Method and device for determining quality difference root cause, and storage medium

Abstract

The application provides a quality difference root cause determination method and device and a storage medium, and is applied to the technical field of Internet. The method comprises the following steps: acquiring original service data carrying quality difference root cause labels, performing feature extraction on the original service data based on a preset feature extraction algorithm, and determining a corresponding feature vector set; determining the spatial relationship between base stations involved in the original service data based on the operator corresponding to the original service data, and constructing a spatial relationship graph of the base stations based on the spatial relationship; performing graph neural network training based on the feature vector set and the spatial relationship graph, determining a trained neural network model; determining model parameters based on the neural network model, performing aggregation calculation based on the model parameters to determine federated model parameters, updating the neural network model based on the federated model parameters, and determining a graph federated model for determining the quality difference root cause. The method of the application achieves the technical effect of improving data analysis accuracy and improving model generalization ability.

Description

Technical Field

[0001] This application relates to the Internet field, and more particularly to a method, apparatus, and storage medium for determining the root causes of poor quality. Background Technology

[0002] With the development of Internet technology, network performance management has become a key area of ​​Internet development. In order to optimize the network performance management system and improve the user experience, root cause analysis of poor network quality has become an important direction for the development of Internet technology.

[0003] Existing methods for analyzing the root causes of poor quality mainly involve extracting feature indicators from actual business experience, using rule-based models or machine learning models for classification and identification, and then conducting root cause analysis based on the classification results.

[0004] In existing technologies, the use of rule-based models and machine learning models cannot fully account for information loss issues in different processes in real business scenarios, nor can they capture the dependencies between different base stations. Furthermore, existing technologies mainly analyze data from a portion of the region without considering the joint learning process between different regions. The lack of joint learning and data fusion leads to the problem of data silos. Consequently, it is impossible to fully utilize data from multiple locations for modeling and analysis, limiting the accuracy and generalization capabilities of the model. Therefore, existing root cause analysis techniques suffer from low data analysis accuracy and low model generalization ability. Summary of the Invention

[0005] This application provides a method, apparatus, and storage medium for determining the root causes of poor quality, in order to solve the technical problems of low data analysis accuracy and low model generalization ability in existing root cause analysis techniques.

[0006] Firstly, this application provides a method for determining the root causes of poor quality, comprising:

[0007] Obtain the original business data with the root cause of poor quality, extract features from the original business data based on the preset feature extraction algorithm, and determine the corresponding feature vector set;

[0008] Based on the operators corresponding to the original service data, determine the spatial relationships between the base stations involved in the original service data, and construct a spatial relationship diagram of the base stations based on the spatial relationships.

[0009] Graph neural network training is performed based on feature vector sets and spatial relationship graphs to determine the trained neural network model.

[0010] The model parameters are determined based on the neural network model, and the federated model parameters are determined by aggregation calculation based on the model parameters. The neural network model is updated based on the federated model parameters to determine the graph federated model used to determine the root causes of poor quality.

[0011] Optionally, a graph neural network is trained based on the feature vector set and spatial relationship graph to determine the trained neural network model, including:

[0012] Determine the graph data corresponding to the spatial relationship graph based on the spatial relationship graph and feature vector set;

[0013] Train a graph neural network based on graph data and obtain the trained neural network model;

[0014] The neural network model takes graph data as input and outputs a feature matrix corresponding to the graph data. The feature matrix represents the root causes of the quality issues in the original business data.

[0015] Optionally, model parameters are determined based on the neural network model, and aggregate calculations are performed based on the model parameters to determine federated model parameters. The neural network model is then updated based on the federated model parameters to determine a graph federated model used to identify the root causes of poor quality, including:

[0016] When the number of iterations of the neural network model is less than the preset number of iterations and the neural network model has not converged, the neural network model is trained based on the preset sample size, preset sample batch and preset number of iterations, and the model parameters corresponding to the neural network model at the preset time point are determined based on the neural network model.

[0017] Aggregate calculations are performed based on model parameters, and the aggregated federated model parameters are obtained. The neural network model is then updated based on the federated model parameters.

[0018] Optionally, the process of determining model parameters based on a neural network model, performing aggregate calculations based on these parameters to determine federated model parameters, updating the neural network model based on the federated model parameters, and determining a graph federated model for identifying the root causes of poor quality, further includes:

[0019] When the number of iterations of the neural network model equals the preset number of iterations, and / or when the neural network model converges, the neural network model is determined as a graph federation model for determining the root causes of poor quality.

[0020] Optionally, obtain the raw business data carrying the root cause labeling of poor quality, including:

[0021] Acquire operator information and business scenario information for data collection; determine data requirements for data collection based on operator information and business scenario information; define data interfaces and formats for data collection based on data requirements; and acquire corresponding raw business data based on data interfaces and formats.

[0022] The raw data is cleaned and standardized, and the processed raw data is stored in the database.

[0023] Optionally, features are extracted from the original business data based on a preset feature extraction algorithm, and the corresponding feature vector set is determined, including:

[0024] Obtain the original business data from the database, extract features from the original business data based on a preset feature extraction algorithm, and determine the original business data after feature extraction as a feature vector set.

