Computer intelligent service management system and method based on network big data

Abstract

The application discloses a computer intelligent service management system and method based on network big data, and belongs to the technical field of operation and maintenance management. The system comprises a data flow generation module, a knowledge network generation module, a candidate set generation module and an action execution module. The data flow generation module is used for forming context-aware data flow carrying management intention; the knowledge network generation module is used for forming knowledge network available for business intention query; the candidate set generation module is used for generating root cause candidate set with priority; and the action execution module is used for executing corresponding automatic repair action. The application achieves the technical effects of intelligent correlation analysis of operation and maintenance data, accurate root cause positioning and system self-healing capability by constructing context-aware data flow with injected business semantics, establishing multi-dimensional context index knowledge network and realizing automatic repair mechanism based on priority evaluation, and solves the problems of weak correlation of operation and maintenance data, low fault positioning efficiency and slow manual response speed in the prior art.

Description

Technical Field

[0001] This invention relates to the field of operation and maintenance management technology, and in particular to a computer intelligent service management system and method based on network big data. Background Technology

[0002] With the widespread adoption of cloud computing, containerization, and microservice architectures, modern software systems are exhibiting highly distributed and dynamically changing characteristics, with the number of service components growing exponentially and network data traffic surging continuously. Against this backdrop, intelligent service management based on network big data has become a key technology for ensuring stable system operation. It achieves intelligent perception and automated management of service status through real-time collection, analysis, and processing of massive amounts of operational data.

[0003] The above-disclosed technical solutions have at least the following technical problems:

[0004] Traditional methods suffer from several drawbacks. The lack of unified data identification standards and business semantic injection mechanisms leads to information silos formed by operational data from different sources, resulting in weak data correlation. Furthermore, the absence of intelligent indexing methods oriented towards business intent leads to low data retrieval efficiency and insufficient root cause localization accuracy during fault analysis. Additionally, reliance on manual experience for fault repair results in slow response times, failing to meet the high availability requirements of modern distributed systems. To address these issues, this invention proposes a solution. Summary of the Invention

[0005] This application provides a computer intelligent service management system and method based on network big data, which solves the problems of weak correlation of operation and maintenance data, low efficiency of fault location and slow manual response in the prior art, and realizes the technical effects of intelligent correlation of operation and maintenance data, efficient root cause location and automated fault repair.

[0006] This application provides a computer intelligent service management system based on network big data, including: a data stream generation module: used to acquire network big data, inject a unified tracking identifier and a management context containing business semantics according to a predefined management strategy, and form a context-aware data stream carrying management intent;

[0007] Knowledge network generation module: used to aggregate the context-aware data stream and, based on the management context, construct a multi-dimensional context index to form a knowledge network that can be queried by business intent;

[0008] Candidate set generation module: When an alarm or query request is received, it is parsed as a management intent query, and the associated context-aware data stream is retrieved from the knowledge network using the multi-dimensional context index to generate a priority root cause candidate set;

[0009] Action execution module: used to execute corresponding automated repair actions based on the root cause candidate set and according to the preset management strategy.

[0010] Furthermore, the steps to form a context-aware data stream carrying management intent include:

[0011] Application data is extracted from the application layer, infrastructure data from the infrastructure layer, and network data from the network layer through the data acquisition interface;

[0012] A unified tracking identifier is generated for each data record through a predefined management strategy rule base, and a unique string is generated through string concatenation operations;

[0013] Inject a management context containing business semantics;

[0014] Based on the acquired network big data, a context-aware data stream carrying a unified tracking identifier and management context is formed;

[0015] Set the sampling frequency parameters for network big data to form a complete context-aware data stream.

[0016] Furthermore, the steps to construct a multi-dimensional contextual index, forming a knowledge network that can be queried based on business intent, include:

[0017] Context-aware data flow through data aggregation pipelines;

[0018] Extract business semantic tags by using the business scenario field, service criticality level field, and compliance requirement field in the management context;

[0019] Construct a multi-dimensional context index, link tags with storage locations through dynamic association operations to form a knowledge network, and calculate edge weights and connect nodes through association construction algorithms;

[0020] Set data aggregation frequency parameters, define algorithm parameters, and form a multi-dimensional contextual index and knowledge network.

[0021] Furthermore, when an alarm or query request is received, the steps to parse it into a management intent query include:

[0022] Receive the query request text through the input interface and decompose the query request text into components;

[0023] Extract intent descriptions from components to generate management intent queries;

[0024] Verify the integrity of the intent query and set the receiving frequency parameters to generate a management intent query.

