set, the training samples can cover more scenarios and situations, so that the big model can learn a wider range of knowledge and improve its reasoning ability.

[0057] When performing data augmentation on the training dataset, the training dataset is first converted into a sentence type to convert the declarative sentences in the training dataset into question sentences, so as to better simulate the situations in which users may ask questions. For example, "task execution failed" is converted into "why did the task execution fail?" or "what is the reason for the task execution failure?" This conversion method can help the model better understand the problem-solving scenarios and different user interaction methods. For the training dataset obtained after the conversion, the sentences need to be expanded using relevant context or background information. For example, when describing anomalies in the DevOps platform, in addition to simply describing "process failure", more detailed information can be added to the data, such as "process failure, due to deployment configuration errors, the service cannot be started". In this way, by introducing more context, the model can better understand the actual problem and background. After adding context information, the training dataset is reconstructed in context, and the semantics are expanded by adjusting the word order in the sentence or introducing different contexts, so as to obtain the reconstructed training dataset. In the above-mentioned semantic expansion process, in addition to generating positive examples, that is, meaningful sentences, negative example sentences that are semantically irrelevant to the training data set can also be obtained. By expanding the training data set through negative example sentences, the model can learn which answers are irrelevant or invalid during the model training process, thereby improving the accuracy and robustness of the model.

[0058] To train the model based on the enhanced training data set, first adjust the model hyperparameters, such as the learning rate and weight initialization method. Initially, the learning rate should be as small as possible, which can be set to 0.0001 or smaller to prevent the model from failing to converge during training. Then use a hierarchical optimization algorithm to enable the model to retain global knowledge (the general knowledge of the pre-trained model) and learn local knowledge (DevOps-specific data). Based on the pre-trained model, use DevOps-specific data for fine-tuning. During the fine-tuning process, some layers of the pre-trained model (such as the embedding layer and the first few Transformer layers) can be fixed, and only the last layer or layers (such as the classification layer or the output layer) can be trained. This allows the model to adapt to specific tasks and data in the DevOps field while retaining global knowledge. After training, the training effect needs to be verified to avoid overfitting and ensure the robustness of the model to multiple scenarios.

[0059] S102: When there is an abnormality in the platform operation status, receive user questions sent by users, fill in the user questions according to the user's historical operation records and the current task parameters of the DevOps platform to generate corresponding question and answer statements.

[0060] When the DevOps platform is in an abnormal operation state, users can send corresponding user questions to the intelligent customer service system through the user interface provided by the intelligent customer service system in the DevOps platform. The intelligent customer service system connects to the large model obtained by the above training, analyzes and infers user questions through the large model, and can generate targeted answers for user questions. Among them, user questions support multiple input methods, including keywords, process IDs, question and answer texts, etc. After receiving the user questions, the intelligent customer service system fills in the user questions accordingly according to the user's historical operation records and the current task parameters of the DevOps platform, thereby generating standardized question and answer statements.

[0061] In one embodiment, historical operation records of users on the DevOps platform are collected, including but not limited to the tasks performed by the user, the tools used, the resources accessed, and the historical records of fault handling. By analyzing the historical operation records, the user's operation preferences can be identified, and the operation preferences include at least one or more of the following: fault handling strategy, operation mode, and operation field. Among them, the fault handling strategy refers to the repair strategy preferred by the user, the operation mode refers to the user's commonly used tools, configuration habits, and operation habits (preferring graphical interface or command line operation, etc.), and the operation field refers to the specific areas of concern to the user (such as build optimization, test coverage, etc.). The DevOps platform usually provides detailed information on the tasks currently being executed or about to be executed, including task type, task description, required resources, etc. According to the current task parameters of the DevOps platform, the task type corresponding to the current DevOps platform can be determined, such as code deployment, performance testing, troubleshooting, etc.

[0062] The identified user operation preferences are associated with the current task type. For example, if the user prefers to use command line operations and the current task is code deployment, it can be considered that the user is more interested in command line operations. Based on this association, interesting question words related to the task type and user operation preferences can be filtered out. Interested question words may involve the use of specific tools, troubleshooting steps, performance optimization suggestions, etc.

[0063] After determining the question words of interest, you can combine or fill them with the questions raised by the user to generate more specific and targeted question and answer statements. For example, if a user asks a question about code deployment, but the question is vague, you can refine or supplement the question based on the filtered question words. Finally, based on these filled question words and question background information, one or more question and answer statements can be generated. The question and answer statements not only answer the user's question, but also provide additional information related to the user's operation preferences and task types, thereby enhancing the pertinence and practicality of the question and answer.

