A method and system for automatically issuing configurations based on business intentions

Abstract

The present invention discloses a method and system for automatically issuing configurations oriented to business intentions, which relates to the field of network management technology, including: collecting natural language instructions input by users, parsing and extracting key information, performing preprocessing, and clarifying user business needs; constructing an intention recognition model, optimizing using historical behavior data, outputting network configuration tasks or actions, and clarifying user business intentions; analyzing the global topology of the network, and constructing a network model using graph theory methods; based on the user's business intentions, using the shortest path algorithm to determine the optimal business link; based on the determined business link and the clear user business intentions, converting business needs into specific network equipment configuration commands, and performing configuration verification and conflict monitoring. The implementation of this invention reduces network configuration errors, improves the automation level of network configuration, makes network management more efficient and accurate, and meets the needs for fast, flexible and efficient network management in modern network environments.

Description

Technical Field

[0001] The present invention relates to the field of network management, and in particular to a method and system for automatically issuing configurations oriented to business intentions. Background Art

[0002] In the contemporary network management field, with the increasing complexity of the network environment and the increasing requirements of enterprises for the accuracy and responsiveness of network configuration, the development of automated network configuration technology has become particularly critical. Automated network configuration technology aims to reduce manual intervention, improve the accuracy and efficiency of configuration tasks, and manage network devices and services through predefined policies and algorithms. The development of this technology field started with simple script automation and gradually evolved to the use of complex machine learning algorithms and artificial intelligence to analyze user needs and optimize network performance. Despite significant progress, existing network automation solutions still face multiple challenges.

[0003] The main shortcoming of existing technologies is that they are not efficient enough in terms of flexibility and accuracy in processing user intent. Most existing systems rely only on static rules and preset configuration templates, which are difficult to adapt to rapidly changing business needs and complex network environments. In addition, these systems often cannot fully understand user intent expressed in natural language, resulting in the need for intermediate conversion and interpretation by professionals, increasing the risk of configuration errors and time delays. In addition, existing automation tools are not smart enough in network topology analysis and optimization, lack the ability to respond to dynamic changes in network status in real time, and cannot effectively support decision-making based on real-time data.

[0004] In response to the above problems, the present invention proposes a method for automatic configuration delivery oriented to business intent. The method realizes the direct conversion from user business intent to specific network configuration by integrating advanced natural language processing technology and graph theory analysis. First, the present invention collects and parses the instructions input by the user through natural language processing technology, so that the system can directly understand and process the user's business needs. Then, the global topology of the network is intelligently analyzed using graph theory methods to optimize network configuration and performance management. Through this method, not only the accuracy and response speed of network configuration are improved, but also it can dynamically adapt to changes in network conditions and business needs. In addition, the present invention also includes a complete set of configuration verification and conflict monitoring mechanisms to ensure the reliability of network configuration and the security of network operation.

[0005] In summary, existing network configuration technologies have problems such as insufficient analysis of business intent and slow response to network optimization. Our invention provides solutions to these problems and belongs to the field of network management and automation technology. The implementation of this invention helps to achieve more efficient and reliable network management and meet the urgent needs for automation and intelligence in modern network environments. Summary of the invention

[0006] In view of the above-mentioned problems, the present invention is proposed.

[0007] Therefore, the technical problem solved by the present invention is: how to understand and extract business intent directly from the user's natural language input in automated network configuration, and how to perform intelligent analysis of the global network topology in real time to support fast and accurate network configuration decisions.

[0008] To solve the above technical problems, the present invention provides the following technical solutions: a method for automatically issuing configurations oriented to business intentions, comprising the following steps:

[0009] Collect natural language instructions input by users and parse them to extract key information.

[0010] Pre-process the extracted key information to clarify user business needs.

[0011] Build an intent recognition model, use the preprocessed key information and combine it with the user's historical behavior data for optimization, output the corresponding network configuration tasks or actions, and clarify the user's business intent.

[0012] Analyze the global topological structure of the network and construct a network model using graph theory methods.

[0013] Based on the user's business intent, the shortest path algorithm is used to determine the optimal business link.

[0014] According to the determined business links and clear user business intentions, the business requirements obtained from the intent recognition model are converted into specific network device configuration commands through mapping tables and template technology, and further configuration verification and conflict monitoring are performed.

[0015] As a preferred solution of the method for automatic configuration delivery based on business intent described in the present invention, the method of collecting natural language instructions input by users and parsing and extracting key information is to use natural language processing technology to analyze the natural language instructions input by users, perform word frequency statistics, part-of-speech tagging and entity recognition, build a context-aware word embedding model, introduce context relevance analysis, build a context-aware word embedding model to convert vocabulary into vectors, transmit the vectors to a hybrid neural network structure, obtain local text features and long-distance text dependencies, and identify user configuration instructions.

[0016] The context-aware word embedding model collects text data containing network configuration-related context, standardizes the collected text data, uses word embedding technology in natural language processing, analyzes the co-occurrence relationship and context information of words in the text, and generates vectors that reflect the semantics of words in a specific network configuration context.