[0025] Optionally, based on the operator corresponding to the original service data, the spatial relationships between the base stations involved in the original service data are determined, and a spatial relationship diagram of the base stations is constructed based on the spatial relationships, including:

[0026] Determine the operator information corresponding to the original service data, and based on the operator information, determine the base station information corresponding to the original service data, and based on the base station information, determine the spatial relationship between base stations;

[0027] The spatial relationships are abstracted and the spatial relationships of the base stations are determined as spatial topology, and a spatial relationship graph corresponding to the spatial relationships of the base stations is constructed.

[0028] Secondly, this application provides an apparatus for determining the root causes of poor quality, comprising:

[0029] The acquisition module is used to acquire the original business data with the root cause of poor quality, extract features from the original business data based on the preset feature extraction algorithm, and determine the corresponding feature vector set.

[0030] The first processing module is used to determine the spatial relationship between base stations involved in the original service data based on the operator corresponding to the original service data, and to construct a spatial relationship diagram of the base stations based on the spatial relationship.

[0031] The second processing module is used to train a graph neural network based on the feature vector set and spatial relationship graph, and to determine the trained neural network model.

[0032] The third processing module is used to determine model parameters based on the neural network model, perform aggregation calculations based on the model parameters to determine federated model parameters, update the neural network model based on the federated model parameters, and determine the graph federated model used to determine the root causes of poor quality.

[0033] Optionally, the device is also used for:

[0034] Determine the graph data corresponding to the spatial relationship graph based on the spatial relationship graph and feature vector set;

[0035] Train a graph neural network based on graph data and obtain the trained neural network model;

[0036] The neural network model takes graph data as input and outputs a feature matrix corresponding to the graph data. The feature matrix represents the root causes of the quality issues in the original business data.

[0037] Optionally, the device is also used for:

[0038] When the number of iterations of the neural network model is less than the preset number of iterations and the neural network model has not converged, the neural network model is trained based on the preset sample size, preset sample batch and preset number of iterations, and the model parameters corresponding to the neural network model at the preset time point are determined based on the neural network model.

[0039] Aggregate calculations are performed based on model parameters, and the aggregated federated model parameters are obtained. The neural network model is then updated based on the federated model parameters.

[0040] Optionally, the device is also used for:

[0041] When the number of iterations of the neural network model equals the preset number of iterations, and / or when the neural network model converges, the neural network model is determined as a graph federation model for determining the root causes of poor quality.

[0042] Optionally, the device is also used for:

[0043] Acquire operator information and business scenario information for data collection; determine data requirements for data collection based on operator information and business scenario information; define data interfaces and formats for data collection based on data requirements; and acquire corresponding raw business data based on data interfaces and formats.

[0044] The raw data is cleaned and standardized, and the processed raw data is stored in the database.

[0045] Optionally, the device is also used for:

[0046] Obtain the original business data from the database, extract features from the original business data based on a preset feature extraction algorithm, and determine the original business data after feature extraction as a feature vector set.

[0047] Optionally, the device is also used for:

[0048] Determine the operator information corresponding to the original service data, and based on the operator information, determine the base station information corresponding to the original service data, and based on the base station information, determine the spatial relationship between base stations;

[0049] The spatial relationships are abstracted and the spatial relationships of the base stations are determined as spatial topology, and a spatial relationship graph corresponding to the spatial relationships of the base stations is constructed.

[0050] A third aspect of this application provides an apparatus for determining the root causes of poor quality, comprising:

[0051] Processor and memory;

[0052] The memory stores the instructions that the computer executes;

[0053] The processor executes computer execution instructions stored in memory, causing the device for determining the root causes of poor quality to perform the method for determining the root causes of poor quality in any of the first aspects.

[0054] Fourthly, this application provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement a method for determining the root causes of poor quality as described in any of the first aspects.

[0055] This application provides a method, apparatus, and storage medium for determining the root causes of poor data quality. The method includes: acquiring original service data labeled with the root causes of poor data quality; extracting features from the original service data based on a preset feature extraction algorithm and determining the corresponding feature vector set; determining the spatial relationship between base stations involved in the original service data based on the operator corresponding to the original service data, and constructing a spatial relationship graph of the base stations based on the spatial relationship; training a graph neural network based on the feature vector set and the spatial relationship graph to determine the trained neural network model; determining the model parameters based on the neural network model, and performing aggregation calculations based on the model parameters to determine the parameters of the federated model; updating the neural network model based on the federated model parameters, and determining the graph federated model used to determine the root causes of poor data quality. The application obtains the feature vector set of the original business data through a preset feature extraction algorithm. Simultaneously, it constructs a topological graph structure of the spatial relationships of base stations involved in the original business data using a graph neural network, identifying base stations as nodes in the graph structure and the feature vector set of the original business data as the feature vector matrix corresponding to each node in the graph structure. Based on the graph neural network, it trains graph data containing the feature vector matrix and determines the trained neural network model. Federated learning is then used to further update the neural network model. The model parameters are aggregated and the results are distributed to the neural network models of each participant in the federated learning process, thus updating the neural network model and determining the final graph federated model for identifying the root causes of quality issues. This application achieves the technical effect of improving data analysis accuracy and enhancing model generalization ability by using federated learning to integrate data from various regions and training the model based on a graph neural network. Attached Figure Description

