Network service quality risk index extraction method and device, equipment and storage medium

Abstract

The invention discloses a network service quality risk index extraction method and device, equipment and a storage medium. According to the scheme, for a target service link hierarchical structure, service quality indexes of all service hierarchies are obtained, risk analysis is carried out, and risk index mapping data of all the service quality indexes are determined. And constructing a directed acyclic graph containing the service quality indexes and the risk indexes according to the risk index mapping data of each service quality index. And determining the risk influence value of each service quality index based on the directed acyclic graph and a preset risk influence value calculation model. And carrying out risk index analysis according to the risk influence value corresponding to each service quality index, and determining a service quality risk index of the target service. Through full-chain business hierarchical modeling and graph structure dynamic evaluation, a complete process from index acquisition to risk index extraction is realized, key risk indexes highly associated with business scenes are accurately acquired, and risk evaluation pertinence and actual effectiveness of index extraction are greatly improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of communication network, in particular to a network service quality risk indicator extraction method and device, equipment and storage medium. BACKGROUND

[0002] Network service quality is a key indicator for measuring the differentiated service capability of different services in a communication network. The core goal is to dynamically allocate network resources according to service types and user needs, thereby ensuring that the performance of critical services meets the standards. Therefore, in the current network service quality evaluation method, the comprehensive service quality is usually taken as the evaluation target, and the evaluation of service risk is often less targeted and adaptable.

[0003] However, service quality risk is still an indispensable part of network service quality. In order to ensure the accurate evaluation of the overall service instruction, the evaluation of service quality risk is gradually included in the evaluation system of network service quality. The problem caused by this is that when network service risk indicators need to be extracted for evaluation, these extracted network service quality risk indicators are difficult to adapt to different services, different scenarios, and even different processing stages in the whole chain of business. The current network service risk indicators have the problems of poor adaptability and insufficient application ability. SUMMARY

[0004] Based on the above problems, the present application provides a network service quality risk indicator extraction method, device, equipment and storage medium, which aims to extract network service quality risk indicators that adapt to different scenarios and processing stages in the whole chain of business.

[0005] The embodiments of the present application disclose the following technical solutions:

[0006] The first aspect of the present application provides a network service quality risk indicator extraction method, which comprises:

[0007] For each service layer in the target service link hierarchical structure, obtain the service quality indicators of each service layer;

[0008] Perform risk analysis on the service quality indicators of each service layer respectively, and determine the risk indicator mapping data of each service quality indicator; the risk indicator mapping data is used to represent the mapping relationship between the service quality indicator and at least one risk indicator;

[0009] According to the risk indicator mapping data of each service quality indicator, a directed acyclic graph composed of multiple service quality indicators and multiple risk indicators is constructed;

[0010] Based on the directed acyclic graph and a preset risk impact value calculation model, the risk impact value of each service quality indicator is determined.

[0011] According to the risk influence value corresponding to each service quality indicator, a risk indicator analysis is performed to determine a service quality risk indicator of the target business.

[0012] In an optional implementation, the risk analysis on each service quality indicator of each service layer is performed to determine risk indicator mapping data of each service quality indicator, including:

[0013] Based on a preset correlation variable parameter, a risk indicator mapping relationship of each service quality indicator is determined; the risk indicator mapping relationship is used to represent a risk indicator having a mapping relationship with the service quality indicator; and the preset correlation variable parameter is used to represent the risk indicator associated with the service quality indicator.

[0014] An influence weight of each risk indicator mapping relationship is determined by performing an influence factor analysis on each risk indicator mapping relationship; and the influence weight is used to represent an influence degree of the service quality indicator on the corresponding risk indicator in the risk indicator mapping relationship.

[0015] The risk indicator mapping relationship and the corresponding influence weight of each service quality indicator are determined as the risk indicator mapping data corresponding to the service quality indicator alone, to obtain the risk indicator mapping data of each service quality indicator.

[0016] In an optional implementation, the nodes of the directed acyclic graph include a plurality of indicator nodes and a plurality of risk nodes, a single indicator node corresponds to a single service quality indicator, and a single risk node corresponds to a single risk indicator.

[0017] The risk indicator mapping data of each service quality indicator is used to construct a directed acyclic graph composed of a plurality of service quality indicators and a plurality of risk indicators, including:

[0018] According to the risk indicator mapping relationship of each service quality indicator, a causal connection relationship between each indicator node and each risk node is constructed; and the causal connection relationship is used to represent a causal relationship and an influence direction between the indicator node and the risk node.

[0019] The influence weight corresponding to each risk indicator mapping relationship alone is used as a carrying weight of each causal relationship.

[0020] Based on the causal connection relationship between a plurality of indicator nodes and a plurality of risk nodes, and the carrying weight of each causal connection relationship, the directed acyclic graph is constructed.

[0021] In an optional implementation, the determining of the risk influence value of each service quality indicator based on the directed acyclic graph and the preset risk influence value calculation model comprises:

[0022] determining, based on the causal connection relationship between each indicator node and each risk node, an associated indicator node corresponding to each indicator node and a relative value of the indicator node and the corresponding associated indicator node; the indicator node and the corresponding associated indicator node point to the same risk node; the relative value is determined based on a current measured value of each indicator node and a preset reference threshold value;

[0023] For each indicator node, the risk influence value is determined based on the relative value of the indicator node and the corresponding associated indicator node and the carrying weight of the corresponding causal connection relationship through the preset risk influence value calculation model, so as to obtain the risk influence value of each service quality indicator.