[0025] Furthermore, the steps for generating a priority-based root cause candidate set include:

[0026] Based on the obtained management intent query and multi-dimensional context index, the knowledge network is traversed, and nodes and edges are filtered using the constraints of the management intent query.

[0027] Retrieve associated context-aware data streams and access specific data storage locations by following dynamic association pointers;

[0028] Collect the retrieved context-aware data stream collection;

[0029] Perform algorithmic analysis to identify potential root causes, which are the source nodes of anomalous patterns in the data stream;

[0030] Calculate the priority score for each potential root cause, sort the list of potential root causes based on the priority scores, and arrange them in descending order of priority scores;

[0031] Output the sorted list as a priority set of root cause candidates.

[0032] Furthermore, based on the preset management strategy, the steps for executing corresponding automated repair actions include:

[0033] Input a priority set of root cause candidates and a preset management strategy;

[0034] Match the highest priority root cause in the root cause candidate set with the conditions of the management strategy;

[0035] Select the corresponding automated remediation action, prepare the execution parameters, which extract the specific target and scope from the root cause candidate set; trigger the automated remediation action.

[0036] Monitor the execution process and record action logs and status changes;

[0037] After execution is complete, update the system status.

[0038] Furthermore, the steps of the association construction algorithm to calculate edge weights and connect nodes include:

[0039] Define a multidimensional similarity calculation function to perform semantic vectorization processing on each business semantic tag string, mapping the tag string into a high-dimensional semantic vector;

[0040] The cosine similarity between semantic vectors is calculated as a semantic similarity component, and the formula is as follows:

[0041] ;

[0042] In the formula, For semantic similarity components, and These represent the semantic vectors of the two labels respectively;

[0043] Extract the timestamp sequence of the tag-related data and calculate the time overlap component. The calculation formula is as follows:

[0044] ;

[0045] In the formula, For time overlap components, and These represent the time interval sets corresponding to the data of the two labels;

[0046] Construct a label co-occurrence matrix and calculate the structural correlation components using the following formula:

[0047] ;

[0048] In the formula, For structurally related components, This indicates the number of queries in which both tags appear together. and Each represents the number of queries that occur independently.

[0049] The weighted edge weights are calculated using a weighted geometric mean, and the formula is as follows:

[0050] ;

[0051] In the formula, To incorporate edge weights, , , These are configurable weight coefficients for semantic, temporal, and structural components, respectively.

[0052] The dynamic connection threshold is obtained from the historical database. When the total edge weight is not less than the dynamic connection threshold, a connection edge is established and the weight value is stored in the edge attribute.

[0053] Furthermore, the steps for calculating the priority score for each potential root cause include:

[0054] Construct a multi-dimensional evaluation system, including four evaluation dimensions: service criticality, scope of impact, intensity of anomalies, and depth of propagation;

[0055] Service criticality is determined by obtaining a standardized score from the service level agreement database, with a value ranging from 0 to 1.

[0056] The scope of influence is calculated using a modified PageRank algorithm, and the formula is as follows:

[0057] ;

[0058] In the formula, For affected components Keyness score, Root cause to component Topological distance;

[0059] Anomaly intensity is calculated through the fusion of multiple indicators, and the calculation formula is as follows:

[0060] ;

[0061] In the formula, For abnormal frequency coefficients, This is the coefficient for the duration of the anomaly. This is the abnormal amplitude coefficient;

[0062] The propagation depth is calculated using the attenuation factor model, and the formula is as follows:

[0063] ;

[0064] In the formula, As the attenuation factor, For the first The number of nodes affected in the layer;

[0065] The final priority score is calculated using a nonlinear combination function, and the formula is as follows:

[0066] ;

[0067] In the formula, As an environmental complexity correction factor, It is a natural constant. To ensure service criticality, For the scope of influence, It is of abnormal strength. To increase the depth of dissemination.

[0068] This application provides a computer intelligent service management method based on network big data, applied to the computer intelligent service management system based on network big data, including the following steps:

[0069] Acquire network big data, inject unified tracking identifiers and management contexts containing business semantics according to predefined management strategies, and form a context-aware data stream carrying management intent;

[0070] The context-aware data streams are aggregated, and a multi-dimensional context index is constructed based on the management context to form a knowledge network that can be queried for business intent.

[0071] When an alarm or query request is received, it is parsed as a management intent query. The multidimensional context index is used to retrieve the associated context-aware data stream from the knowledge network to generate a priority root cause candidate set.