[0056] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0057] Figure 1The flowchart of the method for determining the root causes of poor quality provided in the embodiments of this application Figure 1 ;

[0058] Figure 2 The flowchart of the method for determining the root causes of poor quality provided in the embodiments of this application Figure 2 ;

[0059] Figure 3 A schematic diagram illustrating the construction of a base station spatial relationship diagram for the application of poor quality root cause analysis, provided in an embodiment of this application;

[0060] Figure 4 A schematic diagram of lateral federated learning applied to poor quality root cause analysis provided in this application embodiment. Figure 1 ;

[0061] Figure 5 A schematic diagram of lateral federated learning applied to poor quality root cause analysis provided in this application embodiment. Figure 2 ;

[0062] Figure 6 A schematic diagram of the structure of the device for determining the root causes of poor quality provided in an embodiment of this application;

[0063] Figure 7 This is a hardware structure diagram of the device for determining the root causes of poor quality provided in an embodiment of this application.

[0064] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0065] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0066] Existing methods for analyzing the root causes of poor data quality primarily extract feature indicators from actual business experience, classify and identify them using rule-based models or machine learning models, and then perform root cause analysis based on the classification results. However, existing technologies, due to the use of rule-based and machine learning models, cannot fully consider information loss issues that occur in different processes within real-world business scenarios, nor can they capture the dependencies between different base stations. Furthermore, existing technologies mainly analyze data from partial regions, neglecting the joint learning process between different regions, resulting in a lack of joint learning and data fusion, leading to data silos. Consequently, they cannot fully utilize data from multiple locations for modeling and analysis, limiting the model's accuracy and generalization capabilities. Therefore, existing methods for analyzing the root causes of poor data quality suffer from low data analysis accuracy and low model generalization ability.

[0067] This application provides a method, apparatus, and storage medium for determining the root causes of poor data quality. The method includes: acquiring original service data labeled with the root causes of poor data quality; extracting features from the original service data based on a preset feature extraction algorithm and determining the corresponding feature vector set; determining the spatial relationship between base stations involved in the original service data based on the operator corresponding to the original service data, and constructing a spatial relationship graph of the base stations based on the spatial relationship; training a graph neural network based on the feature vector set and the spatial relationship graph to determine the trained neural network model; determining the model parameters based on the neural network model, and performing aggregation calculations based on the model parameters to determine the parameters of the federated model; updating the neural network model based on the federated model parameters, and determining the graph federated model used to determine the root causes of poor data quality. This application obtains the feature vector set of the original business data through a preset feature extraction algorithm. Simultaneously, it constructs a topological graph structure of the spatial relationships of base stations involved in the original business data using a graph neural network, identifying base stations as nodes in the graph structure and the feature vector set of the original business data as the feature vector matrix corresponding to each node in the graph structure. Based on the graph neural network, it trains graph data containing the feature vector matrix and determines the trained neural network model. Federated learning is then used to further update the neural network model. The model parameters are aggregated and the results are distributed to the neural network models of each participant in the federated learning process, thus updating the neural network model and determining the final graph federated model for identifying the root causes of quality issues. This application solves the technical problems of low data analysis accuracy and low model generalization ability in existing technologies by using federated learning to integrate data from various regions and training the model based on a graph neural network.

[0068] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0069] Figure 1 The flowchart of the method for determining the root causes of poor quality provided in the embodiments of this application Figure 1 .like Figure 1 As shown, the method includes:

[0070] S101. Obtain the original business data with the root cause of poor quality, extract features from the original business data based on the preset feature extraction algorithm, and determine the corresponding feature vector set.

[0071] In this embodiment, the acquisition of the original business data carrying the root cause of poor quality depends on the currently selected operator and the corresponding business scenario; the acquisition of the preset feature extraction algorithm can be: based on the preset time range and preset spatial range, the preset business scenario and the historical poor quality work order data corresponding to the preset operator are obtained, and the preset feature extraction algorithm is designed using the historical poor quality work order data.

[0072] S102. Based on the operators corresponding to the original service data, determine the spatial relationship between the base stations involved in the original service data, and construct a spatial relationship diagram of the base stations based on the spatial relationship.

[0073] In this embodiment, the spatial relationship of base stations may include: the distance relationship between base stations, the propagation direction relationship between base stations, and the coverage area relationship between base stations; the connection relationship between base stations is determined by considering various factors; the base stations are determined as nodes of the graph, the connection relationship between base stations is determined as edges of the graph and edge weights, and the relationship matrix of the graph is determined by the edges of the base stations, thereby constructing a spatial relationship graph of base stations.