[0024] In an optional implementation, the risk indicator analysis based on the risk influence value of each service quality indicator to determine the service quality risk indicator of the target service comprises:

[0025] Based on the risk influence value of each service quality indicator, the service quality indicators with a ranking order greater than a preset threshold value are determined as the service quality risk indicators.

[0026] In an optional implementation, after the determination of the service quality risk indicator of the target service, the method further comprises:

[0027] Based on the service quality risk indicator and a preset historical training sample, the preset risk influence value calculation model is optimized.

[0028] The second aspect of the present application provides a network service quality risk indicator extraction device, which comprises:

[0029] An indicator acquisition module is configured to acquire service quality indicators of each service layer in a target service link hierarchical structure.

[0030] A risk mapping module is configured to perform risk analysis on the service quality indicators of each service layer respectively, and determine risk indicator mapping data of each service quality indicator; the risk indicator mapping data is used to represent a mapping relationship between the service quality indicator and at least one risk indicator.

[0031] A graph construction module is configured to construct a directed acyclic graph composed of a plurality of service quality indicators and a plurality of risk indicators based on the risk indicator mapping data of each service quality indicator.

[0032] An impact evaluation module is configured to determine a risk impact value of each service quality indicator based on the directed acyclic graph and a preset risk impact value calculation model.

[0033] A risk analysis module is configured to perform risk indicator analysis on each service quality indicator according to the risk impact value corresponding to the service quality indicator, and determine a service quality risk indicator of the target service.

[0034] In an optional implementation, the risk mapping module comprises:

[0035] A mapping relationship determination unit is configured to determine a risk indicator mapping relationship of each service quality indicator based on a preset correlation variable parameter; the risk indicator mapping relationship is used to represent a risk indicator that has a mapping relationship with the service quality indicator; and the preset correlation variable parameter is used to represent the risk indicator associated with the service quality indicator.

[0036] An impact weight determination unit is configured to perform impact factor analysis on each risk indicator mapping relationship, and determine an impact weight of each risk indicator mapping relationship; the impact weight is used to represent an influence degree of the service quality indicator on the corresponding risk indicator in the risk indicator mapping relationship.

[0037] A mapping data determination unit is configured to determine the risk indicator mapping relationship and the corresponding impact weight of each service quality indicator as risk indicator mapping data corresponding to the service quality indicator, to obtain the risk indicator mapping data of each service quality indicator.

[0038] The third aspect of the present application provides a network service quality risk indicator extraction device, which comprises a processor and a memory.

[0039] The memory is configured to store program code and transmit the program code to the processor.

[0040] The processor is configured to execute the steps of the network service quality risk indicator extraction method introduced in any implementation manner of the first aspect according to instructions in the program code.

[0041] The fourth aspect of the present application provides a computer readable storage medium for storing program code, which is used to execute the steps of the network service quality risk indicator extraction method introduced in any implementation manner of the first aspect.

[0042] Compared with the prior art, the present application has the following beneficial effects:

[0043] In the technical solution of the present application, firstly, the quality of service indicators of each service layer in the target service link hierarchical structure are acquired; then, the risk indicators of each service quality indicator are determined by respectively performing risk analysis on the service quality indicators of each service layer; thereafter, a directed acyclic graph composed of multiple service quality indicators and multiple risk indicators is constructed according to the risk indicator mapping data of each service quality indicator; then, the risk impact values of each service quality indicator are determined based on the directed acyclic graph and a preset risk impact value calculation model; finally, the risk indicators of the target service are determined by performing risk indicator analysis on the risk impact values corresponding to each service quality indicator. As can be seen, in the technical solution of the present application, a directed acyclic graph based on the service quality indicators and risk indicators of the whole-chain service layer is constructed, and dynamic evaluation is performed by using the graph structure and the risk impact value calculation model, thereby realizing a complete process from indicator collection to risk indicator extraction. The technical solution overcomes the defects of the poor adaptability of the traditional evaluation system, can accurately extract key risk indicators highly related to specific business scenarios, and significantly improves the pertinence of risk evaluation and the effectiveness of indicator extraction in practical applications. BRIEF DESCRIPTION OF DRAWINGS

[0044] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0045] Figure 1 A network service quality risk indicator extraction method flowchart provided by an embodiment of the present application;

[0046] Figure 2 A directed acyclic graph partial structure schematic diagram provided by an embodiment of the present application;

[0047] Figure 3 A network service quality risk indicator extraction device structure schematic diagram provided by an embodiment of the present application;

[0048] Figure 4 A risk mapping module structure schematic diagram provided by an embodiment of the present application. DETAILED DESCRIPTION

[0049] As described above, the current network service quality risk evaluation method generally faces comprehensive quality evaluation, and the index system thereof lacks dynamic correlation and scene adaptation capability, and it is difficult to accurately reflect the core risks in different links and different scenes in the whole-chain business, resulting in insufficient pertinence and practical application value of the evaluation results.

[0050] The inventors propose a network service quality risk indicator extraction method, device, equipment and storage medium.

[0051] First, for each service layer in the target service link hierarchical structure, the service quality indicators of each service layer are obtained. Then, the service quality indicators of each service layer are respectively analyzed for risk, and the risk indicator mapping data of each service quality indicator is determined. Then, according to the risk indicator mapping data of each service quality indicator, a directed acyclic graph composed of multiple service quality indicators and multiple risk indicators is constructed. Then, based on the directed acyclic graph and a preset risk impact value calculation model, the risk impact value of each service quality indicator is determined. Finally, according to the risk impact value corresponding to each service quality indicator, the risk indicator analysis is performed to determine the service quality risk indicator of the target service. In the technical scheme of the present application, by constructing a directed acyclic graph based on the service quality indicators and risk indicators of the whole chain service layer, and using the graph structure and the risk impact value calculation model for dynamic evaluation, the complete process from indicator collection to risk indicator extraction is realized. The defects of the traditional evaluation system are overcome, and the key risk indicators highly related to the specific service scene can be accurately extracted, which significantly improves the pertinence of risk evaluation and the effectiveness of indicator extraction in practical application.

