DevOps deployment pipeline configuration generation method

Through the inference of custom word segmentation and LLM, the problems of insufficient user description analysis and complex pipeline configuration in DevOps are solved, and intelligent analysis and automated configuration of user expectations are realized, accurate configuration files are generated, and user experience is improved. The degree of automation of pipeline configuration.

CN118916040APending Publication Date: 2024-11-08BEIYIN FINANCIAL TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410956164.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-17
Publication Date
2024-11-08

AI Technical Summary

Technical Problem

Existing DevOps solutions cannot effectively resolve users' non-technical descriptions as specific technical configurations, which leads to difficulties for users configuring services. Especially for users with non-technical backgrounds, the pipeline configuration is complex and the user interface is not friendly enough, and lacks Automated pipeline configuration function.

Method used

Through custom word segmentation and word segmentation marking processing, LLM uses the branch name characteristics and the environment data information of the DevOps platform to construct the project stage and the latest products of the service to infer the specific environment of the service deployment, and automatically select the appropriate deployment template and fill it in Required parameters to generate configuration files for deployment. The specific steps include obtaining the text submitted by the user, performing data processing to obtain key terms and operational intent, using NLP library and GLM3 for context analysis, building a sentence dependency tree, determining verbs and semantic roles, converting them to structured formats, and verifying The result is output and the JSON configuration file is finally generated.

Benefits of technology

It realizes intelligent analysis of users' non-technical descriptions, automatically identify the types and details of the service that users expect to deploy, generate accurate technical configuration files, reduces the complexity and learning costs of users when configuring services, and increases the number of non-technical users. Friendly, providing automated pipeline configuration function.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118916040A_ABST
    Figure CN118916040A_ABST
Patent Text Reader

Abstract

The invention provides a DevOps deployment assembly line configuration generation method, and the method comprises the steps: carrying out the self-definition of segmented words, and carrying out the marking of the segmented words; the LLM deduces the specific environment of service deployment according to the specific stage of the project, the branch name features used by the latest product construction of the service and the environment data information of the deployment assembly line existing in the DevOps platform; and automatically generating a configuration file for deployment according to the functions of automatically selecting a proper deployment template and filling necessary parameters according to word segmentation, labels and deployment environment inference results. The non-technical description of the user can be accurately analyzed and converted into specific technical configuration.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of service deployment, and in particular to a DevOps deployment pipeline configuration generation method. Background Art

[0002] In the current DevOps practice, especially in the field of Java-based service deployment, there are a series of challenges. DevOps, as a culture, movement or practice that focuses on software development (Dev) and information technology operations (Ops), aims to increase the frequency of system deployment and achieve faster time to market while maintaining service quality and security. However, in actual operation, users and managers often encounter the following problems:

[0003] Technical challenges faced by users: Many DevOps platform users have insufficient knowledge of DevOps theory and practice, have a poor understanding of the concept of pipelines, and do not know how to achieve automated deployment. Users find it difficult to describe in detail the details of the services they expect to deploy, which makes the configuration process complex and error-prone.

[0004] Challenges for platform managers: Although DevOps platform managers have created a variety of pipeline templates to adapt to different deployment scenarios (such as deploying or updating services in a Kubernetes cluster, deploying services on a Linux host), effective template application is still highly dependent on the user's technical capabilities and understanding of pipeline configuration details.

[0005] Existing DevOps solutions have not yet been able to effectively address the above challenges. The main shortcomings include:

[0006] Deficiencies in service configuration analysis: There is a lack of effective technical means to accurately analyze the types and details of services that users want to configure and deploy.

[0007] Lack of automatic pipeline configuration: Existing systems usually cannot automatically configure pipelines for users. In the absence of a deep understanding of DevOps theory and practice, it is difficult for users to complete this process manually.

[0008] In order to address these shortcomings, existing solutions mainly focus on providing a more intuitive user interface and a simplified configuration process, but the solutions usually fail to go deep enough to intelligently analyze user needs, automatically select and configure pipeline templates. In addition, existing solutions usually lack interface designs that are friendly to non-technical users, and do not support adaptive learning and optimization based on user feedback.