[0074] S103. Train the graph neural network based on the feature vector set and spatial relationship graph, and determine the trained neural network model;

[0075] In this embodiment, the feature vector set corresponds to the feature vector of the node in the spatial relationship graph. The feature vector is used as the feature vector matrix of the spatial relationship graph to represent the node features and to train the graph neural network.

[0076] S104. Determine model parameters based on the neural network model, perform aggregation calculations based on the model parameters to determine federated model parameters, update the neural network model based on the federated model parameters, and determine the graph federated model used to determine the root causes of poor quality.

[0077] In this embodiment, the graph federation model is implemented based on federated learning, which includes horizontal federated learning and vertical federated learning. Different types of federated learning are performed based on the features of the data. Horizontal federated learning is suitable for situations where there is a lot of overlap in the data features of the participants and little overlap in the sample labels. Vertical federated learning is suitable for situations where there is a lot of overlap in the training sample labels of the participants and little overlap in the data features.

[0078] This application provides a method for determining the root causes of poor data quality. The method includes: acquiring original service data labeled with the root causes of poor data quality; extracting features from the original service data based on a preset feature extraction algorithm and determining the corresponding feature vector set; determining the spatial relationship between base stations involved in the original service data based on the operator corresponding to the original service data, and constructing a spatial relationship graph of the base stations based on the spatial relationship; training a graph neural network based on the feature vector set and the spatial relationship graph to determine the trained neural network model; determining the model parameters based on the neural network model, and performing aggregation calculations based on the model parameters to determine the parameters of the federated model; updating the neural network model based on the federated model parameters, and determining the graph federated model used to determine the root causes of poor data quality. This application obtains the feature vector set of the original business data through a preset feature extraction algorithm. Simultaneously, it constructs a topological graph structure of the spatial relationships of base stations involved in the original business data using a graph neural network, identifying base stations as nodes in the graph structure and the feature vector set of the original business data as the feature vector matrix corresponding to each node in the graph structure. Based on the graph neural network, it trains graph data containing the feature vector matrix and determines the trained neural network model. Federated learning is then used to further update the neural network model. The model parameters are aggregated and the results are distributed to the neural network models of each participant in the federated learning process, thus updating the neural network model and determining the final graph federated model for identifying the root causes of quality issues. This application solves the technical problems of low data analysis accuracy and low model generalization ability in existing technologies by using federated learning to integrate data from various regions and training the model based on a graph neural network.

[0079] Figure 2 The flowchart of the method for determining the root causes of poor quality provided in the embodiments of this application Figure 2 .like Figure 2 As shown, the method includes:

[0080] S201. Obtain operator information and business scenario information for data collection; determine data requirements for data collection based on operator information and business scenario information; define data interfaces and formats for data collection based on data requirements; obtain corresponding raw business data based on data interfaces and formats; perform data cleaning and standardization on the raw data; and store the processed raw data in the database.

[0081] In this embodiment, the business scenario information can be: mobile communication and data communication.

[0082] In the first example, a graph federation model corresponding to the root causes of poor data quality is trained based on cell data. The data collection requirements include: daily-level cell-level PM data, daily-level cell-level MR data, cell basic and CM data, cell handover data, and historical work order data of poor-quality cells. For each type of data, a corresponding data interface and format are defined, which may include: field name, poor-quality data type, data table name, and data unit. The corresponding raw business data is collected through the data interface and format. Data preprocessing is performed based on the raw business data, including: removing duplicate data, cleaning data, and filling missing data. The preprocessed data is then stored in the database.

[0083] In this context, a cell refers to a specific area within a wireless communication system, covered by one or more base stations. Each cell has a unique identifier called a Cell ID. Cell basic data describes the basic information and characteristics of a cell; this data is typically provided by mobile network operators or network equipment providers for management purposes. Configuration Management Data (CM) records and manages network equipment configuration information, including various configuration parameters and settings to ensure proper operation and performance optimization. Measurement Report (MR) data monitors and evaluates the mobile network quality within a cell; MR data is generated by mobile devices using the mobile network and records network performance information at specific times and locations. Cell handover data records and analyzes data related to cell handover during mobile network operations. Cell handover refers to a mobile device switching from one cell to another to maintain network connectivity continuity and stability.

[0084] S202. Obtain the original business data from the database, extract features from the original business data based on the preset feature extraction algorithm, and determine the original business data after feature extraction as a feature vector set;

[0085] In this embodiment, in the second example, network quality poor work order data in mobile network operation is obtained, and a preset feature extraction algorithm is designed based on the experience of processing network quality poor work order data. The extracted features are then standardized to reduce the impact of different dimensions of data. The preset feature extraction algorithm is used to extract features from the original business data in the database, and a feature vector set after feature extraction is obtained.

[0086] S203. Determine the operator information corresponding to the original service data, and based on the operator information, determine the base station information corresponding to the original service data, and based on the base station information, determine the spatial relationship between base stations; perform abstraction processing based on the spatial relationship, determine the spatial relationship of the base stations as a spatial topology, and construct the spatial relationship diagram corresponding to the spatial relationship of the base stations.