[0064] S103: Input the question and answer statements into the big model, analyze the operation log of the DevOps platform and the question and answer statements through the big model, and output several candidate exception descriptions corresponding to the question and answer statements.

[0065] The intelligent customer service system inputs the received Q&A statements into the big model connected to it. The big model analyzes the Q&A statements and the running logs of the DevOps platform. By parsing the error logs and running status data, it analyzes the potential causes of the exceptions. For example, the keyword "database connection failure" that appears in the log may correspond to the problem of resource pool exhaustion. Then, the cause of the exception is located through the learned knowledge and context, and then combined with the keywords in the Q&A statement, several candidate exception descriptions corresponding to the Q&A statement are output. The candidate exception description refers to the type of exception, such as excessive server load, abnormal network connection, etc.

[0066] S104: Perform pattern matching on the operation log through a preset knowledge base to output a corresponding target exception description.

[0067] Candidate exception descriptions are obtained through preliminary reasoning of question-answer statements and log analysis by the big model, and there may be multiple of them. To accurately locate the cause of the exception, it is necessary to combine the knowledge base to perform pattern matching on the operation log, so as to filter out more accurate target exception descriptions from the candidate exception descriptions as the exception problems that occur on the current DevOps platform. The knowledge base can match the operation log and the exception description by rule matching (regular expression) or pattern matching (Bayesian classifier, etc.). For example, if the log contains "connection refused", the pattern matching rule will locate it as "network connection abnormality"; if the log contains "resource not found", the knowledge base will match it to "file path or configuration error". By verifying and supplementing the candidate exception descriptions through the knowledge base, the recognition accuracy of the exception description can be effectively improved.

[0068] S105: When the candidate exception description matches the target exception description, an exception repair strategy corresponding to the candidate exception description is obtained based on the knowledge base, and the exception repair strategy is pushed to the user.

[0069] When the candidate anomaly description inferred by the model is consistent with the target anomaly description matched by the knowledge base, the cause of the anomaly can be directly determined, and the anomaly repair strategy corresponding to the candidate anomaly description is generated based on the solution template provided in the knowledge base, and the anomaly repair strategy is pushed to the user. If the candidate anomaly description is inconsistent with the target anomaly description, the knowledge base can generate corresponding supplementary information to guide the large model to adjust the reasoning direction until a matching candidate anomaly description is derived.

[0070] In one embodiment, the embodiment of the present application can provide a personalized exception repair strategy push mechanism based on the user's operation preferences. That is to say, according to the operation preferences, the target path corresponding to the exception repair strategy is determined. The target path refers to an operation sequence consisting of multiple operations to implement the exception repair strategy. The operation may include selecting the push order of the strategy, selecting the display format of the strategy, and determining the policy organization form of the strategy. After obtaining the target path, the exception repair strategy is pushed to the user according to the target path, which can better adapt to the user's actual operation preferences, provide a push method for the exception repair strategy that is more in line with their usage habits, and improve the user experience.

[0071] Specifically, according to the fault handling strategy in the operation preference, the operation frequency corresponding to the candidate abnormal description of the user is determined, and then, according to the operation frequency, the user's fitness for the abnormal repair strategy is determined. Fitness is used to indicate the user's familiarity with the abnormal repair strategy. The higher the fitness, the more familiar the user is with the implementation process of such abnormal repair strategy. After obtaining the fitness, the fitness is compared with the preset fitness, and the policy organization form corresponding to the abnormal repair strategy is determined according to the size relationship between the fitness and the preset fitness. The policy organization form includes policy details and policy prompts. When the fitness is less than the preset fitness, it means that the user is not very familiar with the implementation process of the abnormal repair strategy. At this time, the policy organization form selected is the policy details, and a detailed tutorial of the abnormal repair strategy needs to be generated. On the contrary, when the fitness is not less than the preset fitness, it means that the user has handled similar abnormal situations. At this time, it is only necessary to provide the user with a simple policy prompt to help the user locate the cause of the abnormality and the strategy. The amount of information provided by the policy is adjusted according to the user's fitness to avoid excessive information provision to familiar users, while also ensuring that unfamiliar users obtain sufficient information support, which helps to reduce unnecessary customer service intervention and reduce operating costs.