[0017] The text data containing network configuration related context includes technical documents, user operation logs and historical configuration commands.

[0018] The hybrid neural network structure includes a convolutional neural network and a recurrent neural network.

[0019] As a preferred solution of the method for automatic configuration delivery based on business intent described in the present invention, the extracted key information is preprocessed to clarify the user's business needs by refining and cleaning the data based on the identified user configuration instructions, standardizing the extracted key information, unifying the expressions of different terms, and adjusting and verifying according to the user's historical behavior data and the contextual information of the current network environment, so as to match the requirements extracted from the user configuration instructions with the actual network configuration requirements.

[0020] As a preferred solution of the method for automatic configuration delivery for business intent described in the present invention, the construction of the intent recognition model, the optimization using the pre-processed key information and combining with the user's historical behavior data is based on the Transformer architecture, the input layer of the BERT model is customized, and a dedicated word embedding layer is constructed. Professional terminology of network technology and configuration commands is trained, and professional text data extracted from technical documents and configuration logs is used. By increasing the frequency of occurrence of professional terms and adjusting the dimension of the embedding vector, the intent recognition model's understanding of the terms is completed, the user's historical configuration data is collected and analyzed, and the intent recognition model is optimized according to the results of the historical data analysis.

[0021] The customization of the input layer of the BERT model includes building a dedicated word embedding layer. On the basis of the BERT model, the vocabulary of the processing is expanded, and the professional terms, device names and operation commands in the field of network configuration are added to the original vocabulary. The professional vocabulary that has not been covered but is frequently used is included in the original vocabulary. The dedicated word embedding layer is trained by using a customized training data set, and the expression is:

[0022]

[0023] in, is the position index of the target word currently being processed in the text, is the total number of words in the text, is the size of the context window, For a from arrive integer, To index the target word, for The context words, Target word The word vector representation of Context words The output word vector of is the total number of words in the vocabulary, is the transpose of the output word vector.

[0024] The customized training data set includes a network configuration manual, an actual configuration script, and a historical configuration log.

[0025] As a preferred solution of the method for automatic delivery of business intent-oriented configuration described in the present invention, the global topological structure of the network is analyzed and the network model is constructed using a graph theory method, which is to collect information about network devices and connections and perform standardized processing, and to construct a graph theory model using standardized network data. Network devices are represented as nodes in the graph, and connections are represented as edges. The attributes of nodes and edges are set according to actual network data, and the shortest path algorithm is used to analyze the connectivity between nodes and the optimality of the path, and the results of the graph theory analysis are used to construct the network model.

[0026] The connection information includes the location, model, IP address, and connection status of the device.

[0027] As a preferred solution of the method for automatically issuing configurations for business intent described in the present invention, wherein: the determination of the optimal business link based on the business intent of the user using the shortest path algorithm is based on matching the requirements extracted from the user configuration instructions with the actual network configuration requirements, using the Dijkstra algorithm, calculating the shortest path between two points in the graph theory model, initializing the distance and predecessor node of each vertex, the distance from the starting vertex to any vertex in the graph is infinite, setting the distance of the starting vertex to 0, selecting the unvisited node with the minimum distance, continuously updating the distance of adjacent vertices, and outputting the optimal business link;

[0028] Continuously select vertices with the smallest distance value that have not been visited. After the current vertex is selected, it will be marked as visited and will not be considered in subsequent operations. If the length of the new path is shorter than the known shortest path from the starting point to the vertex, the shortest path length of the adjacent vertices of the current vertex is updated, and the predecessor node of the current adjacent vertex is set to the current vertex.

[0029] As a preferred solution of the method for automatic configuration issuance oriented to business intentions described in the present invention, wherein: according to the determined business links and clear user business intentions, the business requirements obtained from the intent recognition model are converted into specific network device configuration commands through mapping tables and template technology. The method is to parse the user's specific business intentions and optimal business link requirements, use a mapping table to convert abstract business requirements into technical network configuration parameters, and automatically generate specific configuration commands through template technology to generate configuration instructions that are directly applied to network devices.

[0030] The configuration verification and conflict monitoring described above is to deploy automated test scripts to simulate network operations after completing the configuration, verify whether the new configuration works as expected, and enter the verification and conflict monitoring stage after implementing the new configuration in the actual network, monitor real-time data to verify the actual performance of the configuration, compare the new configuration with the existing network configuration, detect potential configuration conflicts, analyze the dependency of the new configuration on other devices or services in the network, and update the network management rule base and configuration policies based on the detected conflicts and problems. If the configuration conflict is caused by insufficient management process, improve the configuration management process to complete the configuration verification and conflict monitoring.

[0031] Another object of the present invention is to provide a business intent-oriented configuration automation distribution system, which can directly parse and convert the user's business intent into specific network configuration commands through natural language processing technology, and also use graph theory methods to perform real-time and dynamic analysis of network topology to ensure that the recommended network configuration is optimized, solving the problems of prior art in terms of flexibility, response speed and accuracy.