[0087] In this embodiment, the spatial relationship of the base stations is abstracted and transformed into a spatial topology graph structure. In the third example, the distance relationship, direction relationship and coverage area relationship of the base stations need to be considered during the construction of the graph, and the weight of each edge between the base stations needs to be determined when constructing the topology graph.

[0088] In this embodiment, the spatial relationship graph can be constructed using the K-nearest neighbor method, the minimum spanning tree method, and the shortest path method.

[0089] S204. Determine the graph data corresponding to the spatial relationship graph based on the spatial relationship graph and the feature vector set; train the graph neural network based on the graph data and obtain the trained neural network model;

[0090] In this embodiment, the input data of the neural network model is graph data, and the output data is the feature matrix corresponding to the graph data. The feature matrix is ​​the root cause of the quality problems corresponding to the original business data.

[0091] In the fourth example, a graph convolutional neural network is used to train the model. The graph data is defined as G = (V, E, A, X), where V is the set of points in the graph, corresponding to the base stations; E is the set of edges in the graph, corresponding to the connection relationships between the base stations, i.e., the connections constructed by the spatial relationship graph; A ∈ R n×n It is the adjacency matrix of the graph, representing the topological relations of the spatial relational graph; X∈R n×d It is the feature vector matrix of the graph nodes, i.e. the feature vector set, where each node contains a d-dimensional feature vector; the graph data G is used as the model input data, and the model output data is the root cause of the quality difference. The model is trained by calculating the error between the output data and the feature vector of the node.

[0092] The training of graph convolutional neural networks includes:

[0093] Define the symmetric normalized Laplace matrix L = I of the graph data G. N -D -1 / 2 AD -1 / 2 =UΛU T , where I N Let A be the identity matrix of the graph, D be the adjacency matrix of the graph, U be the eigenvector matrix of the normalized graph Laplacian operator, and Λ be the diagonal matrix composed of eigenvalues. Given a signal As input, another signal g w= diag(w), where w∈R n As parameters, and acting as filters, convolution in the spectral domain is defined as:

[0094] g w ★x=Ug w (Λ)U T x

[0095] Where w0 and w1 are parameters, and further, by applying the constraint w = w0 = -w1, the formula can be simplified to:

[0096]

[0097] In order to solve To address the gradient explosion and / or vanishing problems, graph convolutional neural networks introduce a renormalization operation: Finally, the convolution calculation formula can be simplified as follows:

[0098]

[0099] in, It is the input matrix. It is a parameter matrix. The convolutional matrix, F, F ′ These are the input dimension and the output dimension, respectively. It is the adjacency matrix of graph data G plus self-connections, that is, each vertex adds an edge to itself. N It is the identity matrix. yes The degree matrix, i.e.

[0100] S205. When the number of iterations of the neural network model is less than the preset number of iterations and the neural network model has not converged, train the neural network model based on the preset sample size, preset sample batch and preset number of iterations, and determine the model parameters corresponding to the neural network model at the preset time point based on the neural network model; perform aggregation calculation based on the model parameters, obtain the federated model parameters after aggregation calculation, and update the neural network model based on the federated model parameters.

[0101] In this embodiment, in order to improve the utilization rate of operator data in different regions and eliminate the data silo problem between different regions and institutions, this application constructs a graph federation model based on graph neural networks to address the root cause problem of poor operator data quality, thereby achieving joint modeling and collaborative optimization of multiple regions and institutions without disclosing user privacy.

[0102] In the fifth example, a graph federated model is constructed based on lateral federated learning. The federated averaging algorithm is used for lateral federated learning, specifically:

[0103] Determine the optimization objective of the algorithm:

[0104]

[0105] The total number of participants in federated learning is K, ω∈R. d P represents the d-dimensional model parameters. k and n k Let be the sample set and the number of samples for the k-th participant, respectively, and let n be the total number of samples from all participants. F is the error calculated using the gradient descent algorithm on the sample set of the k-th participant. k (ω) is the average error calculated by the gradient descent algorithm on the sample set of the k-th participant, and f(ω) is the mean error calculated by the gradient descent algorithm for all participants in the federated learning. A parameter ω is set for the federated model of the graph federated model at time t. t .

[0106] The specific execution process of this algorithm can be divided into three steps:

[0107] (1) The parameters ω of the federated model t Sent to participants in this round of training;

[0108] (2) The selected participants use the gradient descent algorithm on their own local dataset, setting the batch size B and the number of iterations Epoch E, to calculate the gradient g. k The model parameters are then updated, specifically as follows:

[0109]

[0110]

[0111] in, The model gradient is computed locally for each participant. New model parameters are applied to the k-th participant, where η is an adjustable learning rate;

[0112] (3) These participants send the new model parameters to the parameter server of the federated learning, and the parameter server updates the federated model parameters based on the model parameters returned by each participant:

[0113]

[0114] In this formula, ω has no superscript. t+1 These are the parameters for the new round of federal modeling.