[0072] The specific calculation process of fitness is as follows: First, based on the historical operation records, determine the average operation frequency of all users for the candidate anomaly description and the policy matching coefficient of the user for the anomaly repair strategy. The policy matching coefficient is obtained based on the average repair time corresponding to the anomaly repair strategy, which measures the degree of match between the strategy and the user's historical processing situation. If the average repair time of a certain strategy is short, and the user can quickly solve the problem when using the strategy in the past, then the degree of match between the strategy and the user is high, and the policy matching coefficient is also correspondingly high. Then, calculate the ratio between the user's current operation frequency and the average operation frequency. This ratio reflects the degree of deviation of the user's current frequency of using the strategy from the overall average level. Multiply the obtained ratio by the policy matching coefficient to obtain the user's fitness for the anomaly repair strategy. The higher the fitness value, the more frequently the user uses the strategy.

[0073] Furthermore, it is determined whether the anomaly repair strategy exists in the user's fault handling strategy. If so, it means that the user has selected this anomaly repair strategy when handling similar anomalies in the past. At this time, the priority corresponding to this strategy in the knowledge base needs to be updated. The priority reflects the effectiveness of the strategy and the user's preference for it. For the solutions selected by the user in historical problems, the priority is increased when adjusting the priority, so that when presenting candidate repair strategies, it can better meet the user's needs and expectations.

[0074] Furthermore, the corresponding operation habits are determined according to the user's operation mode. The operation habits are used to determine the display format corresponding to the abnormality repair strategy. For example, if the user prefers graphical operation, the abnormality repair strategy will be pushed to the user in the form of buttons or menu paths on the interface, while for users who are accustomed to using command lines, specific commands will be provided.

[0075] At this point, according to the above-obtained strategy organization form, operation habits and candidate order, the target path corresponding to the exception repair strategy is determined. Through the target path, personalized recommendations can be made for users, which are more in line with the user's operation preferences. For example, the final exception repair strategy pushed is a retry task, which is pushed to the platform interface in the form of command line instructions when pushed to the user. At the same time, considering that the user's adaptability to the exception repair strategy is low, the command line instructions also come with detailed tutorials. In addition, for the selected retry task, the priority of the exception repair strategy in the knowledge base will be increased accordingly. The knowledge base will also be regularly updated according to the platform usage and expert feedback to ensure that the answers provided by the system are always valid. By regularly collecting new problems and user feedback in the platform operation, they are sorted and included in the knowledge base, and the accuracy of the content is ensured through the expert verification mechanism. At the same time, the newly added knowledge base content is synchronized to the large model through fine-tuning or online learning to ensure that the model always has the latest domain knowledge and can accurately answer questions related to new technologies.

[0076] The above are embodiments of the method proposed in this application. Based on the same idea, some embodiments of this application also provide devices and non-volatile computer storage media corresponding to the above methods.

[0077] Figure 2 A schematic diagram of the structure of an intelligent customer service implementation device for a DevOps platform provided in an embodiment of the present application. Figure 2 As shown, including:

[0078] at least one processor; and,

[0079] at least one processor is communicatively connected to a memory; wherein,

[0080] The memory stores instructions that can be executed by at least one processor, and the instructions are executed by at least one processor so that the at least one processor can execute an intelligent customer service implementation method for a DevOps platform as described in any of the above items.

[0081] The embodiment of the present application provides a non-volatile computer storage medium storing computer executable instructions, wherein the computer executable instructions are configured as follows:

[0082] A method for implementing intelligent customer service for a DevOps platform as described in any of the above items.

[0083] Each embodiment in this application is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device and medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiments.

[0084] The devices and media provided in the embodiments of the present application correspond one-to-one to the methods. Therefore, the devices and media also have similar beneficial technical effects as the corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the devices and media will not be repeated here.

[0085] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0086] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0087] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0088] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0089] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0090] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0091] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0092] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0093] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A method for implementing intelligent customer service for a DevOps platform, characterized in that: The method comprises: Collecting real-time operation data of the DevOps platform, inputting the real-time operation data into a preset big model, parsing the real-time operation data through the big model, and obtaining the platform operation status corresponding to the DevOps platform; When there is an abnormality in the operation status of the platform, receiving a user question sent by a user, and filling in the user question according to the user's historical operation record and the current task parameters of the DevOps platform to generate a corresponding question and answer statement; Input the question and answer statement into the big model, analyze the operation log of the DevOps platform and the question and answer statement through the big model, and output several candidate exception descriptions corresponding to the question and answer statement; Performing pattern matching on the operation log through a preset knowledge base to output a corresponding target exception description; In the case where the candidate exception description matches the target exception description, an exception repair strategy corresponding to the candidate exception description is acquired based on the knowledge base, and the exception repair strategy is pushed to the user.