[0032] To solve the above technical problems, the present invention provides the following technical solutions: a business intent-oriented configuration automatic delivery system, comprising: an intent recognition module, a network analysis module, a configuration generation module, a configuration verification module, and a monitoring and feedback module.

[0033] The intent recognition module parses the natural language instructions input by the user and the business intentions expressed, uses deep learning and natural language processing technology to understand the user's instructions and needs, and converts abstract needs into clear technical parameters and configuration goals.

[0034] The network analysis module uses graph theory algorithms to analyze the current state and topological structure of the network, calculate the optimal path in the network, and determine the performance bottlenecks and improvement points of the network.

[0035] The configuration generation module obtains technical requirements from the intent recognition module and network data provided by the network analysis module. The configuration generation module uses mapping tables and template technology to automatically generate specific network configuration commands.

[0036] The configuration verification module verifies the correctness and effect of the configuration through simulation and automated testing before the configuration command is applied to the network device.

[0037] The monitoring and feedback module is when a new configuration is deployed to the network, the monitoring and feedback module starts to monitor the performance and stability of the network in real time.

[0038] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the above-mentioned method for automatically delivering configurations oriented to business intent are implemented.

[0039] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the above-mentioned method for automatically delivering configurations oriented to business intent.

[0040] Beneficial effects of the present invention: The present invention realizes the direct conversion from user natural language instructions to specific network configuration commands through natural language processing and graph analysis technology, accurately parses the user's business intentions, and allows the system to directly understand and respond to the configuration needs of non-technical users. The intention recognition model optimized by user historical data further improves the accuracy of command parsing and the ability of personalized configuration. By analyzing the global topology of the network through graph theory methods, the present invention not only optimizes the data flow path and improves network efficiency, but also enhances the reliability and stability of the network. Finally, the mapping table and template technology convert abstract business requirements into specific network device configuration commands, greatly improving the speed and accuracy of configuration. The implementation of this invention significantly reduces network configuration errors, improves the level of automation of network configuration, makes network management more efficient and accurate, and meets the needs of fast, flexible and efficient network management in modern network environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them:

[0042] Figure 1 An overall flow chart of a method for automatically delivering configurations oriented to business intent provided by the first embodiment of the present invention;

[0043] Figure 2 A structural diagram of a business intent-oriented configuration automatic delivery system provided in the second embodiment of the present invention. DETAILED DESCRIPTION

[0044] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0045] Example 1

[0046] Reference Figure 1 As an embodiment of the present invention, a method for automatically delivering configurations oriented to business intentions is provided, which is characterized by:

[0047] Collect natural language instructions input by users and parse them to extract key information.

[0048] Pre-process the extracted key information to clarify user business needs.

[0049] Build an intent recognition model, use the preprocessed key information and combine it with the user's historical behavior data for optimization, output the corresponding network configuration tasks or actions, and clarify the user's business intent.

[0050] Analyze the global topological structure of the network and construct a network model using graph theory methods.

[0051] Based on the user's business intent, the shortest path algorithm is used to determine the optimal business link.

[0052] According to the determined business links and clear user business intentions, the business requirements obtained from the intent recognition model are converted into specific network device configuration commands through mapping tables and template technology, and further configuration verification and conflict monitoring are performed.

[0053] Collecting natural language instructions input by users and parsing them to extract key information is to use natural language processing technology to perform word frequency statistics, part-of-speech tagging and entity recognition on the collected data, build a context-aware word embedding model, introduce context relevance analysis, build a context-aware word embedding model to convert vocabulary into vectors, transfer the vectors to a hybrid neural network structure, obtain local text features and long-distance text dependencies, and identify user configuration instructions.

[0054] The context-aware word embedding model collects text data containing network configuration-related context, standardizes the collected text data, uses word embedding technology in natural language processing, analyzes the co-occurrence relationship and contextual information of words in the text, and generates vectors that reflect the semantics of words in a specific network configuration context.

[0055] Text data containing network configuration-related context includes technical documentation, user operation logs, and historical configuration commands.

[0056] The hybrid neural network structure includes convolutional neural network and recurrent neural network. The convolutional neural network extracts local features in the text, and the recurrent neural network captures the long-term dependencies in the text sequence.

[0057] It is further explained that text data containing network configuration-related context, such as technical documents, user operation logs, and historical configuration commands, are selected as the basic data set for model training. The collected text data is standardized, including word segmentation, removal of stop words, and standardized terminology to ensure data quality and consistency. The word embedding technology in natural language processing, such as Word2Vec or GloVe, is used to train the word vector model. This process generates a vector representation that can reflect the semantics of the word in a specific network configuration context by analyzing the co-occurrence relationship and context information of words in the text. In order to enhance the context-awareness of the word vector, position encoding and context window technology are introduced to emphasize the relationship between the word and its surrounding words in order to better capture context-specific semantic information. By comparing the application effect in the actual network configuration task, the word vector model is evaluated and optimized to ensure that the model can effectively capture and apply the context characteristics of the network configuration.

[0058] Through this series of steps, the constructed context-aware word embedding model can effectively convert natural language related to network configuration into semantically rich vector form, providing support for subsequent automated processing and command generation. The construction of this model not only focuses on the vectorization of a single word, but also emphasizes the practicality and accuracy in specific network configuration tasks.