[0052] In order to enable personnel in the technical field to better understand the present application, the technical solutions in the embodiments of the present application will be described clearly and completely in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0053] Referring to Figure 1 , the figure is a network service quality risk indicator extraction method flowchart provided by the present application. As Figure 1 shown, the method comprises the following steps:

[0054] S101, for each service layer in the target service link hierarchical structure, the service quality indicators of each service layer are obtained.

[0055] In the embodiments of the present application, the target service link hierarchical structure is a whole chain service model of Internet infrastructure service, which divides complex network services into multiple levels according to functional logic, including user access layer, network transmission layer, service bearing layer, data layer and operation and management layer. Each service layer represents an independent functional module in the service delivery process, and undertakes a specific service responsibility.

[0056] The service quality indicators are technical parameter items used to quantitatively evaluate the performance of each layer, including network performance indicators such as delay, jitter, packet loss rate, and throughput, and service availability indicators such as mean time to repair (MTTR) and service interruption time. These indicators (including names, units, etc.) are defined based on national standards and industry specifications, forming a complete indicator system that provides a conceptual basis and analysis object for subsequent risk assessment.

[0057] In the embodiments of the present application, the reference thresholds of these indicators are also defined at this step, for example, an end-to-end delay of more than 200 ms is high risk.

[0058] In an example implementation, the indicators can be systematically extracted based on national standards and industry specifications, such as the standard indicator items of the network transmission layer from the specification documents such as the "Guidelines for Quality of Service Evaluation of Telecommunication Networks and Internet", including end-to-end delay (unit: ms), jitter (unit: ms), packet loss rate (unit: %), throughput (unit: Mbps), etc.

[0059] In another example implementation, specific business indicators can be defined based on the standard indicators for different business types (such as video streaming, real-time communication, e-commerce, etc.). In the definition process, the name, definition, unit of measurement, and reference threshold of each indicator need to be clearly defined to form a structured indicator definition library.

[0060] The embodiments of the present application can cover all chain business links from user access to service bearing by comprehensively obtaining service quality indicators of each layer, ensuring the comprehensiveness and representativeness of risk assessment.

[0061] S102, respectively performing risk analysis on the service quality indicators of each business layer to determine risk indicator mapping data of each service quality indicator.

[0062] In the embodiments of the present application, risk analysis is a business impact deduction based on the service quality indicators obtained in S101. The purpose is to identify what negative consequences, i.e., risk indicators, may be caused by the abnormality of each service quality indicator.

[0063] The risk indicator mapping data is the output of risk analysis, which is a structured data used to represent the mapping relationship between the service quality indicators and at least one risk indicator, and can accurately represent which risk indicators and their impact degree may be caused by a certain service quality indicator.

[0064] In the embodiments of the present application, the mapping relationship can be constructed based on the expert knowledge base. Through analysis based on business experience, the risk indicators that can be triggered by each service quality indicator can be determined. The data mining technology can also be used to automatically learn the mapping relationship from the historical operation and maintenance data. Through analysis of the correlation between the abnormality of each indicator in the historical fault event and the occurrence of the risk event, the association rule mining algorithm (such as the Apriori algorithm) or the machine learning classification model is used to automatically identify the strong correlation between the indicators and the risks. This way can discover potential and unattended mapping relationships by manual experience, and improve the integrity of the mapping data.

[0065] Taking the network output layer as an example, Table 1 is a service quality indicator and risk indicator mapping data relationship table provided by the embodiments of the present application, which clearly shows the risk indicators and impact descriptions corresponding to part of the service quality indicators in the network transmission layer.

[0066] Table 1

[0067]

[0068] In an optional implementation, the specific steps of determining the risk indicator mapping data of each service quality indicator of the hierarchical service quality indicators of each business include:

[0069] S1021, determining the risk indicator mapping relationship of each service quality indicator based on the preset correlation variable parameters.

[0070] In the embodiments of the present application, the risk indicator mapping relationship is used to represent the risk indicators that have a mapping relationship with the service quality indicators. The preset correlation variable parameters are used to represent the risk indicators associated with the service quality indicators. These parameters are defined in advance based on the business knowledge base and related industry experience, and establish a semantic bridge from the technical indicators to the business risks.

[0071] In an example implementation, the mapping relationship shown in Table 1 can be referred to to establish the corresponding relationship between the service quality indicators and the risk indicators. For example, based on the preset correlation variable parameters, it is determined that the “end-to-end delay” indicator has a mapping relationship with the “user experience degradation” and “payment link failure” risk indicators, and the “BGP (Border Gateway Protocol) route hijacking times” indicator has a mapping relationship with the “data leakage” and “service unavailable” risk indicators. The establishment of this mapping relationship can be based on business impact analysis to ensure that each service quality indicator is associated with the core business risks that it can trigger.

[0072] S1022, performing impact factor analysis on each risk indicator mapping relationship to determine the impact weight of each risk indicator mapping relationship.

[0073] In the embodiments of the present application, the influence weight is used to represent the influence degree of the service quality indicator on the corresponding risk indicator in the risk indicator mapping relationship. The influence weight is a quantitative value, usually ranging from 0 to 1, and the greater the value, the higher the influence degree of the indicator on the corresponding risk.