[0009] The problems with existing technologies are: Insufficient analysis of user needs: Existing DevOps solutions are usually unable to accurately analyze users' non-technical descriptions into specific technical configurations, causing users to encounter difficulties when configuring services. Especially for users with non-technical backgrounds, this lack of analysis ability means that they cannot effectively use the DevOps platform.

[0010] Complexity of pipeline configuration: Existing systems often require users to have certain technical knowledge to manually select and configure pipeline templates, which is a major challenge for users who lack in-depth understanding of DevOps theory and practice.

[0011] Unfriendly user interface: The user interface design of most DevOps solutions is not intuitive enough, especially for non-technical users, which increases their learning cost and difficulty in using the platform.

[0012] Lack of intelligent user demand analysis: Existing DevOps solutions are usually unable to effectively convert users' non-technical descriptions into specific technical configurations. The problem that the present invention needs to solve is how to intelligently analyze user input and automatically identify the service type and details that users want to deploy.

[0013] Insufficient automated pipeline configuration: Current DevOps tools and platforms rarely provide the ability to automatically select and configure pipelines, especially in the absence of user technical expertise. Summary of the invention

[0014] In view of the above problems, the present invention is proposed to provide a DevOps deployment pipeline configuration generation method that overcomes the above problems or at least partially solves the above problems.

[0015] According to one aspect of the present invention, a DevOps deployment pipeline configuration generation method is provided, the configuration generation method comprising:

[0016] Customize word segmentation and perform word segmentation tagging;

[0017] LLM infers the specific environment of service deployment based on the specific stage of the project, the branch name characteristics used by the latest artifact build of the service, and the environmental data information of the deployment pipeline that already exists in the DevOps platform;

[0018] The function of automatically selecting a suitable deployment template and filling in necessary parameters based on the word segmentation, label and deployment environment inference results will automatically generate a configuration file for deployment.

[0019] Optionally, the specific environment of the service deployment includes: a development environment or a test environment.

[0020] Optionally, automatically generating a configuration file for deployment includes: selecting a correct template and filling in parameters according to project-specific requirements.

[0021] Optionally, the custom word segmentation specifically includes:

[0022] Get the text submitted by the user;

[0023] The text is processed using a data processing method to obtain key terms, relationships and operational intentions.

[0024] Optionally, the processing of the text using a data processing method specifically includes:

[0025] Regular expressions are used to remove non-text content, punctuation marks, and special characters to obtain preprocessed data;

[0026] Standardizing the preprocessed data to obtain a standard text;

[0027] Using the NLP library to segment the standard text into words or phrases;

[0028] Use pre-trained named entity recognition models to identify technical terms;

[0029] Conduct contextual analysis based on the technical terms described;

[0030] Convert the recognition and analysis results into a structured format to obtain output results;

[0031] The output result is verified.

[0032] Optionally, the performing context analysis according to the technical term specifically includes:

[0033] Using GLM3, understand the relationships between terms and their semantic roles in text;

[0034] Determine operational intent based on the contextual use of the term;

[0035] Use the NLP library to build a sentence dependency tree and understand the grammatical dependencies between terms;

[0036] Determine the semantic roles of verbs and related terms in a sentence to understand the goal and manner of the action.

[0037] Optionally, the function of automatically selecting a suitable deployment template and filling in necessary parameters based on the word segmentation, label and deployment environment inference results will automatically generate a configuration file for deployment, specifically including:

[0038] The word segmentation results, labels, and deployment environment information transmitted from the upstream system;

[0039] Input data is passed in a structured data format, including key information;

[0040] Outputs a fully configured deployment template file with all required parameters filled in;

[0041] Use the declarative API to create a pipeline based on a deployment template file;

[0042] The output format depends on the deployment system requirements, usually a JSON configuration file.

[0043] Optionally, the key information specifically includes: project name, service type, product URL address, deployment environment, service listening port and deployment type.

[0044] Optionally, the input data is passed in a structured data format, including key information; the output fully configured deployment template file specifically includes:

[0045] Select an appropriate deployment template from the preset template library based on the input environment type and service type;

[0046] Parse key information in the input data and fill it in according to the placeholders defined in the template;

[0047] Convert the template with filled parameters into the final deployment file, ensuring that all necessary parameters are filled correctly and comply with syntax rules.