[0059] Natural language processing technology is used to collect and parse natural language instructions input by users. This process first processes the collected data through word frequency statistics, part-of-speech tagging, and entity recognition. Then, the constructed context-aware word embedding model is used to convert the vocabulary into vectors, which are trained in the context of specific network configuration commands to adapt to specific application requirements.

[0060] The vocabulary transformed into vectors is input into a hybrid neural network structure that combines convolutional neural networks (CNN) and recurrent neural networks (RNN). Convolutional neural networks specialize in extracting local features in text, while recurrent neural networks are responsible for capturing long-term dependencies in text sequences. The output of this hybrid network is a high-level representation of user intent that accurately reveals the user's configuration needs and instructions.

[0061] Ultimately, these high-level representations are used to generate specific network device configuration commands. This generation process involves mapping the high-level user intent representation output by the hybrid neural network to specific network configuration commands. The mapping process uses predefined templates and rules that are optimized based on the configuration syntax and parameters of different devices. The generated configuration commands are then verified and conflict-checked to ensure compatibility with existing network configurations before being officially issued to network devices for execution.

[0062] In this way, the present invention forms an effective closed-loop system, which not only demonstrates how the system can accurately capture the intentions from user input, but also explains how to convert these intentions into actual executable network configuration commands.

[0063] Preprocess the extracted key information to clarify user business needs. Refine and clean the data based on the identified user configuration instructions, standardize the identified keywords and parameters, unify the expression of different terms, and adjust and verify according to the user's historical configuration data and the context information of the current network environment, and match the requirements extracted from the user configuration instructions with the actual network configuration requirements.

[0064] What needs to be further explained is data reception and preliminary analysis:

[0065] In the first step of "identifying user configuration instructions", natural language processing technology is used to convert the user's language input into a preliminary word vector or tokenized data structure. This step mainly focuses on extracting key information from the original input, structuring it preliminarily, and identifying keywords and command statements.

[0066] When entering the second step of "preprocessing data and clarifying user business needs", the data will be further refined and cleaned based on the first step. For example, the system will standardize the identified keywords and parameters, unify the expression of different terms, such as unifying "set IP" and "set IP address" into "configure IP", and further verify and correct the correctness of this information according to the context, and check the validity of the IP address format.

[0067] In this step, the user's historical configuration data and the contextual information of the current network environment are integrated to ensure that the requirements extracted from the user's instructions match the actual network status. This includes considering previous configuration habits, device compatibility issues, etc., to provide support for generating accurate configuration commands.

[0068] Finally, a detailed business requirements model will be built based on the preprocessed data. This model will include all necessary configuration details, such as device type, required operations, parameter settings, etc., which are refined results obtained from a series of processing from the user's original instructions.

[0069] Building an intent recognition model and optimizing it using user historical data is based on the Transformer architecture. The input layer of the BERT model is customized, a dedicated word embedding layer is built, and professional terminology for network technology and configuration commands is trained. Professional text data extracted from technical documents and configuration logs is used to increase the frequency of professional terms and adjust the dimension of the embedding vector to complete the intent recognition model's understanding of the terms. The user's historical configuration data is collected and analyzed, and the intent recognition model is optimized based on the results of historical data analysis.

[0070] Customizing the input layer of the BERT model includes building a dedicated word embedding layer. Based on the BERT model, the vocabulary of processing is expanded. Professional terms, device names, and operation commands in the field of network configuration are added to the original vocabulary. Professional vocabulary that has not yet been covered but is frequently used is included in the original vocabulary. The dedicated word embedding layer is trained by using a customized training data set. The expression is:

[0071]

[0072] in, is the position index of the target word currently being processed in the text, is the total number of words in the text, is the size of the context window, For a from arrive integer, To index the target word, for The context words, Target word The word vector representation of Context words The output word vector of is the total number of words in the vocabulary, is the transpose of the output word vector.

[0073] The customized training data set includes network configuration manual, actual configuration scripts and historical configuration logs.

[0074] It should be further explained that the field of network configuration contains a large number of specific professional terms, which may be expressed differently in different manufacturers, devices and technical standards. For example, the same network operation may be named as different commands in different router configuration languages. Traditional intent recognition models are mostly based on general language processing, and they often use a wide range of training data sets that lack in-depth training on specific network technology terms. Therefore, it is difficult for these models to understand and correctly process professional vocabulary and complex commands in the field of network configuration. Network configuration requirements are not only complex but also very dynamic. As the network environment and business needs change, the relevant configuration commands and parameter settings need to be adjusted accordingly. Traditional models lack the ability to learn and adapt from real-time changing network environments, which limits their application effects in dynamic network configuration scenarios.

[0075] BERT is a Transformer-based language model. The intent recognition optimization based on the Transformer and BERT models of the present invention is optimized specifically for complex terms and operation commands for network configuration by customizing the input layer and word embedding layer of the BERT model. This includes expanding the vocabulary of the model to include professional terms in the field of network configuration, and adjusting the dimensions of word vectors to better reflect the semantic relationships and usage contexts of these terms. This customized approach enables the model to accurately understand and process specific network configuration commands, significantly improving the accuracy of command parsing and the response speed of the system, thereby reducing configuration errors and improving the efficiency of network management.