[0074] In an example implementation, an experience-based scoring method can be used to determine the influence weight. Each mapping relationship is independently scored manually in combination with multiple sets of network service scene data, and then the final influence weight is obtained by weighted average. For example, for the mapping relationship of "end-to-end delay → user experience decline", the influence weight mean given by the user of the present method can be 0.6 based on historical experience and business importance; for the mapping relationship of "BGP route hijacking times → data leakage", the influence weight can be 0.95 due to the severity of the security threat.

[0075] In another example implementation, the correlation coefficient between the indicator anomaly and the risk event occurrence probability can be calculated through historical data analysis, and the normalized correlation coefficient can be used as the influence weight.

[0076] S1023, the risk indicator mapping relationship of each service quality indicator and the corresponding influence weight are determined as the risk indicator mapping data corresponding to the service quality indicator, to obtain the risk indicator mapping data of each service quality indicator.

[0077] In the embodiments of the present application, the risk indicator mapping data is a structured data object, which contains complete service quality indicators, mapped risk indicators and influence weights. These data are organized in a unified format to ensure consistency and standardization in subsequent processing.

[0078] In an example implementation, the risk indicator mapping data can be organized in JSON format. For example, for the "end-to-end delay" indicator, its risk indicator mapping data can be represented as:

[0079]

[0080] In another implementation, the mapping data can be stored in a relational database table structure, a service quality indicator table, a risk indicator table and a mapping relationship table are established, and complete mapping information is maintained through foreign key association. This structured data organization method facilitates subsequent query, update and expansion, and provides accurate data basis for the construction of directed acyclic graph.

[0081] S103, according to the risk indicator mapping data of each service quality indicator, a directed acyclic graph composed of multiple service quality indicators and multiple risk indicators is constructed.

[0082] A directed acyclic graph (DAG) is a special graph data structure composed of vertices and directed edges, and there is no loop in the graph.

[0083] In the embodiment of the present application, the nodes of the directed acyclic graph constructed in this step include a plurality of indicator nodes and a plurality of risk nodes, wherein a single indicator node represents a single service quality indicator, and a single risk node represents a single risk indicator. The directed edges are used to connect the indicator nodes and the risk nodes, indicating the causal relationship between the service quality indicators and the risk indicators, and the weight value carried on the edge represents the degree of influence.

[0084] In an optional implementation, according to the risk indicator mapping data of each service quality indicator, a directed acyclic graph composed of a plurality of service quality indicators and a plurality of risk indicators is constructed, specifically including the following steps:

[0085] S1031, according to the risk indicator mapping relationship of each service quality indicator, the causal connection relationship between each indicator node and each risk node is constructed.

[0086] In the embodiment of the present application, the causal connection relationship is used to represent the causal relationship and the influence direction between the indicator nodes and the risk nodes. It reflects the direct influence path of the service quality indicator abnormality on the business risk.

[0087] In an example implementation, the risk indicator mapping data generated in step S102 can be traversed to create a corresponding causal connection for each mapping relationship. For example, for the "end-to-end delay" indicator node, according to its risk indicator mapping relationship, two directed edges pointing to the "user experience decline" risk node and the "payment link failure" risk node are created. Similarly, for the "BGP route hijacking times" indicator node, directed edges pointing to the "data leakage" and "service unavailable" risk nodes are created.

[0088] S1032, the influence weight corresponding to each risk indicator mapping relationship is taken as the carrying weight of each causal relationship.

[0089] In the embodiment of the present application, the carrying weight is an important attribute of the causal connection relationship, which quantifies the influence strength of the indicator node on the risk node. The weight value directly comes from the influence weight in the risk indicator mapping data determined in S102, ensuring the consistency of the graph model parameters and the business knowledge.

[0090] In an example implementation, the corresponding impact weight in the mapping data can be set as the weight attribute of the directed edge when the directed edge is created. For example, when the edge of "end-to-end latency → user experience degradation" is established, the weight carried by it is set to 0.6; when the edge of "end-to-end latency → payment link failure" is established, the weight carried by it is set to 0.8. These weight values will directly participate in the subsequent risk propagation calculation and determine the transmission strength of the risk impact.

[0091] S1033, based on the causal connection relationship between the plurality of index nodes and the plurality of risk nodes, and the weight carried by each causal connection relationship, a directed acyclic graph is constructed.

[0092] In the embodiments of the present application, the construction of the directed acyclic graph needs to organize all nodes and edges according to a specific data structure, and ensure the connectivity and acyclicity of the graph. The finally formed graph structure contains complete risk propagation path information, which provides a calculation basis for complex risk assessment.

[0093] In an example implementation, a graph database (such as Neo4j) or a graph computing library (such as NetworkX) can be used to construct and store the directed acyclic graph. The construction process includes: first creating all index nodes and risk nodes, then adding directed edges according to the causal connection relationship established in S1031, and setting the weight carried by each edge determined in S1032. After the construction is completed, the acyclicity of the graph is verified using a topological sorting algorithm (such as Kahn algorithm), and for the possible loop, appropriate adjustment is needed, for example, by deleting or adjusting the weight of the edge to eliminate the loop.

[0094] Referring to Figure 2 , Figure 2 A directed acyclic graph partial structure diagram is provided in the embodiments of the present application. As shown in Figure 2 , the graph shows the causal relationship network between the index nodes and the risk nodes, where the circular nodes represent the service quality indicators, the rectangular nodes represent the risk indicators, and the numbers on the directed edges represent the weight carried. For example, the "end-to-end latency" index node affects the "user experience degradation" risk node through an edge with a weight of 0.6, and affects the "payment link failure" risk node through an edge with a weight of 0.8, forming a complete risk conduction path.