[0048] The present invention provides a DevOps deployment pipeline configuration generation method, which includes: customizing word segmentation and performing word segmentation tagging processing; LLM infers the specific environment of service deployment based on the specific stage of the project, the branch name features used by the latest product construction of the service, and the environmental data information of the deployment pipeline that already exists in the DevOps platform; the function of automatically selecting a suitable deployment template and filling in necessary parameters based on the word segmentation, label and deployment environment inference results will automatically generate a configuration file for deployment. It can accurately parse the user's non-technical description and convert it into a specific technical configuration.

[0049] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented according to the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0051] Figure 1 A flowchart of a DevOps deployment pipeline configuration generation method provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0052] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0053] The terms "comprises" and "having" and any variations thereof in the description embodiments, claims and drawings of the present invention are intended to cover non-exclusive inclusions, for example, including a series of steps or units.

[0054] The technical solution of the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments.

[0055] like Figure 1 As shown, a DevOps deployment pipeline configuration generation method includes:

[0056] 1. Customize word segmentation and word segmentation labeling process

[0057] Word segmentation:

[0058] Input includes: text submitted by the user, such as "Please help me deploy the ncr-gateway service on the cloud". Format: string format, which may contain professional terms and technical commands.

[0059] The output includes: processed key terms and their relationships, as well as operational intent.

[0060] Format: A structured data format, such as JSON, containing keywords and their attributes and relationships, such as {"action":"deployment","platform":"cloud","service":"ncr-gateway"}.

[0061] The processing logic unit is divided into four functional modules, including:

[0062] Preprocessing modules include:

[0063] Text cleaning: Use regular expressions to remove non-text content (such as HTML tags), punctuation marks, special characters, etc.

[0064] Standardization: Unify different variants of terms, such as unifying all "cloud" and "on the cloud" as "cloud" to reduce the complexity of subsequent processing.

[0065] Word segmentation and term identification modules, including:

[0066] Tokenization: Use NLP libraries such as spaCy or NLTK to segment text into words or phrases.

[0067] Term Identification: Use a pre-trained Named Entity Recognition (NER) model to identify specific technical terms. Enhanced by specific entities included in the training dataset.

[0068] Terminology context analysis module, including:

[0069] Contextual Analysis: Using GLM3, to understand the relationships between terms and their semantic roles in the text.

[0070] Intent identification: Determine the intent of an operation based on the contextual usage of a term, e.g. a deployment action may involve specific services and platforms.

[0071] Dependency parsing: Use the NLP library to build a dependency tree for a sentence to understand the grammatical dependencies between terms.

[0072] Semantic Role Labeling (SRL): Determine the semantic roles of verbs and related terms in a sentence, such as "agent" and "target", to understand the target and method of the operation.

[0073] Post-processing and format output modules include: Data structuring: converting recognition and analysis results into a structured format (such as JSON) to facilitate use by other systems or modules.

[0074] Error detection and correction: Verify the output results to ensure that all information is accurate, without omissions or wrong associations.

[0075] Word segmentation and tagging processing, including: technical implementation details of highly customized entity recognition models for specific technical terms.

[0076] Input: The text output from the preprocessing module, which has been processed by basic word segmentation. Format: Text string or word sequence after word segmentation.

[0077] Output: The output of each vocabulary unit after labeling, including the specific attributes of the vocabulary and additional information related to it.

[0078] Format: structured data format, such as JSON, for example:

[0079] ncr: Identified as a 3-digit project code. The label may also contain other relevant information about the project, such as project phase and associated services.

[0080] gateway: Identifies as a service under the Java Spring framework and provides additional information such as common listening ports and the latest product download address.

[0081] On-cloud and deployment: Combined with "ncr", it is identified as an operation to deploy a service in a specific namespace on the Kubernetes cloud platform. The system also identifies whether this is the first deployment of the service and provides corresponding pipeline configuration parameters accordingly.

[0082] Processing logic unit, including:

[0083] Professional knowledge base integration module

[0084] LLM knowledge base: Uses the GLM3 large language model as the basis to understand and encode professional terminology through training models.