[0115] S206. When the number of iterations of the neural network model equals the preset number of iterations, and / or when the neural network model converges, the neural network model is determined as a graph federation model for determining the root causes of poor quality.

[0116] In this embodiment, in the sixth example, a new round of local model training for the participants is performed based on the obtained federated model parameters. It is determined whether the current number of iterations of the neural network model exceeds the preset number of iterations or whether the mean error calculated by the gradient descent algorithm for all participants in the federated learning is less than or equal to the preset convergence value, i.e., the neural network model has converged. If so, the new round of training of the neural network model is stopped, and the neural network model with the latest federated model parameters is determined as the graph federated model.

[0117] By executing S201 to S206, federated learning and graph network construction are used to train the model, which improves the representation of data characteristics and increases the utilization rate of data, thereby achieving the technical effects of improving the model's generalization ability and increasing the accuracy of data analysis.

[0118] Figure 3 This is a schematic diagram illustrating the construction of a base station spatial relationship diagram for root cause analysis of poor quality, provided in an embodiment of this application. Figure 4 A schematic diagram of lateral federated learning applied to poor quality root cause analysis provided in this application embodiment. Figure 1 , Figure 5 A schematic diagram of lateral federated learning applied to poor quality root cause analysis provided in this application embodiment. Figure 2 .like Figure 3 As shown, base stations are treated as graph nodes, and the spatial relationships between base stations are defined as the connection relationships between graph nodes. A spatial relationship graph of base stations is then constructed based on these connection relationships. Figure 4 and Figure 5 As shown, the data features of each participant overlap significantly while the sample labels overlap less. Therefore, a horizontal federated learning approach is chosen for model training. Each participant must first train its own graph neural network model based on local data and upload the model parameters to the parameter server of the federated learning. The parameter server aggregates the model parameters of each participant and forms a federated model parameter. The federated model parameter is then distributed to each participant, and the participant's neural network model is updated. When the entire training process reaches the preset maximum iteration value or the model training converges, the training process stops, and the final graph federated model is obtained.

[0119] This application provides a method for determining the root causes of poor data quality. The method includes: acquiring original service data labeled with the root causes of poor data quality; extracting features from the original service data based on a preset feature extraction algorithm and determining the corresponding feature vector set; determining the spatial relationship between base stations involved in the original service data based on the operator corresponding to the original service data, and constructing a spatial relationship graph of the base stations based on the spatial relationship; training a graph neural network based on the feature vector set and the spatial relationship graph to determine the trained neural network model; determining the model parameters based on the neural network model, and performing aggregation calculations based on the model parameters to determine the parameters of the federated model; updating the neural network model based on the parameters of the federated model to determine the graph federated model used to determine the root causes of poor data quality. Compared to existing technologies, expressing the relationship between service data and base stations through network topology graphs improves the accuracy of extracting data features carrying root cause markers of poor quality. Simultaneously, using graph neural networks for model training extends deep learning to unstructured graph data, enhancing the model's graph structure data mining capabilities. Federated learning is performed concurrently with graph neural network training to acquire raw service data from various regions, aggregate the model parameters trained by the neural network models from each region, and update the local models, thus achieving the goal of joint learning among all parties. Compared to existing technologies, federated learning expands the sample size of training data while maintaining data confidentiality, thereby improving the model's generalization ability and the accuracy of data analysis. This solves the technical problems of low data analysis accuracy and low model generalization ability in existing technologies.

[0120] Figure 6 This is a schematic diagram of the device for determining the root causes of poor quality provided in an embodiment of this application. The device in this embodiment can be in the form of software and / or hardware. Figure 6 As shown in the embodiment of this application, a device 600 for determining the root causes of poor quality includes: an acquisition module 601, a first processing module 602, a second processing module 603, and a third processing module 604.

[0121] The acquisition module 601 is used to acquire the original business data carrying the root cause label of poor quality, extract features from the original business data based on the preset feature extraction algorithm, and determine the corresponding feature vector set;

[0122] The first processing module 602 is used to determine the spatial relationship between base stations involved in the original service data based on the operator corresponding to the original service data, and to construct a spatial relationship diagram of the base stations based on the spatial relationship.

[0123] The second processing module 603 is used to train a graph neural network based on a feature vector set and a spatial relationship graph, and to determine the trained neural network model.

[0124] The third processing module 604 is used to determine model parameters based on the neural network model, perform aggregation calculations based on the model parameters to determine federated model parameters, update the neural network model based on the federated model parameters, and determine the graph federated model used to determine the root causes of poor quality.

[0125] In one possible implementation, the device is also used for:

[0126] Determine the graph data corresponding to the spatial relationship graph based on the spatial relationship graph and feature vector set;

[0127] Train a graph neural network based on graph data and obtain the trained neural network model;

[0128] The neural network model takes graph data as input and outputs a feature matrix corresponding to the graph data. The feature matrix represents the root causes of the quality issues in the original business data.