[0095] The embodiments of the present application reflect the complex risk correlation relationship through the visual expression of the directed acyclic graph, and provide an intuitive reference for the subsequent risk impact value calculation.

[0096] S104, based on the directed acyclic graph and the preset risk impact value calculation model, determining the risk impact value of each service quality indicator.

[0097] In the embodiments of the present application, the preset risk influence value calculation model is a mathematical model for quantifying the influence degree of the service quality indicators on the risk indicators, and the core formula is based on the node association relationship and parameter design in the directed acyclic graph.

[0098] The risk influence value is a comprehensive embodiment of the possibility and severity of the corresponding risk caused by the service quality indicators, and is the core basis for judging the risk level of the indicators.

[0099] In an optional implementation, based on the directed acyclic graph and the preset risk influence value calculation model, the risk influence values of the service quality indicators are determined, specifically including the following steps:

[0100] S1041, based on the causal connection relationship between each indicator node and each risk node, determining the corresponding associated indicator node of each indicator node, and the relative value of the indicator node and the corresponding associated indicator node.

[0101] In the embodiments of the present application, the indicator node and the corresponding associated indicator node point to the same risk node.

[0102] The relative value is determined based on the current measured value of each indicator node and the preset reference threshold value, and reflects the abnormality degree of the current state of the indicator.

[0103] In an example implementation, the identification of the associated indicator node can be realized by a graph traversal algorithm. Taking the "throughput" indicator node as an example, it points to the "service unavailable" risk node, and the "service interruption time" indicator node and the "BGP route hijacking times" indicator node also point to the risk node. Therefore, the associated indicator nodes corresponding to the "throughput" indicator node include the "service interruption time" indicator node and the "BGP route hijacking times" indicator node.

[0104] Taking the "jitter" indicator node as an example, it points to the "video lag" risk node, and the "packet loss rate" indicator node also points to the risk node. Therefore, the associated indicator nodes corresponding to the "jitter" indicator node include the "packet loss rate" indicator node.

[0105] The relative value can be determined by a threshold normalization method, for example, the current measured value of "end-to-end delay" is 280ms, and the preset reference threshold value is 200ms. Therefore, the relative value is calculated as min(1.0, 280 / 200)=1.0; the current value of "jitter" is 25ms, and the preset reference threshold value is 30ms. Therefore, the relative value is 25 / 30≈0.83; the current value of "packet loss rate" is 0.8%, and the preset reference threshold value is 1%. Therefore, the relative value is 0.8% / 1%=0.8.

[0106] S1042, for each index node, according to the relative value of the index node and the corresponding associated index node, and the carrying weight of the corresponding causal connection relationship, determining a risk influence value through a preset risk influence value calculation model to obtain the risk influence value of each service quality index.

[0107] In the embodiments of the present application, the preset risk influence value calculation model is a mathematical calculation framework that integrates graph propagation theory and risk aggregation algorithm. The model takes the relative value of the index node and the carrying weight of the edge as input, and calculates the influence degree of each node on the overall risk through a specific propagation algorithm. The calculation process considers various influence paths such as direct association and indirect transmission, ensuring the comprehensiveness and accuracy of the evaluation results.

[0108] In an example implementation, the risk influence value calculation model uses the following calculation formula:

[0109] ;

[0110] wherein, the risk influence value of the index node , is a set of all risk nodes directly pointed to by the index node , is a risk node in the set , is the carrying weight of the edge from the index node to the risk node , is the relative value of the index node compared to the threshold value, , is a set of all index nodes pointing to the risk node , is any one index node (i.e. associated index node) in the set except the index node being calculated, and are the carrying weight and relative value of the associated index node respectively.

[0111] For example, when calculating the risk impact value of the "BGP route hijacking count" indicator node, the risk nodes it points to include "Service Unavailability" with a weight of 0.9 and "Data Leakage" with a weight of 0.95. The relative value of the "BGP route hijacking count" indicator node is 0.8. Its associated indicator nodes include the "Throughput" indicator node and the "Service Interruption Time" indicator node, which are associated with the "Service Unavailability" risk node. The "Throughput" indicator node carries a weight of 0.7 towards the "Service Unavailability" risk node, with a relative value of 1.0, while the "Service Interruption Time" indicator node carries a weight of 0.9 towards the "Service Unavailability" risk node, with a relative value of 0.5. The formula for calculating the risk impact value of the "BGP route hijacking count" indicator node using the risk impact value calculation model is as follows:

[0112] RiskScore (Number of BGP route hijackings) = 0.95 × 0.8 + (0.9 × 0.8 + 0.7 × 1.0 + 0.9 × 0.5) = 2.63.

[0113] It is understood that the above-mentioned risk impact value calculation model is only an example risk impact value calculation model provided in the embodiments of this application. In practical applications, an appropriate risk impact value calculation model can be selected according to the complexity of the business scenario to achieve a balance between accuracy and performance.

[0114] This application's embodiments combine the risk impact value calculation logic of associated nodes and relative values, considering both the independent impact of single indicators and the synergistic effect of multiple indicators, thereby improving the comprehensiveness and accuracy of the calculation results and providing a more precise quantitative basis for risk indicator screening.

[0115] S105 conducts risk indicator analysis based on the risk impact value corresponding to each service quality indicator to determine the service quality risk indicators for the target business.

[0116] In this embodiment of the application, risk indicator analysis is the process of interpreting and making decisions based on the calculated risk impact values, with the aim of identifying the key risk indicators that have the greatest impact on business.

[0117] Service quality risk indicators refer to a small number of key indicators that require priority attention and handling after analysis and screening. These indicators typically have high risk impact values ​​and pose a significant threat to business stability and security. The analysis process includes numerical ranking, threshold screening, and result interpretation, ultimately outputting a list of risk indicators with clear guidance.