[0085] Vectorized storage: BGE_M3 embedding technology is used to convert terms and related information into vector form. This embedding can retain the semantic information of the terms, facilitating similarity calculation and retrieval.

[0086] Entity recognition and attribute assignment module, including:

[0087] Data preparation; Data collection;

[0088] Data extraction: Extract information from the DevOps platform, including system name, project name, 3-digit project code, service name, listening port number, development language, deployment frequency, etc. This data usually exists in project configuration files, deployment scripts, and CI / CD process records.

[0089] Data formatting: Format the collected data into JSON format to facilitate subsequent processing. For example, the data of each service can be represented as a JSON object.

[0090] Data annotation, including:

[0091] Automatic labeling: Automatically generate labels using existing project documents and DevOps metadata. For example, project names, service names, etc. can be directly used as labels for entities.

[0092] Manual review: Professionals review the results of automatic annotation to ensure the accuracy and completeness of the annotation.

[0093] Feature engineering, including:

[0094] Data analysis

[0095] JSON parsing: Parse the JSON format data exported from the DevOps platform. This data may contain system name, project name, 3-digit project code, service name, etc.

[0096] Data extraction: Extract the required information from the parsed structure, such as the service development language, listening port number, deployment frequency, etc.

[0097] Data cleaning

[0098] Missing value processing: Processing of missing data. Optional strategies include filling, deleting, or estimating using statistical methods (such as mean, median).

[0099] Outlier detection: Check outliers in the data set and process them to ensure data quality.

[0100] Feature extraction, including:

[0101] Text data processing

[0102] Lexical features: Extract lexical-level features such as n-gram and TF-IDF from text information such as project names and service names.

[0103] Embedding features: Use the pre-trained GLM model and BGE_M3 algorithm to generate high-quality embedding vectors for text data. These vectors can capture the deep semantic information of text data.

[0104] Processing of numerical data, including:

[0105] Standardization: Standardize numerical features such as listening port numbers and deployment frequencies (for example, Z-score standardization or minimum-maximum standardization) to eliminate the impact of different dimensions.

[0106] Discretization: Discretize some continuous variables, such as converting deployment frequency into high / medium / low frequency categories.

[0107] Processing of categorical data, including:

[0108] Encoding: Encode categorical data such as the development language of a service. Common methods include one-hot encoding or label encoding.

[0109] Feature combination and optimization, including:

[0110] Feature Selection

[0111] Information gain: Use methods such as information gain to evaluate the contribution of each feature to the target variable (such as entity recognition) and select the most valuable features.

[0112] Dimensionality reduction: If the feature dimension is too high, you can consider using methods such as PCA (principal component analysis) and LDA (linear discriminant analysis) to reduce the dimension.

[0113] The feature set includes:

[0114] Interaction features: Exploring the interaction between different features, such as the combination of service name and service development language, may have a better prediction effect on entity recognition.

[0115] Model Training

[0116] Model architecture selection

[0117] Select the base model: GLM3 is selected as the base model because GLM3 has shown superior performance in multiple language tasks, especially when dealing with complex text and entity relationships.

[0118] Adaptation model: Considering the needs of entity recognition, it may be necessary to introduce specific layers based on GLM3, such as CRF (conditional random field) layer, to optimize the boundary recognition of entities.

[0119] Training data preparation

[0120] Dataset division: The preprocessed and vectorized data is divided into three parts: training set, validation set and test set. The ratio is usually 70% training set, 15% validation set, and 15% test set.

[0121] Input formatting: Format the data into a format acceptable to GLM3, which usually includes the tokenids of the input text and the corresponding attention mask.

[0122] Training process

[0123] Initialization and Configuration

[0124] Parameter setting: Set parameters such as learning rate, batch size, training cycle (Epochs), etc. These parameters may need to be optimized through multiple experiments.

[0125] Optimizer selection: Choose an optimizer suitable for large pre-trained models, such as Adam or AdamW. These optimizers can help the model converge more stably during training.

[0126] Fine-tuning

[0127] Load pre-trained model: Load the pre-trained GLM3 model and the vectors generated by the BGE_M3 algorithm.