[0072] Based on the root cause candidate set and in accordance with the preset management strategy, corresponding automated repair actions are performed.

[0073] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages:

[0074] By injecting unified tracking identifiers and business semantic management contexts to form context-aware data streams, the benefits of improved service management intelligence and fault response speed are achieved. Furthermore, in the knowledge network construction process, the use of multi-dimensional context indexing and association construction algorithms to dynamically associate data tags enables efficient and accurate data query and association analysis capabilities. Moreover, in the root cause candidate set generation process, the introduction of a multi-dimensional evaluation system and a non-linear priority calculation model achieves accurate identification of core fault root causes, effectively avoiding misoperation and delayed processing, thereby ensuring the stability and reliability of the service system. Attached Figure Description

[0075] Figure 1 A schematic diagram of the structure of a computer intelligent service management system based on network big data provided in an embodiment of this application;

[0076] Figure 2 A flowchart illustrating computer intelligent service management based on network big data, provided for embodiments of this application. Detailed Implementation

[0077] This application provides a computer intelligent service management system and method based on network big data, which solves the problems of weak correlation of operation and maintenance data, low efficiency of fault location and slow manual response in the prior art. By constructing a context-aware data stream injected with business semantics, establishing a multi-dimensional context index knowledge network, and implementing an automated repair mechanism based on priority evaluation, it achieves the technical effects of intelligent correlation analysis of operation and maintenance data, accurate root cause location and system self-healing capability.

[0078] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.

[0079] like Figure 1 As shown, this application embodiment provides a computer intelligent service management system based on network big data, including: a data stream generation module: used to acquire network big data, the network big data including application data, infrastructure data, and network data, and inject a unified tracking identifier and a management context containing business semantics, such as: the business scenario, service criticality level, and compliance requirements, according to a predefined management strategy, to form a context-aware data stream carrying management intent;

[0080] Knowledge network generation module: used to aggregate the context-aware data stream and, based on the management context, construct a multi-dimensional context index that dynamically associates business semantic tags with specific data storage locations, forming a knowledge network that can be queried by business intent;

[0081] Candidate set generation module: When an alarm or query request is received, it is used to parse it into a management intent query, retrieve the associated context-aware data stream from the knowledge network using the multi-dimensional context index, and perform algorithm analysis to generate a priority root cause candidate set;

[0082] Action execution module: Based on the root cause candidate set and according to preset management strategies, such as "prioritizing core business", it executes corresponding automated repair actions, such as resource expansion, service restart, and traffic scheduling.

[0083] Furthermore, the steps to form a context-aware data stream carrying management intent include:

[0084] Application data is extracted from the application layer, infrastructure data from the infrastructure layer, and network data from the network layer through the data acquisition interface. The application data includes user interaction logs and business operation records, the infrastructure data includes server resource utilization and storage capacity indicators, and the network data includes transmission latency and bandwidth usage.

[0085] A unified tracking identifier is generated for each data record through a predefined management policy rule base. This tracking identifier is composed of a timestamp sequence and a source address, and a unique string is generated through string concatenation operations.

[0086] Inject a management context containing business semantics. The management context consists of a business scenario field, a service criticality level field, and a compliance requirement field. The business scenario field extracts keywords from the data content and maps them to a preset scenario classification table through a pattern matching algorithm. The service criticality level field determines the level value based on the business scenario field through a priority assignment algorithm. The compliance requirement field verifies that the data conforms to preset specifications and adds corresponding tags through a compliance check algorithm.

[0087] Based on the acquired network big data, a data stream processing pipeline is used during the injection process to sequentially pass the network big data into the identifier generation module and the management context injection module, forming a context-aware data stream carrying a unified tracking identifier and management context. This data stream is stored in a structured format, including header fields storing the tracking identifier and context, and body fields storing the original data content.

[0088] The sampling frequency parameters for network big data are set to control the collection interval of the data acquisition interface, ensuring continuous generation of data streams. Algorithm parameters are defined, including threshold parameters for the pattern matching algorithm for adjusting keyword extraction accuracy, weight parameters for the priority assignment algorithm for level calculation, and rule set parameters for the compliance check algorithm for standard verification. These parameters ensure that the injection process is executed automatically, forming a complete context-aware data stream.

[0089] Furthermore, the steps to construct a multi-dimensional contextual index, forming a knowledge network that can be queried based on business intent, include:

[0090] The data stream is context-aware through a data aggregation pipeline that uses a batch processing mode to input the data stream in segments.