[0118] In the embodiments of this application, risk indicator analysis can be implemented using a variety of techniques.

[0119] In an example implementation, the top N indicators can be selected as the key risk indicators based on a ranking Top-N screening method, in descending order of risk impact values.

[0120] In another example implementation, the threshold of risk impact values can be set as a threshold gate for screening, and all indicators exceeding the threshold can be included in the key risk indicator set.

[0121] In practical applications, a hybrid strategy can be adopted in combination with business characteristics, such as a lower threshold for core businesses and a higher threshold for non-core businesses.

[0122] In the embodiments of the present application, the analysis results can be output in the form of a structured report, including detailed information such as indicator name, current value, risk score, and supporting decision-making needs at different levels.

[0123] In an example implementation, the analysis result output form is as follows:

[0124] { "service quality risk indicator list": [ {"indicator name": "end-to-end latency", "current value": "250ms", "risk score": 0.82}, {"indicator name": "BGP route hijacking times", "current value": "3 times / month", "risk score": 0.91} …… ].

[0125] In an optional implementation, determining the service quality risk indicators of the target business specifically includes:

[0126] Based on the numerical value of each service quality indicator, the risk impact value is sorted in descending order, and the service quality indicators with a ranking position greater than a preset threshold are determined as the service quality risk indicators.

[0127] In the embodiments of the present application, numerical value sorting is a process of arranging all service quality indicators in descending order according to the calculated risk impact values, aiming to identify the key indicators with the highest risk contribution.

[0128] The preset threshold is a judgment standard for screening key risk indicators, which can be a fixed ranking position (such as the top 10 ranking positions) or a dynamic risk score threshold (such as a score > 0.8).

[0129] The service quality risk indicators are a set of key indicators that need to be prioritized after sorting and screening, and these indicators have the greatest potential threat to business stability and security.

[0130] In the embodiments of the present application, the sorting and screening process can adopt both batch processing and stream processing. Batch processing sorts and screens the full amount of indicators periodically, which is suitable for scenarios with low real-time requirements. Stream processing realizes real-time sorting and threshold judgment based on a stream computing engine (such as Flink or Storm), and can timely discover sudden risk indicators. The threshold setting can adopt an adaptive mechanism in combination with business characteristics, and dynamically adjust the threshold standard according to business load, time period and other factors.

[0131] This step screens the mass indicators to the core risk indicators, and clearly defines the risk prevention and control focus. The network service quality risk indicators extracted in the embodiments of the present application can help the business party to reasonably allocate resources, avoid resource waste of comprehensive prevention and control, and provide clear objects for risk early warning and optimization adjustment, so that the control work is more targeted and efficient, and the stability and security of the target business service quality are ensured.

[0132] The embodiments of the present application build a service quality indicator system based on full-chain business layering; then an index and risk mapping relationship is established by pre-setting an associated variable parameter, a structured mapping data is formed by analyzing the influence weight; then a directed acyclic graph containing index nodes, risk nodes and weight causal edges is constructed, and the risk transmission path is intuitively presented; based on the graph and a pre-designed calculation model, the risk influence value of each index node is calculated in combination with the relative value of the index node and the influence of the associated node; finally, the core service quality risk indicators are determined through sorting and screening. The full-process closed loop from index acquisition, risk modeling to quantitative evaluation and index screening is realized, which not only avoids the risk evaluation blind area, but also accurately locks the high-risk link, realizes the extraction of network service quality risk indicators suitable for different scenarios and processing stages of full-chain business, provides a scientific decision basis for network service risk prevention and control, effectively improves the resource utilization efficiency, and ensures the stability and security of the full-chain business service quality.

[0133] Optionally, in order to improve the prediction accuracy of the risk influence value calculation model on the actual business risk and ensure the consistency of the risk influence value calculation result with the real risk scenario, after step S105, the network service quality risk indicator extraction method further includes:

[0134] Based on the service quality risk indicators and the pre-set historical training samples, the pre-set risk influence value calculation model is optimized.

[0135] In the embodiments of the present application, the pre-set historical training sample is a structured data set containing historical service quality indicator data, corresponding risk event records and actual influence degree, which covers indicator samples of different business layers and different abnormal degrees, and provides data support for model optimization.

[0136] In an example implementation, the model optimization can adopt an iterative parameter adjustment strategy: first, merge the historical data corresponding to the quality of service risk indicators (including the measured values of the indicators, the calculation results of the risk influence values, and the actual risk levels) with the preset historical training samples to construct an expanded training set; second, minimize the error between the model prediction value (risk influence value) and the label by using the gradient descent algorithm, dynamically adjust the distribution logic of the influence weight in the model, for example, for samples in the historical data where the actual influence of the "high-risk indicator" is underestimated, increase the weight coefficient of the corresponding correlation; finally, verify the generalization ability of the optimized model by using the cross-validation method, if the prediction accuracy on the validation set improves by more than 5%, save the optimized model parameters, otherwise, backtrack to the previous round of iterative parameters. By continuously incorporating new service quality risk indicator data for iterative optimization, the model can adapt to the dynamic changes of the business link and improve the timeliness and accuracy of risk assessment.

[0137] Based on the network service quality risk indicator extraction method provided in the foregoing embodiments, the application also provides a network service quality risk indicator extraction device. Figure 3 A structural schematic diagram of a network service quality risk indicator extraction device provided by an embodiment of the application is shown in FIG. 3. Figure 3 As shown in FIG. 3, the network service quality risk indicator extraction device includes an indicator acquisition module 301, a risk mapping module 302, a graph construction module 303, an influence evaluation module 304, and a risk analysis module 305.