[0128] Fine-tuning goal: Fine-tune GLM3, focusing on adjusting the model to adapt to the entity and attribute data extracted from the DevOps platform.

[0129] Loss function: Use a loss function suitable for entity recognition, such as cross entropy loss, calculate the loss for each entity label, and optimize the prediction of entity boundaries through the CRF layer.

[0130] Training Execution

[0131] Batch processing: Feed data in batches for model training, and each batch contains multiple data points.

[0132] Forward propagation: The model performs forward propagation on each batch of data to calculate the output.

[0133] Loss calculation: Calculate the loss between the output and the true label.

[0134] Backpropagation: Perform backpropagation to update the model weights.

[0135] Attribute and additional information extraction module

[0136] Information retrieval: Retrieving information related to a specific entity from a vectorized knowledge base.

[0137] Dynamic Updates: Information in the knowledge base is dynamically updated based on the latest input and feedback to keep the data fresh.

[0138] Output formatting module, including:

[0139] Data formatting: Convert internal vector data into user-readable JSON format.

[0140] Error handling: Perform data integrity and consistency verification to ensure that all output data meets expected specifications.

[0141] 2. Environmental inference processing

[0142] LLM infers the specific environment (such as development environment or test environment) of service deployment based on information such as the specific stage of the project, the branch name characteristics used by the latest artifact build of the service, and the environmental data of the deployment pipeline that already exists in the DevOps platform.

[0143] The input includes the project name, service name, project stage, branch name of the latest artifact build, and environment data of the existing deployment pipeline.

[0144] Format: Input data is in structured JSON format, containing all necessary metadata and feature information.

[0145] Output the inferred service deployment environment, such as "development environment" and "test environment".

[0146] Format: The output will also be in structured JSON format, clearly indicating the inferred deployment environment.

[0147] Processing logic unit

[0148] Data preprocessing module

[0149] Input: Collect data from multiple sources, such as DevOps platform-continuous delivery state, DevOps platform-project management state tools, etc.

[0150] Data cleaning: remove invalid or incomplete records, standardize text formats (such as date and time stamp formats), and handle missing values.

[0151] Data transformation: Convert all input data into a format acceptable to the model, such as encoding categorical data into one-hot encoding.

[0152] Feature extraction module

[0153] Key feature extraction: Extract key information from the processed data, such as project stage, branch name, deployment history, etc.

[0154] Feature conversion: Convert text data into numerical vectors through NLP techniques, such as using TF-IDF or WordEmbedding.

[0155] TF-IDF: Converts text into a weight matrix that represents the importance of words in the document.

[0156] Word Embeddings: Use pre-trained word embedding models (such as Word2Vec or GLM3’s own embedding layer) to transform text.

[0157] Environmental Inference Module

[0158] Fine-tuning the implementation details of LLM

[0159] Training data preparation: Collect a large amount of real data about the project, including deployment status in various environments and branch usage details.

[0160] Model training: Based on the existing GLM3, fine-tune it using project-specific data. Focus on performance on different features, such as the correlation between project stage and deployment environment.

[0161] Model validation: Through cross-validation and other techniques, the model's inference ability on different types of data is evaluated to ensure the accuracy and generalization ability of the model.

[0162] Output formatting module

[0163] Data serialization: Convert the environment classification output by the model into JSON format.

[0164] Data verification: Perform format and logic verification on output data to ensure data integrity and correctness.

[0165] 3. Automatic configuration generation

[0166] We designed an automatic configuration parameter generation module to automatically select the appropriate deployment template and fill in the necessary parameters based on the word segmentation, label and deployment environment inference results. This module will highly automatically generate configuration files for deployment, including selecting the correct template and filling in parameters according to project-specific requirements.

[0167] The input includes the word segmentation results, labels, and deployment environment information passed from the upstream system (such as the environment inference module).

[0168] Format: Input data is passed in a structured data format (such as JSON), containing key information such as project name, service type, artifact URL address, deployment environment, service listening port, deployment type, etc.

[0169] Output: A fully configured deployment template file with all required parameters populated. Use the declarative API to create a pipeline based on the deployment template file.