[0076] The present invention uses customized training data sets for model training, which uses professional text data extracted from network configuration manuals, actual configuration scripts, and historical configuration logs. These data directly reflect the actual operations and scenarios of network configuration. In this way, the intent recognition model not only learns the standard operations of network configuration, but also optimizes its performance based on historical data to better adapt to the configuration habits of specific users and the configuration requirements of specific networks. The use of such historical data significantly improves the practicality of the model and the relevance of configuration recommendations.

[0077] Analyzing the global topology of the network and building a network model using graph theory methods involves collecting information about network devices and connections and standardizing it. Using standardized network data to build a graph theory model, network devices are represented as nodes in the graph, connections are represented as edges, and the properties of nodes and edges are set according to the actual network data. The shortest path algorithm is used to analyze the connectivity between nodes and the optimality of the path, and the results of the graph theory analysis are used to build a network model.

[0078] The connection information includes the device's location, model, IP address, and connection status.

[0079] It is further noted that the output of the intent recognition model includes key actions and parameters in the user's instructions, such as device type, configured attributes and values, etc. For example, if the output is "Set the router's IP address to 192.168.1.1", the model will recognize "Set" as the action, "router" as the device type, "IP address" as the configuration item, and "192.168.1.1" as the parameter value.

[0080] Map to network topology.

[0081] Use the configuration tasks and parameters obtained to map to the actual network topology. This usually requires the network management system to already have a detailed database or map of network devices and connections.

[0082] Based on the output device type and configuration items, the system can determine which specific network device or device group needs to be configured or updated.

[0083] The present invention applies graph theory methods to analyze or update the network structure based on the existing network topology. For example, if the configuration task involves adding or modifying routes, the system needs to calculate and update the paths that may be affected. Graph theory algorithms, such as the Dijkstra algorithm, are used to determine the affected service links and the optimal path to ensure that the efficiency and stability of the network will not be negatively affected.

[0084] Based on the information obtained from the intent recognition model and the results of graph theory analysis, specific network configuration instructions are generated. These instructions need to ensure the overall performance of the network while meeting the specific needs of users.

[0085] At this stage, you also need to consider the compatibility, security, and impact of existing configurations of the device to avoid conflicts or security vulnerabilities.

[0086] Before finalizing any configuration changes, perform a thorough validation to ensure that all new configurations will work properly and will not disrupt other parts of the network.

[0087] Configuration changes can be tested in a simulated or sandbox environment to ensure their correctness and effectiveness before being deployed in the actual network.

[0088] Through these steps, the output obtained from the intent recognition model is not only directly applied to actual network configuration tasks, but can also be integrated into global network management and optimization. This approach ensures the accuracy, relevance and ultimate successful execution of network configuration activities, while also making network management more efficient and automated. The key to this process is to efficiently convert the configuration instructions obtained by natural language processing into specific network operations, and to ensure the correct implementation of these operations in the entire network architecture.

[0089] Based on the user's business intention, the shortest path algorithm is used to determine the optimal business link. The Dijkstra algorithm is used to match the requirements extracted from the user configuration instructions with the actual network configuration requirements. The shortest path between two points in the graph theory model is calculated, and the distance and predecessor node of each vertex are initialized. The distance from the starting vertex to any vertex in the graph is infinite. The distance of the starting vertex is set to 0, and the unvisited node with the minimum distance is selected. The distance of the adjacent vertices is continuously updated to output the optimal business link. The expression is:

[0090] ,

[0091] in, dist[v] is the shortest distance from the starting point to the vertex v, prev[v] It is the previous vertex of vertex v in the shortest path tree. Undefined means that at the beginning of the algorithm, no vertex’s predecessor is known. dist[s] Indicates that the distance from the starting vertex S to itself is 0, for each neighbor of For each adjacent point v of vertex u, find the shortest path from u to v. is the weight of the edge from u to v, is a trial value for the length of a new path from vertex u to vertex v.

[0092] Continue to select the vertex u that has not been visited and has the smallest dist value. After selection, the current vertex u will be marked as visited and will not be considered in subsequent operations. If the length alt of the new path is shorter than the known shortest path dist[v] from the starting point to the vertex v, then update the value of the current dist[v] and set the predecessor node of vertex v to u.

[0093] It should be further explained that in traditional network configurations, determining the optimal path for data transmission usually requires network engineers to manually configure routing, which is not only time-consuming but also prone to errors, especially in large or complex network environments. In addition, traditional methods are often unable to dynamically adjust quickly according to changes in business needs, resulting in the inability to maximize network performance. The present invention sets the initial distance of each vertex to infinity (indicating unknown or unreachable), except that the distance of the starting vertex is set to 0. This initialization ensures that the algorithm starts from the starting point and can correctly evaluate the distance to all other points.