[0138] The indicator acquisition module 301 is configured to acquire the service quality indicators of each business layer in the target business link hierarchical structure.

[0139] The risk mapping module 302 is configured to perform risk analysis on the service quality indicators of each business layer respectively, and determine the risk indicator mapping data of each service quality indicator; the risk indicator mapping data is used to represent the mapping relationship between the service quality indicator and at least one risk indicator.

[0140] The graph construction module 303 is configured to construct a directed acyclic graph composed of multiple service quality indicators and multiple risk indicators according to the risk indicator mapping data of each service quality indicator.

[0141] The influence evaluation module 304 is configured to determine the risk influence value of each service quality indicator based on the directed acyclic graph and a preset risk influence value calculation model.

[0142] The risk analysis module 305 is configured to perform risk indicator analysis according to the respective risk influence values of each service quality indicator, and determine the service quality risk indicators of the target business.

[0143] The embodiments of the present application realize the extraction of network service quality risk indexes suitable for different scenes and processing stages of the whole chain business by the mutual combination of the functions of the index acquisition module 301, the risk mapping module 302, the graph construction module 303, the influence evaluation module 304 and the risk analysis module 305.

[0144] In an optional implementation manner, Figure 4 A structural schematic diagram of a risk mapping module provided by the embodiments of the present application is shown in FIG. 4. Figure 4 As shown in FIG. 4, the sampling module includes a mapping relationship determination unit 401, an influence weight determination unit 402 and a mapping data determination unit 403.

[0145] The mapping relationship determination unit 401 is configured to determine risk index mapping relationships of each service quality index based on preset correlation variable parameters; the risk index mapping relationship is configured to represent a risk index having a mapping relationship with a service quality index; and the preset correlation variable parameter is configured to represent a risk index associated with the service quality index.

[0146] The influence weight determination unit 402 is configured to analyze influence factors of each risk index mapping relationship respectively, and determine influence weights of each risk index mapping relationship; and the influence weight is configured to represent an influence degree of the service quality index on the corresponding risk index in the risk index mapping relationship.

[0147] The mapping data determination unit 403 is configured to determine the risk index mapping relationship and the corresponding influence weight of each service quality index as risk index mapping data corresponding to the service quality index alone, so as to obtain risk index mapping data of each service quality index.

[0148] In an optional implementation manner, the nodes of the directed acyclic graph include a plurality of index nodes and a plurality of risk nodes, a single index node corresponds to a single service quality index, a single risk node corresponds to a single risk index, and the graph construction module 303 is specifically configured to:

[0149] construct a causal connection relationship between each index node and each risk node according to the risk index mapping relationship of each service quality index; the causal connection relationship is configured to represent a causal relationship and an influence direction between the index node and the risk node;

[0150] use the influence weight corresponding to each risk index mapping relationship as a carrying weight of each causal relationship;

[0151] construct the directed acyclic graph based on the causal connection relationship between the plurality of index nodes and the plurality of risk nodes, and the carrying weight of each causal connection relationship.

[0152] In an optional implementation manner, the influence evaluation module 304 is specifically configured to:

[0153] Based on the causal connection relationship between each index node and each risk node, determine the corresponding associated index nodes of each index node, and the relative values of the index nodes and the corresponding associated index nodes; the index nodes and the corresponding associated index nodes point to the same risk node; the relative values are determined based on the current measured values of each index node and the preset reference threshold;

[0154] For each index node, according to the relative values of the index nodes and the corresponding associated index nodes, and the carrying weight of the corresponding causal connection relationship, determine the risk influence value through the preset risk influence value calculation model to obtain the risk influence values of each service quality index.

[0155] In an optional implementation, the risk analysis module 305 is specifically configured to:

[0156] Based on the risk influence values of each service quality index, the numerical values are sorted, and the service quality indexes with a ranking order greater than a preset threshold are determined as service quality risk indexes.

[0157] In an optional implementation, the network service quality risk index extraction device further includes a model optimization module configured to perform model optimization on the preset risk influence value calculation model based on the service quality risk indexes and a preset historical training sample.

[0158] In addition, an embodiment of the present application also provides a network service quality risk index extraction device, which includes a processor and a memory.

[0159] The memory is configured to store program code and transmit the program code to the processor.

[0160] The processor is configured to execute the steps of the network service quality risk index extraction method introduced in any of the above method embodiments according to the instructions in the program code.

[0161] In addition, an embodiment of the present application also provides a computer readable storage medium, which stores a computer program, when the program is run by a processor, the network service quality risk index extraction method as introduced in any of the method embodiments is implemented.

[0162] It should be noted that each of the embodiments of the present specification is described in a progressive manner, and the same or similar parts between each embodiment can be understood by referring to each other, and each embodiment focuses on the difference from other embodiments. In particular, for the device and equipment embodiments, since they are basically similar to the method embodiments, they are described more simply, and the relevant parts can be understood by referring to the part of the method embodiment. The above-described device and equipment embodiments are only illustrative, and the units described as separate components can or can not be physically separated, and the components indicated as units can or can not be physical units, that is, they can be located in one place, or they can be distributed on multiple network units. Part or all of the modules can be selected to achieve the purpose of the embodiments of the present application according to actual needs. Those skilled in the art can understand and implement it without creative labor.