[0129] In one possible implementation, the device is also used for:

[0130] When the number of iterations of the neural network model is less than the preset number of iterations and the neural network model has not converged, the neural network model is trained based on the preset sample size, preset sample batch and preset number of iterations, and the model parameters corresponding to the neural network model at the preset time point are determined based on the neural network model.

[0131] Aggregate calculations are performed based on model parameters, and the aggregated federated model parameters are obtained. The neural network model is then updated based on the federated model parameters.

[0132] In one possible implementation, the device is also used for:

[0133] When the number of iterations of the neural network model equals the preset number of iterations, and / or when the neural network model converges, the neural network model is determined as a graph federation model for determining the root causes of poor quality.

[0134] In one possible implementation, the device is also used for:

[0135] Acquire operator information and business scenario information for data collection; determine data requirements for data collection based on operator information and business scenario information; define data interfaces and formats for data collection based on data requirements; and acquire corresponding raw business data based on data interfaces and formats.

[0136] The raw data is cleaned and standardized, and the processed raw data is stored in the database.

[0137] In one possible implementation, the device is also used for:

[0138] Obtain the original business data from the database, extract features from the original business data based on a preset feature extraction algorithm, and determine the original business data after feature extraction as a feature vector set.

[0139] In one possible implementation, the device is also used for:

[0140] Determine the operator information corresponding to the original service data, and based on the operator information, determine the base station information corresponding to the original service data, and based on the base station information, determine the spatial relationship between base stations;

[0141] The spatial relationships are abstracted and the spatial relationships of the base stations are determined as spatial topology, and a spatial relationship graph corresponding to the spatial relationships of the base stations is constructed.

[0142] This application provides an apparatus for determining the root causes of poor data quality. The apparatus includes: an acquisition module, used to acquire raw service data carrying labels of root causes of poor data quality, extract features from the raw service data based on a preset feature extraction algorithm, and determine the corresponding feature vector set; a first processing module, used to determine the spatial relationship between base stations involved in the raw service data based on the operator corresponding to the raw service data, and construct a spatial relationship graph of the base stations based on the spatial relationship; a second processing module, used to train a graph neural network based on the feature vector set and the spatial relationship graph, and determine the trained neural network model; a third processing module, used to determine the model parameters based on the neural network model, perform aggregation calculations based on the model parameters to determine the parameters of the federated model, update the neural network model based on the federated model parameters, and determine the graph federated model used to determine the root causes of poor data quality. Compared to existing technologies, expressing the relationship between service data and base stations through network topology graphs improves the accuracy of extracting data features carrying root cause markers of poor quality. Simultaneously, using graph neural networks for model training extends deep learning to unstructured graph data, enhancing the model's graph structure data mining capabilities. Federated learning is performed concurrently with graph neural network training to acquire raw service data from various regions, aggregate the model parameters trained by the neural network models from each region, and update the local models, thus achieving the goal of joint learning among all parties. Compared to existing technologies, federated learning expands the sample size of training data while maintaining data confidentiality, thereby improving the model's generalization ability and the accuracy of data analysis. This solves the technical problems of low data analysis accuracy and low model generalization ability in existing technologies.

[0143] Figure 7 This is a hardware structure diagram of the device for determining the root causes of poor quality provided in an embodiment of this application. (See diagram below.) Figure 7 As shown, the device 700 for determining the root cause of poor quality includes:

[0144] Processor 701 and memory 702;

[0145] The memory stores instructions that the computer executes;

[0146] The processor executes computer execution instructions stored in memory 702, causing the device for determining the root causes of poor quality to perform the method for determining the root causes of poor quality as described above.

[0147] It should be understood that the processor 701 described above can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor. The memory 702 may include high-speed random access memory (RAM), and may also include non-volatile memory (NVM), such as at least one disk storage device, and may also be a USB flash drive, external hard drive, read-only memory, disk, or optical disc, etc.

[0148] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement a method for determining the root causes of poor quality.

[0149] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily essential to this application.

[0150] It should be further noted that although the steps in the flowchart are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowchart may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.

[0151] It should be understood that the above-described device embodiments are merely illustrative, and the device of this application can also be implemented in other ways. For example, the division of units / modules in the above embodiments is only a logical functional division, and there may be other division methods in actual implementation. For example, multiple units, modules, or components may be combined, or integrated into another system, or some features may be ignored or not executed.

[0152] Furthermore, unless otherwise specified, the functional units / modules in the various embodiments of this application can be integrated into one unit / module, or each unit / module can exist physically separately, or two or more units / modules can be integrated together. The integrated units / modules described above can be implemented in hardware or as software program modules.

[0153] When integrated units / modules are implemented in hardware, the hardware can be digital circuits, analog circuits, etc. The physical implementation of the hardware structure includes, but is not limited to, transistors, memristors, etc. Unless otherwise specified, the processor can be any suitable hardware processor, such as a CPU, GPU, FPGA, DSP, and ASIC, etc. Unless otherwise specified, the storage unit can be any suitable magnetic or magneto-optical storage medium, such as Resistive Random Access Memory (RRAM), Dynamic Random Access Memory (DRAM), Static Random Access Memory (SRAM), Enhanced Dynamic Random Access Memory (EDRAM), High-Bandwidth Memory (HBM), Hybrid Memory Cube (HMC), etc.