[0094] The algorithm continuously selects unvisited vertices with the minimum distance estimate and updates the distance of the vertex's neighboring points. In this process, if the path from the current vertex to the neighboring point is shorter than the known path, the path length is updated and the path source is recorded, which ensures that the shortest path of each node can be found and updated. Through the above steps, the algorithm finally outputs the shortest path from the starting point to any other point in the network. These paths are the optimal business links.

[0095] The present invention can respond quickly and accurately to changes in business requirements through automated shortest path calculation, provide optimal routes for data transmission, and thus maximize the utilization efficiency of network resources and reduce data transmission delays. Integrated automation tools can monitor changes in network status and business requirements in real time, and dynamically adjust data paths, which is difficult to achieve in traditional manual configuration methods. Automated path calculation reduces network failures that may be caused by human configuration errors and improves network stability and reliability. Through automated business link optimization, network administrators can manage the network more effectively and free up human resources to handle more complex tasks.

[0096] According to the determined business links and user intentions, the abstract business requirements are converted into specific network device configuration commands through mapping tables and template technology. The purpose is to parse the user's specific business intentions and optimal business link requirements, use mapping tables to convert abstract business requirements into technical network configuration parameters, and automatically generate specific configuration commands through template technology to generate configuration instructions that are directly applied to network devices.

[0097] Configuration verification and conflict monitoring is to deploy automated test scripts to simulate network operations after completing the configuration, verify whether the new configuration works as expected, and enter the verification and conflict monitoring stage after implementing the new configuration in the actual network. Real-time data is monitored to verify the actual performance of the configuration, and the new configuration is compared with the existing network configuration to detect potential configuration conflicts. The dependency of the new configuration on other devices or services in the network is analyzed. Based on the detected conflicts and problems, the network management rule base and configuration policies are updated. If the configuration conflict is caused by insufficient management processes, the configuration management process is improved to complete configuration verification and conflict monitoring.

[0098] It is further explained that in the present invention, the process of converting abstract business requirements into specific network device configuration commands through mapping tables and template technology according to the determined business links and user intentions is first based on the user's specific business intentions and the optimal business link determined by the shortest path algorithm. This process begins with parsing user requirements and the network link requirements determined thereby, and then using mapping tables to convert these high-level requirements into technical network configuration parameters. These mapping tables predefine the correspondence between business requirements and technical parameters, for example, data transmission efficiency may be mapped to bandwidth requirements and routing priorities. Subsequently, template technology is used to automatically generate specific configuration commands, and these templates fill in preset configuration scripts or command lines according to the input parameters, thereby generating configuration instructions that can be directly applied to network devices. This whole set of processes not only ensures the faithful execution of business requirements, but also greatly improves the efficiency and accuracy of network configuration, while ensuring that the new configuration will not conflict with the existing settings through automated verification and conflict detection mechanisms, thereby maintaining the stability and reliability of the network.

[0099] Deploy automated test scripts to simulate network operations and verify that the new configuration works as expected. This can include traffic simulation, performance benchmarking, and recovery testing to ensure the validity of the configuration and that performance indicators meet requirements.

[0100] After implementing the new configuration in the actual network, monitor real-time data to verify the actual performance of the configuration. This includes monitoring key performance indicators such as network traffic, latency, and error rates to ensure that they are within acceptable ranges. Compare the new configuration with the existing network configuration to detect potential configuration conflicts. This includes issues such as IP address conflicts, inconsistent routing tables, and overlapping policies. This process can be automated using dedicated network management tools. Analyze the dependencies of the new configuration on other devices or services in the network. Ensure that the new configuration does not disrupt existing dependencies or cause service interruptions. Rule base and policy updates: Based on detected conflicts and issues, update the network management rule base and configuration policies to prevent future conflicts.

[0101] If it is found that the configuration conflict is caused by insufficient management process, the configuration management process should be improved, such as introducing stricter change control procedures, enhanced audit trails, and improved permission management. Collect feedback from network operators, system administrators, and end users on the performance of the new configuration. Based on the collected feedback and monitoring data, make necessary adjustments and optimizations to the configuration to improve the performance and reliability of the network.

[0102] Example 2

[0103] Reference Figure 2, as an embodiment of the present invention, provides a system for a business intent-oriented configuration automatic delivery method, characterized in that it includes an intent recognition module, a network analysis module, a configuration generation module, a configuration verification module, and a monitoring and feedback module.

[0104] The intent recognition module parses the natural language instructions input by the user and the business intentions expressed, uses deep learning and natural language processing technology to understand the user's instructions and needs, and converts abstract needs into clear technical parameters and configuration goals.

[0105] The network analysis module uses graph theory algorithms to analyze the current state and topology of the network, calculate the optimal path in the network, and determine the performance bottlenecks and improvement points of the network.

[0106] The configuration generation module obtains technical requirements from the intent recognition module and network data provided by the network analysis module. The configuration generation module uses mapping tables and template technology to automatically generate specific network configuration commands.

[0107] The configuration verification module verifies the correctness and effect of the configuration through simulation and automated testing before the configuration command is applied to the network device.