[0170] Format: The output format depends on the deployment system requirements, usually a JSON configuration file.

[0171] Processing logic unit

[0172] Template selection module: Select the appropriate deployment template from the preset template library based on the input environment type and service type. For example, different configuration templates are selected depending on whether the service is running in a development environment or a production environment.

[0173] Parameter filling module:

[0174] Technical implementation: Parse the key information in the input data and fill it in according to the placeholders defined in the template. This includes settings such as database address, service port, environment-specific variables, etc.

[0175] Template rendering module: converts the template with filled parameters into the final deployment file, ensuring that all necessary parameters are filled correctly and conform to the syntax rules.

[0176] Beneficial effects: The present invention aims to use the ChatGLM model to automate the creation and configuration process of DevOps deployment pipelines based on natural language. This includes collecting user needs through conversations, parsing user needs, intelligently selecting pipeline templates and configuring pipeline templates, and continuous optimization based on user feedback.

[0177] The above specific implementation methods further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above are only specific implementation methods of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A DevOps deployment pipeline configuration generation method, characterized in that: The configuration generation method comprises: Customize word segmentation and perform word segmentation tagging; LLM infers the specific environment of service deployment based on the specific stage of the project, the branch name characteristics used by the latest artifact build of the service, and the environmental data information of the deployment pipeline that already exists in the DevOps platform; The function of automatically selecting a suitable deployment template and filling in necessary parameters based on the word segmentation, label and deployment environment inference results will automatically generate a configuration file for deployment.

2. A DevOps deployment pipeline configuration generation method according to claim 1, characterized in that: The specific environment of the service deployment specifically includes: a development environment or a test environment.

3. A DevOps deployment pipeline configuration generation method according to claim 1, characterized in that: The automatic generation of a configuration file for deployment includes: selecting a correct template and filling in parameters according to project-specific requirements.

4. A DevOps deployment pipeline configuration generation method according to claim 1, characterized in that: The custom word segmentation specifically includes: Get the text submitted by the user; The text is processed using a data processing method to obtain key terms, relationships and operational intentions.

5. A DevOps deployment pipeline configuration generation method according to claim 4, characterized in that: The method of processing the text using data processing specifically includes: Regular expressions are used to remove non-text content, punctuation marks, and special characters to obtain preprocessed data; Standardizing the preprocessed data to obtain a standard text; Using the NLP library to segment the standard text into words or phrases; Use pre-trained named entity recognition models to identify technical terms; Conduct contextual analysis based on the technical terms described; Convert the recognition and analysis results into a structured format to obtain output results; The output result is verified.

6. A DevOps deployment pipeline configuration generation method according to claim 5, characterized in that: The context analysis based on the technical terms specifically includes: Using GLM3, understand the relationships between terms and their semantic roles in text; Determine operational intent based on the contextual use of the term; Use the NLP library to build a sentence dependency tree and understand the grammatical dependencies between terms; Determine the semantic roles of verbs and related terms in a sentence to understand the goal and manner of the action.

7. A DevOps deployment pipeline configuration generation method according to claim 1, characterized in that: The function of automatically selecting a suitable deployment template and filling in necessary parameters based on the word segmentation, label and deployment environment inference results will automatically generate a configuration file for deployment, specifically including: The word segmentation results, labels, and deployment environment information transmitted from the upstream system; Input data is passed in a structured data format, including key information; Outputs a fully configured deployment template file with all required parameters filled in; Use the declarative API to create a pipeline based on a deployment template file; The output format depends on the deployment system requirements, usually a JSON configuration file.

8. A DevOps deployment pipeline configuration generation method according to claim 1, characterized in that: The key information specifically includes: project name, service type, product URL address, deployment environment, service listening port and deployment type.

9. A DevOps deployment pipeline configuration generation method according to claim 1, characterized in that: The input data is passed in a structured data format, including key information; Output fully configured deployment template file includes: Select an appropriate deployment template from the preset template library based on the input environment type and service type; Parse key information in the input data and fill it in according to the placeholders defined in the template; Convert the template with filled parameters into the final deployment file, ensuring that all necessary parameters are filled correctly and comply with syntax rules.