[0091] Business semantic tags are extracted by the business scenario field, service criticality level field, and compliance requirement field in the management context. This extraction is achieved through a tag generation algorithm, which uses a field parsing function to decompose the context and applies a mapping table to generate tag strings.

[0092] A multi-dimensional contextual index is constructed, which consists of a hash table structure. Each index entry has a business semantic label as its key and a list of pointers to specific data storage locations as its values. The labels are linked to the storage locations through dynamic association operations, which include key-value pair insertion and update functions, forming a knowledge network. This knowledge network is represented by a graph structure, where nodes are business semantic labels and edges are the associations between labels. An association construction algorithm is used to calculate edge weights and connect nodes. This algorithm uses a similarity calculation function based on label string comparison.

[0093] Set the data aggregation frequency parameter to control the processing interval of the aggregation pipeline and ensure timely index updates. Define algorithm parameters, including the parsing depth parameter of the tag generation algorithm for field decomposition control, the conflict resolution parameter of the dynamic association operation for hash table maintenance, and the similarity threshold parameter of the association construction algorithm for edge connection decisions, forming a multi-dimensional contextual index and knowledge network that supports subsequent query operations.

[0094] Furthermore, when an alarm or query request is received, the steps to parse it into a management intent query include:

[0095] The system receives query request text via an input interface, which processes incoming data using a standardized protocol.

[0096] The application parsing algorithm decomposes the query request text into components. This algorithm includes a tokenization function to segment the text into phrases and a natural language processing module to identify intent entities.

[0097] The intent description is extracted from the component, which is achieved through an entity recognition algorithm, where the entity recognition algorithm uses a pre-trained model to match phrases to an intent category table; a management intent query is generated, which consists of intent categories and a parameter list, and is assembled through a query construction function;

[0098] Verify the integrity of the intent query. This verification uses an integrity check algorithm to scan the parameter list and confirm that all required fields exist. It also sets the receiving frequency parameter to form a management intent query, which is used for subsequent retrieval operations.

[0099] Furthermore, the steps for generating a priority-based root cause candidate set include:

[0100] Based on the obtained management intent query and multi-dimensional context index, the knowledge network is traversed, and nodes and edges are filtered using the constraints of the management intent query.

[0101] Retrieve associated context-aware data streams and access specific data storage locations by following dynamic association pointers;

[0102] Collect the retrieved context-aware data stream collection;

[0103] The algorithm analysis is performed. This algorithm is a graph traversal algorithm combined with a causal reasoning model. It performs pattern matching and dependency calculation on the context-aware data stream set to identify potential root causes. The root cause is the source node of the abnormal pattern in the data stream.

[0104] Calculate the priority score for each potential root cause, determine it based on the service criticality level and scope of impact using a weighted summation formula, and sort the potential root causes list based on the priority scores in descending order.

[0105] Output the sorted list as a priority set of root cause candidates.

[0106] Furthermore, based on the preset management strategy, the steps for executing corresponding automated repair actions include:

[0107] Input a priority set of root cause candidates and a preset management strategy, which is a set of rules including condition-action pairs;

[0108] The highest priority root cause in the root cause candidate set is matched with the conditions of the management strategy through logical comparison operations.

[0109] Select the corresponding automated repair action, which includes resource expansion, service restart, or traffic scheduling; prepare the execution parameters, which extract specific targets and scopes from the root cause candidate set; trigger the automated repair action by calling the system interface to achieve resource expansion, service restart, or traffic scheduling.

[0110] Monitor the execution process and record action logs and status changes;

[0111] After execution, update the system status to ensure that the repair actions have been applied to the target components, so that the automated repair actions strictly correspond to the analysis results of the root cause candidate set.

[0112] Furthermore, the steps of the association construction algorithm to calculate edge weights and connect nodes include:

[0113] Define a multidimensional similarity calculation function. This function takes two business semantic label strings and their associated spatiotemporal metadata as input parameters, performs semantic vectorization processing on each business semantic label string, and maps the label string to a high-dimensional semantic vector.

[0114] The cosine similarity between semantic vectors is calculated as a semantic similarity component, and the formula is as follows:

[0115] ;

[0116] In the formula, For semantic similarity components, and These represent the semantic vectors of the two labels respectively;

[0117] Extract the timestamp sequence of the tag-related data and calculate the time overlap component. The calculation formula is as follows:

[0118] ;

[0119] In the formula, For time overlap components, and These represent the time interval sets corresponding to the data of the two labels;

[0120] Construct a label co-occurrence matrix and calculate the structural correlation components using the following formula:

[0121] ;

[0122] In the formula, For structurally related components, This indicates the number of queries in which both tags appear together. and Each represents the number of queries that occur independently.