[0108] The monitoring and feedback module is when a new configuration is deployed to the network, the monitoring and feedback module starts to monitor the performance and stability of the network in real time.

[0109] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program codes.

[0110] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0111] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0112] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0113] Example 3

[0114] In this embodiment, in order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments. This embodiment experiments the existing traditional method and the method of this embodiment respectively.

[0115] This simulation experiment aims to verify the effectiveness and superiority of a business intent-oriented configuration automation delivery method. The experiment simulates a real network environment, collects the user's natural language input and applies it to the automated network configuration system to verify the application effect of the present invention in network management practice. The experimental process includes parsing user instructions, preprocessing data, intent recognition, network topology analysis, shortest path calculation, generating configuration commands, and configuration verification and conflict monitoring.

[0116] In the experimental setting, the input user intent was set as "increase the bandwidth from the data center to the office area" or "set firewall rules to block access from specific IPs." These commands were preprocessed to clarify the instructions and analyzed by the built intent recognition model, which is based on the optimized BERT architecture and is specially trained to understand network configuration related terms.

[0117] "Increase bandwidth" can refer to a variety of specific actions, such as increasing total bandwidth, optimizing bandwidth usage, or changing bandwidth configuration, and preprocessing can help clarify the specific intent of these instructions. Users may use non-standard or colloquial expressions, such as "block" instead of "disable". In the preprocessing step, these terms are converted into standard terms that the system can recognize and process.

[0118] The global topology of the network is analyzed by graph theory, and the Dijkstra algorithm is used to calculate the optimal service link. Based on the analysis results, the system automatically generates accurate network device configuration commands through mapping tables and template technology. Finally, the system performs configuration verification and conflict monitoring to ensure the correctness of the new configuration and the stability of the network.

[0119]

[0120] As can be seen from the table, the accuracy of user intent recognition is as high as 89% on average, which shows the efficient processing capability of the optimized BERT model for professional network configuration terms. The time for topology analysis and path optimization is controlled within a short range, with an average topology analysis time of only 2.12 seconds and an average path optimization time of 113 milliseconds, reflecting the significant advantages of the present invention in improving network configuration speed.

[0121] In addition, the number of configuration commands generated and the high conflict detection rate (96.5% on average) indicate that the system can not only effectively generate the necessary configuration commands, but also effectively prevent potential configuration problems. The overall success rate averaged 92.5%, which further confirms the effectiveness of this automated delivery method in improving the accuracy and reliability of network configuration.

[0122] Through these data, the beneficial effects of the present invention are fully reflected, including improving configuration accuracy, accelerating configuration speed, optimizing network performance and enhancing system stability.

[0123] Table 2 Data comparison table

[0124]

[0125] In this simulation experiment, the generated data table shows the significant differences between the present invention and traditional methods in key performance indicators such as user intent accuracy, configuration time and success rate through intuitive comparison. From the data, it can be seen that the present invention has improved the user intent accuracy by an average of 10%, reduced the configuration time by an average of about 50%, and increased the success rate by an average of 7%. These data clearly demonstrate the significant advantages of the present invention in improving the accuracy and efficiency of network configuration and reducing the configuration error rate. Through this innovative automation method, the present invention not only effectively optimizes network performance, but also greatly improves the stability and reliability of the entire system.

[0126] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for automatically issuing configurations based on business intentions, characterized in that: include: Collect natural language instructions input by users and parse them to extract key information; Pre-process the extracted key information to clarify user business needs; Build an intent recognition model, use the preprocessed key information and combine it with the user's historical behavior data for optimization, output the corresponding network configuration tasks or actions, and clarify the user's business intent; Analyze the global topological structure of the network and construct a network model using graph theory methods; Based on the user's business intent, the shortest path algorithm is used to determine the optimal business link; According to the determined business links and clear user business intentions, the business requirements obtained from the intent recognition model are converted into specific network equipment configuration commands through mapping tables and template technology, and configuration verification and conflict monitoring are further performed; the collection of natural language instructions input by the user and parsing and extracting key information is to use natural language processing technology to analyze the natural language instructions input by the user, perform word frequency statistics, part-of-speech tagging and entity recognition, build a context-aware word embedding model, introduce context relevance analysis, build a context-aware word embedding model to convert vocabulary into vectors, transfer the vectors to the use of a hybrid neural network structure, obtain local text features and long-distance text dependencies, and identify the user's configuration instructions; The context-aware word embedding model collects text data containing network configuration-related context, standardizes the collected text data, uses word embedding technology in natural language processing, analyzes the co-occurrence relationship and context information of words in the text, and generates vectors that reflect the semantics of words in a specific network configuration context; The text data containing the network configuration related context includes technical documents, user operation logs and historical configuration commands; The hybrid neural network structure includes a convolutional neural network and a recurrent neural network; the extracted key information is preprocessed to clarify the user's business needs, and the data is refined and cleaned based on the identified user configuration instructions, the extracted key information is standardized, the expression of different terms is unified, and the user's historical behavior data and the context information of the current network environment are adjusted and verified, and the requirements extracted from the user configuration instructions are matched with the actual network configuration requirements; the construction of the intent recognition model, the optimization using the preprocessed key information and combining the user's historical behavior data is based on the Transformer architecture, the input layer of the BERT model is customized, a dedicated word embedding layer is constructed, and professional terms for network technology and configuration commands are trained. Professional text data extracted from technical documents and configuration logs are used to increase the frequency of occurrence of professional terms and adjust the dimension of the embedding vector to complete the understanding of the terminology by the intent recognition model, collect and analyze the user's historical configuration data, and optimize the intent recognition model based on the results of the historical data analysis; The customization of the input layer of the BERT model includes building a dedicated word embedding layer. On the basis of the BERT model, the vocabulary of the processing is expanded, and the professional terms, device names and operation commands in the field of network configuration are added to the original vocabulary. The professional vocabulary that has not been covered but is frequently used is included in the original vocabulary. The dedicated word embedding layer is trained by using a customized training data set, and the expression is: Where t is the position index of the target word currently being processed in the text, N is the total number of words in the text, c is the size of the context window, j is an integer from -c to c, and w t is the index target word, w t+j w t The context words, The target word w t The word vector representation of is the context word w t+j The output word vector of , W is the total number of words in the vocabulary, is the transpose of the output word vector; The customized training data set includes a network configuration manual, an actual configuration script, and a historical configuration log.