2. According to the method for implementing intelligent customer service for DevOps platform according to claim 1, it is characterized in that: According to the historical operation records of the user and the current task parameters of the DevOps platform, the user question is filled to generate a corresponding question and answer statement, specifically including: Analyze the historical operation records of the user to determine the operation preference of the user; wherein the operation preference includes at least one or more of the following: fault handling strategy, operation mode, and operation field; Determine the task type corresponding to the DevOps platform according to the current task parameters of the DevOps platform; Associating the task type with the user's operation preference to filter out the question words of interest corresponding to the task type according to the operation preference; The user question is filled in according to the associated words of interest to generate a corresponding question-and-answer sentence.

3. According to the method for implementing intelligent customer service for DevOps platform according to claim 2, it is characterized in that: Before pushing the anomaly repair strategy to the user, the method further includes: Determine, according to the operation preference, a target path corresponding to the anomaly repair strategy; wherein the target path refers to an operation sequence consisting of a plurality of operations for implementing the anomaly repair strategy; According to the target path, the abnormality repair strategy is pushed to the user.

4. The method for implementing intelligent customer service for a DevOps platform according to claim 3, characterized in that: Determining a target path corresponding to the abnormality repair strategy according to the operation preference specifically includes: Determine, according to the fault handling strategy in the operation preference, a strategy organization form corresponding to the abnormality repair strategy; Determine whether the abnormality repair strategy exists in the fault handling strategy, and if so, update the priority corresponding to the abnormality repair strategy in the knowledge base so that the knowledge base determines the candidate order of the abnormality repair strategy according to the priority; According to the operation mode, determining the operation habit corresponding to the user; wherein the operation habit is used to determine the display format corresponding to the abnormality repair strategy; According to the strategy organization form, the operation habits and the candidate order, a target path corresponding to the abnormal repair strategy is determined.

5. The method for implementing intelligent customer service for a DevOps platform according to claim 4, characterized in that: According to the fault handling strategy in the operation preference, a strategy organization form corresponding to the abnormal repair strategy is determined, specifically including: Determining, according to the fault handling strategy in the operation preference, the operation frequency corresponding to the candidate abnormality description by the user; Determining, according to the operation frequency, the adaptability of the user to the abnormality repair strategy; The fitness is compared with a preset fitness, and a policy organization form corresponding to the abnormal repair strategy is determined according to a size relationship between the fitness and the preset fitness; wherein the policy organization form includes policy details and policy prompts.

6. The method for implementing intelligent customer service for a DevOps platform according to claim 5, characterized in that: Determining the user's adaptability to the abnormality repair strategy according to the operation frequency specifically includes: Determine, according to the historical operation records, an average operation frequency corresponding to the abnormal repair strategy and a strategy matching coefficient of the user for the abnormal repair strategy; wherein the strategy matching coefficient is obtained according to an average repair time corresponding to the abnormal repair strategy; A ratio between the operating frequency and the average operating frequency is calculated, and a product of the ratio and the strategy matching coefficient is used as the adaptability of the user to the abnormality repair strategy.

7. The method for implementing intelligent customer service for a DevOps platform according to claim 1, characterized in that: Before collecting the real-time operation data of the DevOps platform, the method further includes: Collecting development and operation data corresponding to the DevOps platform, clustering the development and operation data, and constructing a training data set according to the clustering results obtained after clustering; The training data set is enhanced, and based on the enhanced training data set, a large model for intelligent customer service question and answering on the DevOps platform is trained.

8. The method for implementing intelligent customer service for a DevOps platform according to claim 7, characterized in that: Performing data enhancement on the training data set, specifically including: Performing sentence type conversion on the training data set to convert declarative sentences in the training data set into question sentences; For the converted training data set, reconstructing the context of the training data set to obtain the reconstructed training data set; Obtain negative example sentences that are unrelated to the semantics in the training data set, and expand the data in the reconstructed training data set according to the negative example sentences.

9. An intelligent customer service implementation device for a DevOps platform, characterized in that: The device comprises: at least one processor; and, a memory communicatively coupled to the at least one processor; In which, the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the intelligent customer service implementation method for a DevOps platform as described in any one of claims 1-8.

10. A non-volatile computer storage medium storing computer executable instructions, characterized in that: The computer executable instructions are configured to: A method for implementing intelligent customer service for a DevOps platform as described in any one of claims 1 to 8.

Images