[0123] The weighted edge weights are calculated using a weighted geometric mean, and the formula is as follows:

[0124] ;

[0125] In the formula, To incorporate edge weights, , , These are configurable weight coefficients for semantic, temporal, and structural components, respectively.

[0126] The dynamic connection threshold is obtained from the historical database. This threshold is adaptively adjusted according to the network density. When the overall edge weight is not less than the dynamic connection threshold, a connection edge is established and the weight value is stored in the edge attribute.

[0127] Furthermore, the steps for calculating the priority score for each potential root cause include:

[0128] Construct a multi-dimensional evaluation system, including four evaluation dimensions: service criticality, scope of impact, intensity of anomalies, and depth of propagation;

[0129] Service criticality is determined by obtaining a standardized score from the service level agreement database, with a value ranging from 0 to 1.

[0130] The scope of influence is calculated using a modified PageRank algorithm, and the formula is as follows:

[0131] ;

[0132] In the formula, For affected components Keyness score, Root cause to component Topological distance;

[0133] Anomaly intensity is calculated through the fusion of multiple indicators, and the calculation formula is as follows:

[0134] ;

[0135] In the formula, For abnormal frequency coefficients, This is the coefficient for the duration of the anomaly. This is the abnormal amplitude coefficient;

[0136] The propagation depth is calculated using the attenuation factor model, and the formula is as follows:

[0137] ;

[0138] In the formula, As the attenuation factor, For the first The number of nodes affected in the layer;

[0139] The final priority score is calculated using a nonlinear combination function, and the formula is as follows:

[0140] ;

[0141] In the formula, This is an environmental complexity correction factor, calculated using real-time system load and network topology complexity. It is a natural constant. To ensure service criticality, For the scope of influence, It is of abnormal strength. To increase the depth of dissemination.

[0142] like Figure 2 As shown, this application provides a computer intelligent service management method based on network big data, applied to the computer intelligent service management system based on network big data, including the following steps: acquiring network big data, injecting a unified tracking identifier and a management context containing business semantics according to a predefined management strategy, forming a context-aware data stream carrying management intent;

[0143] The context-aware data streams are aggregated, and a multi-dimensional context index is constructed based on the management context to form a knowledge network that can be queried for business intent.

[0144] When an alarm or query request is received, it is parsed as a management intent query. The multidimensional context index is used to retrieve the associated context-aware data stream from the knowledge network to generate a priority root cause candidate set.

[0145] Based on the root cause candidate set and in accordance with the preset management strategy, corresponding automated repair actions are performed.

[0146] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0147] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0148] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0149] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0150] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0151] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A computer intelligent service management system based on network big data, characterized in that, include: Data stream generation module: used to acquire network big data, inject unified tracking identifiers and management contexts containing business semantics according to predefined management strategies, and form context-aware data streams carrying management intentions; The steps to create a context-aware data stream carrying management intent include: Application data is extracted from the application layer, infrastructure data from the infrastructure layer, and network data from the network layer through the data acquisition interface; A unified tracking identifier is generated for each data record through a predefined management strategy rule base, and a unique string is generated through string concatenation operations; Inject a management context containing business semantics; Based on the acquired network big data, a context-aware data stream carrying a unified tracking identifier and management context is formed; Set the sampling frequency parameters for network big data to form a complete context-aware data stream; Knowledge network generation module: used to aggregate the context-aware data stream and, based on the management context, construct a multi-dimensional context index to form a knowledge network that can be queried by business intent; Candidate set generation module: When an alarm or query request is received, it is parsed as a management intent query, and the associated context-aware data stream is retrieved from the knowledge network using the multi-dimensional context index to generate a priority root cause candidate set; The steps for generating a priority-based root cause candidate set include: Based on the obtained management intent query and multi-dimensional context index, the knowledge network is traversed, and nodes and edges are filtered using the constraints of the management intent query. Retrieve associated context-aware data streams and access specific data storage locations by following dynamic association pointers; Collect the retrieved context-aware data stream collection; Perform algorithmic analysis to identify potential root causes, which are the source nodes of anomalous patterns in the data stream; Calculate the priority score for each potential root cause, sort the list of potential root causes based on the priority scores, and arrange them in descending order of priority scores; Output the sorted list as a priority-based root cause candidate set; The steps for calculating the priority score for each potential root cause include: Construct a multi-dimensional evaluation system, including four evaluation dimensions: service criticality, scope of impact, intensity of anomalies, and depth of propagation; Service criticality is determined by obtaining a standardized score from the service level agreement database, with a value ranging from 0 to 1. The scope of influence is calculated using a modified PageRank algorithm, and the formula is as follows: ； In the formula, For affected components Keyness score, Root cause to component Topological distance; Anomaly intensity is calculated through the fusion of multiple indicators, and the calculation formula is as follows: ； In the formula, For abnormal frequency coefficients, This is the coefficient for the duration of the anomaly. This is the abnormal amplitude coefficient; The propagation depth is calculated using the attenuation factor model, and the formula is as follows: ； In the formula, As the attenuation factor, For the first The number of nodes affected in the layer; The final priority score is calculated using a nonlinear combination function, and the formula is as follows: ； In the formula, This is a correction factor for environmental complexity. It is a natural constant. To ensure service criticality, For the scope of influence, It is of abnormal strength. For the depth of dissemination; Action execution module: used to execute corresponding automated repair actions based on the root cause candidate set and according to the preset management strategy.