2. The method for automatically issuing configurations based on business intent according to claim 1, characterized in that: The analysis of the global topological structure of the network and the construction of the network model using the graph theory method is to collect information about network devices and connections and perform standardized processing, and use standardized network data to build a graph theory model, in which network devices are represented as nodes in the graph, and connections are represented as edges. The attributes of the nodes and edges are set according to the actual network data, and the shortest path algorithm is used to analyze the connectivity between nodes and the optimality of the path, and the results of the graph theory analysis are used to build the network model; The connection information includes the location, model, IP address, and connection status of the device.

3. The method for automatically issuing configurations based on business intent as described in claim 2, characterized in that: The method of using the shortest path algorithm to determine the optimal service link based on the service intention of the user is to use the Dijkstra algorithm after matching the requirements extracted from the user configuration instructions with the actual network configuration requirements, calculate the shortest path between two points in the graph theory model, initialize the distance and predecessor node of each vertex, set the distance from the starting vertex to any vertex in the graph to be infinite, set the distance of the starting vertex to 0, select the unvisited node with the minimum distance, continuously update the distance of adjacent vertices, and output the optimal service link; Continuously select vertices with the smallest distance value that have not been visited. After the current vertex is selected, it will be marked as visited and will not be considered in subsequent operations. If the length of the new path is shorter than the known shortest path from the starting point to the vertex, the shortest path length of the adjacent vertices of the current vertex is updated, and the predecessor node of the current adjacent vertex is set to the current vertex.

4. The method for automatically issuing configurations based on business intentions as claimed in claim 3, characterized in that: The method of converting the business requirements obtained from the intent recognition model into specific network device configuration commands through mapping tables and template technology according to the determined business links and clear user business intentions is to analyze the user's specific business intentions and optimal business link requirements, use mapping tables to convert abstract business requirements into technical network configuration parameters, and automatically generate specific configuration commands through template technology to generate configuration instructions directly applied to network devices; The configuration verification and conflict monitoring described above is to deploy automated test scripts to simulate network operations after completing the configuration, verify whether the new configuration works as expected, and enter the verification and conflict monitoring stage after implementing the new configuration in the actual network, monitor real-time data to verify the actual performance of the configuration, compare the new configuration with the existing network configuration, detect potential configuration conflicts, analyze the dependency of the new configuration on other devices or services in the network, and update the network management rule base and configuration policies based on the detected conflicts and problems. If the configuration conflict is caused by insufficient management process, improve the configuration management process to complete the configuration verification and conflict monitoring.

5. A system using a business intention-oriented configuration automatic delivery method as claimed in any one of claims 1 to 4, characterized in that: It includes intention recognition module, network analysis module, configuration generation module, configuration verification module and monitoring and feedback module; The intent recognition module analyzes the natural language instructions input by the user and the business intent expressed, uses deep learning and natural language processing technology to understand the user's instructions and needs, and converts abstract needs into clear technical parameters and configuration goals; The network analysis module uses graph theory algorithms to analyze the current state and topology of the network, calculate the optimal path in the network, and determine the performance bottlenecks and improvement points of the network; The configuration generation module is based on the technical requirements obtained by the intention recognition module and the network data provided by the network analysis module. The configuration generation module uses a mapping table and template technology to automatically generate specific network configuration commands; The configuration verification module verifies the correctness and effect of the configuration through simulation and automated testing before the configuration command is applied to the network device; The monitoring and feedback module is when a new configuration is deployed to the network, the monitoring and feedback module starts to monitor the performance and stability of the network in real time.

6. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the processor implements the steps of a method for automatically delivering configurations oriented to business intent according to any one of claims 1 to 4.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of a method for automatically delivering configurations oriented to business intent as described in any one of claims 1 to 4 are implemented.

Images