[0154] If the integrated unit / module is implemented as a software program module and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned memory includes various media capable of storing program code, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard drive, magnetic disk, or optical disk.

[0155] In the above embodiments, the descriptions of each embodiment have their own emphasis. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments. The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification.

[0156] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this application are indicated by the following claims.

[0157] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.

Claims

1. A method for determining the root causes of poor quality, characterized in that, include: Obtain raw business data with labels indicating poor quality root causes, extract features from the raw business data based on a preset feature extraction algorithm, and determine the corresponding feature vector set; Based on the operator corresponding to the original service data, determine the spatial relationship between the base stations involved in the original service data, and construct a spatial relationship diagram of the base stations based on the spatial relationship; Based on the spatial relationship graph and the feature vector set, determine the graph data corresponding to the spatial relationship graph; A graph neural network is trained based on the graph data, and the trained neural network model is obtained; wherein the input data of the neural network model is the graph data, the output data is the feature matrix corresponding to the graph data, and the feature matrix is ​​the root cause of the quality difference corresponding to the original business data; Based on the neural network model, model parameters are determined, and based on the model parameters, aggregate calculations are performed to determine federated model parameters. Based on the federated model parameters, the neural network model is updated to determine a graph federated model used to determine the root causes of the quality defects.

2. The method according to claim 1, characterized in that, The process of determining model parameters based on the neural network model, performing aggregation calculations based on the model parameters to determine federated model parameters, updating the neural network model based on the federated model parameters, and determining a graph federated model for determining the root causes of poor quality includes: When the number of iterations of the neural network model is less than the preset number of iterations and the neural network model has not converged, the neural network model is trained based on the preset sample size, preset sample batch and preset number of iterations, and the model parameters corresponding to the neural network model at the preset time point are determined based on the neural network model. Aggregate calculations are performed based on the model parameters, and the aggregated federated model parameters are obtained. The neural network model is then updated based on the federated model parameters.

3. The method according to claim 2, characterized in that, The process of determining model parameters based on the neural network model, performing aggregation calculations based on the model parameters to determine federated model parameters, updating the neural network model based on the federated model parameters, and determining a graph federated model for determining the root causes of poor quality further includes: When the number of iterations of the neural network model is equal to the preset number of iterations, and / or when the neural network model converges, the neural network model is determined as the graph federation model used to determine the root causes of the quality defects.

4. The method according to claim 1, characterized in that, The acquisition of raw business data carrying the root cause labeling of poor quality includes: The system acquires operator information and business scenario information for data collection, determines data requirements for data collection based on the operator information and business scenario information, defines data interfaces and formats for data collection based on the data requirements, and acquires corresponding raw business data based on the data interfaces and formats. The raw data is cleaned and standardized, and the processed raw data is stored in the database.

5. The method according to claim 4, characterized in that, The step of extracting features from the original business data based on a preset feature extraction algorithm and determining the corresponding feature vector set includes: Obtain the original business data from the database, extract features from the original business data based on the preset feature extraction algorithm, and determine the original business data after feature extraction as the feature vector set.

6. The method according to claim 5, characterized in that, The step of determining the spatial relationships between base stations involved in the original service data based on the corresponding operator, and constructing a spatial relationship diagram of the base stations based on the spatial relationships, includes: The operator information corresponding to the original service data is determined, and the base station information corresponding to the original service data is determined based on the operator information, and the spatial relationship between the base stations is determined based on the base station information; The spatial relationships are abstracted and the spatial relationships of the base stations are determined as spatial topology, and a spatial relationship graph corresponding to the spatial relationships of the base stations is constructed.

7. A device for determining the root causes of poor quality, characterized in that, include: The acquisition module is used to acquire the original business data carrying the root cause label of poor quality, extract features from the original business data based on a preset feature extraction algorithm, and determine the corresponding feature vector set. The first processing module is used to determine the spatial relationship between base stations involved in the original service data based on the operator corresponding to the original service data, and to construct a spatial relationship diagram of the base stations based on the spatial relationship. The second processing module is used to determine the graph data corresponding to the spatial relationship graph based on the spatial relationship graph and the feature vector set; The third processing module is used to train a graph neural network based on the graph data and obtain the trained neural network model; wherein the input data of the neural network model is the graph data, the output data is the feature matrix corresponding to the graph data, and the feature matrix is ​​the root cause of the quality difference corresponding to the original business data; The fourth processing module is used to determine model parameters based on the neural network model, perform aggregation calculations based on the model parameters to determine federated model parameters, update the neural network model based on the federated model parameters, and determine a graph federated model for determining the root causes of the quality defects.

8. A device for determining the root causes of poor quality, characterized in that, include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory to implement the method for determining the root causes of poor quality as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method for determining the root causes of poor quality as described in any one of claims 1 to 6.

Images