2. The computer intelligent service management system based on network big data as described in claim 1, characterized in that, The steps to build a multidimensional contextual index and form a knowledge network that can be queried for business intent include: Context-aware data flow through data aggregation pipelines; Extract business semantic tags by using the business scenario field, service criticality level field, and compliance requirement field in the management context; Construct a multi-dimensional context index, link tags with storage locations through dynamic association operations to form a knowledge network, and calculate edge weights and connect nodes through association construction algorithms; Set data aggregation frequency parameters, define algorithm parameters, and form a multi-dimensional contextual index and knowledge network.

3. The computer intelligent service management system based on network big data as described in claim 1, characterized in that, When an alarm or query request is received, the steps to parse it into a management intent query include: Receive the query request text through the input interface and decompose the query request text into components; Extract intent descriptions from components to generate management intent queries; Verify the integrity of the intent query and set the receiving frequency parameters to generate a management intent query.

4. The computer intelligent service management system based on network big data as described in claim 1, characterized in that, According to the preset management strategy, the steps for performing corresponding automated repair actions include: Input a priority set of root cause candidates and a preset management strategy; Match the highest priority root cause in the root cause candidate set with the conditions of the management strategy; Select the corresponding automated remediation action, prepare the execution parameters, which extract the specific target and scope from the root cause candidate set; trigger the automated remediation action. Monitor the execution process and record action logs and status changes; After execution is complete, update the system status.

5. The computer intelligent service management system based on network big data as described in claim 2, characterized in that, The steps of the association construction algorithm to calculate edge weights and connect nodes include: Define a multidimensional similarity calculation function to perform semantic vectorization processing on each business semantic tag string, mapping the tag string into a high-dimensional semantic vector; The cosine similarity between semantic vectors is calculated as a semantic similarity component, and the formula is as follows: ； In the formula, For semantic similarity components, and These represent the semantic vectors of the two labels respectively; Extract the timestamp sequence of the tag-related data and calculate the time overlap component. The calculation formula is as follows: ； In the formula, For time overlap components, and These represent the time interval sets corresponding to the data of the two labels; Construct a label co-occurrence matrix and calculate the structural correlation components using the following formula: ； In the formula, For structurally related components, This indicates the number of queries in which both tags appear together. and Each represents the number of queries that occur independently. The weighted edge weights are calculated using a weighted geometric mean, and the formula is as follows: ； In the formula, To incorporate edge weights, , , These are configurable weight coefficients for semantic, temporal, and structural components, respectively. The dynamic connection threshold is obtained from the historical database. When the total edge weight is not less than the dynamic connection threshold, a connection edge is established and the weight value is stored in the edge attribute.

6. A computer intelligent service management method based on network big data, applied to the computer intelligent service management system based on network big data as described in any one of claims 1-5, characterized in that, Includes the following steps: Acquire network big data, inject unified tracking identifiers and management contexts containing business semantics according to predefined management strategies, and form a context-aware data stream carrying management intent; The context-aware data streams are aggregated, and a multi-dimensional context index is constructed based on the management context to form a knowledge network that can be queried for business intent. When an alarm or query request is received, it is parsed as a management intent query. The multidimensional context index is used to retrieve the associated context-aware data stream from the knowledge network to generate a priority root cause candidate set. Based on the root cause candidate set and in accordance with the preset management strategy, corresponding automated repair actions are performed.

Images