[0163] The above is only one specific embodiment of the present application, but the protection scope of the present application is not limited to this. Any skilled person in the art can easily think of changes or replacements within the technical scope disclosed in the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A network service quality risk indicator extraction method, characterized by, The method comprises the following steps: For each service layer in the target service link hierarchy, obtain the service quality indicators of each service layer; Perform risk analysis on the service quality indicators of each service layer respectively to determine the risk indicator mapping data of each service quality indicator; the risk indicator mapping data is used to represent the mapping relationship between the service quality indicator and at least one risk indicator; According to the risk indicator mapping data of each service quality indicator, a directed acyclic graph composed of multiple service quality indicators and multiple risk indicators is constructed; Based on the directed acyclic graph and a preset risk impact value calculation model, the risk impact value of each service quality indicator is determined; According to the risk impact value corresponding to each service quality indicator, risk indicator analysis is performed to determine the service quality risk indicator of the target service.

2. The method of claim 1, wherein, The risk analysis of each service layer service quality indicator and the determination of the risk indicator mapping data of each service quality indicator include: Based on a preset correlation variable parameter, the risk indicator mapping relationship of each service quality indicator is determined; the risk indicator mapping relationship is used to represent the risk indicators that have a mapping relationship with the service quality indicators; the preset correlation variable parameter is used to represent the risk indicators associated with the service quality indicators; The influence weight of each risk indicator mapping relationship is determined by analyzing the influence factors of each risk indicator mapping relationship; the influence weight is used to represent the influence degree of the service quality indicator on the corresponding risk indicator in the risk indicator mapping relationship; The risk indicator mapping relationship and the corresponding influence weight of each service quality indicator are determined as the risk indicator mapping data corresponding to the service quality indicator alone to obtain the risk indicator mapping data of each service quality indicator.

3. The method of claim 2, wherein, The nodes of the directed acyclic graph comprise multiple indicator nodes and multiple risk nodes, a single indicator node corresponds to a single service quality indicator, and a single risk node corresponds to a single risk indicator; According to the risk indicator mapping data of each service quality indicator, a directed acyclic graph composed of multiple service quality indicators and multiple risk indicators is constructed, which includes: According to the risk indicator mapping relationship of each service quality indicator, a causal connection relationship between each indicator node and each risk node is constructed; the causal connection relationship is used to represent the causal relationship and influence direction between the indicator node and the risk node; The influence weight corresponding to each risk indicator mapping relationship is used as the carrying weight of each causal relationship; Based on the causal connection relationship between multiple indicator nodes and multiple risk nodes and the carrying weight of each causal connection relationship, the directed acyclic graph is constructed.

4. The method of claim 3, wherein, The determination of the risk impact value of each service quality indicator based on the directed acyclic graph and the preset risk impact value calculation model includes: Determine, based on the causal connection relationship between each of the index nodes and each of the risk nodes, an associated index node corresponding to each of the index nodes, and a relative value of the index node and the corresponding associated index node; the index node and the corresponding associated index node point to the same risk node; the relative value is determined based on a current measured value of each index node and a preset reference threshold value; For each of the index nodes, determine the risk impact value based on the relative value of the index node and the corresponding associated index node, and the carrying weight of the corresponding causal connection relationship, through the preset risk impact value calculation model, to obtain the risk impact value of each of the service quality indicators.

5. The method of claim 1, wherein, The risk index analysis based on the risk impact value of each of the service quality indicators includes: Based on the risk impact value of each of the service quality indicators, perform numerical size sorting, and determine the service quality indicators with a ranking order greater than a preset threshold as the service quality risk indicators.

6. The method of claim 1, wherein, After determining the service quality risk indicators of the target service, the method further includes: Based on the service quality risk indicators and a preset historical training sample, perform model optimization on the preset risk impact value calculation model.

7. A network service quality risk indicator extraction apparatus characterized by comprising: Including: An index acquisition module is configured to acquire service quality indicators of each service layer in a target service link hierarchical structure; A risk mapping module is configured to perform risk analysis on the service quality indicators of each service layer respectively, and determine risk indicator mapping data of each service quality indicator; the risk indicator mapping data is used to represent a mapping relationship between the service quality indicator and at least one risk indicator; A graph construction module is configured to construct a directed acyclic graph composed of a plurality of service quality indicators and a plurality of risk indicators based on the risk indicator mapping data of each service quality indicator; An impact evaluation module is configured to determine a risk impact value of each service quality indicator based on the directed acyclic graph and a preset risk impact value calculation model; A risk analysis module is configured to perform risk index analysis based on the risk impact value corresponding to each of the service quality indicators, and determine the service quality risk indicators of the target service.

8. The apparatus of claim 7, wherein, The risk mapping module includes: A mapping relationship determination unit is configured to determine a risk indicator mapping relationship of each service quality indicator based on a preset associated variable parameter; the risk indicator mapping relationship is used to represent a risk indicator having a mapping relationship with the service quality indicator; the preset associated variable parameter is used to represent the risk indicators associated with the service quality indicators; An impact weight determination unit is configured to analyze influence factors of each of the risk indicator mapping relationships respectively, and determine an impact weight of each of the risk indicator mapping relationships; the impact weight is used to represent an influence degree of the service quality indicator on the corresponding risk indicator in the risk indicator mapping relationship. A mapping data determination unit is configured to determine the risk indicator mapping data corresponding to each service quality indicator, so as to obtain the risk indicator mapping data of each service quality indicator.

9. A network service quality risk indicator extraction apparatus characterized by comprising: Comprise: A processor and a memory: The memory is configured to store program code and transmit the program code to the processor; The processor is configured to execute the steps of the network service quality risk indicator extraction method according to the instructions in the program code.

10. A computer-readable storage medium, characterized in that, The computer readable storage medium is configured to store program code, and the program code is configured to execute the steps of the network service quality risk indicator extraction method. The computer readable storage medium is configured to store program code, and the program code is configured to execute the steps of the network service quality risk indicator extraction method.

Images