Methods and apparatus for sample data generation, model training, and determination of causes of abnormal operation

Abstract

This specification provides a method and apparatus for generating sample data, training a model, and determining the causes of abnormal operation. The sample data generation method includes: determining the causes of abnormal operation of a target application using an automated attribution model corresponding to the target application; classifying each cause of abnormal operation to obtain the type of cause of abnormal operation; and obtaining pre-configured operation data indicators for the target application; obtaining abnormal operation data corresponding to the cause of abnormal operation type from the historical abnormal operation data of the target application, and determining the data indicator value of the abnormal operation data under the operation data indicator item; and using the cause of abnormal operation type and data indicator value as model training sample data. Through this embodiment, for occasional application anomalies, the causes of application anomalies can be determined without relying on massive amounts of data on similar anomalies, improving the accuracy of determining the causes of application anomalies.

Description

Technical Field

[0001] This document relates to the field of computer technology, and in particular to a method and apparatus for generating sample data, training models, and determining the causes of abnormal operation. Background Technology

[0002] Determining the cause of an application's abnormal operation can guide developers to further optimize the application. Existing technologies utilize automated cause analysis systems to pinpoint the reasons for application anomalies; however, these systems rely on massive amounts of data on similar anomalies. For occasional application anomalies, the lack of similar data reduces the accuracy of these systems in locating the cause. Summary of the Invention

[0003] This specification provides a method and apparatus for generating sample data, training models, and determining the causes of abnormal operation. For occasional application anomalies, it can determine the cause of application anomalies without relying on massive amounts of data on similar anomalies, thereby improving the accuracy of determining the cause of application anomalies.

[0004] Firstly, embodiments of this specification provide a method for generating sample data, including:

[0005] By using the automated attribution model corresponding to the target application, at least one cause of abnormal operation of the target application is determined, and the at least one cause of abnormal operation is classified to obtain the type of abnormal operation cause.

[0006] From the historical abnormal operation data of the target application, obtain the abnormal operation data corresponding to the abnormal operation reason type, and determine the data indicator value of the abnormal operation data under the preset operation data indicator item;

[0007] Determine the correspondence between the abnormal operation cause type and the data indicator value, and use the abnormal operation cause type and the data indicator value as model training sample data according to the correspondence.

[0008] Secondly, the embodiments of this specification provide a model training method, including:

[0009] Obtain the model training sample data generated by the method described in the first aspect above;

[0010] Using the model training sample data, a pre-built neural network structure is trained until convergence. The converged neural network structure is used as the model for determining the cause of abnormal operation of the target application.

[0011] The parameters in the neural network structure represent the contribution weights of each data indicator value corresponding to the same abnormal operation cause type to that same abnormal operation cause type.

[0012] Thirdly, the embodiments of this specification provide a method for determining the cause of abnormal operation, including:

[0013] Obtain the abnormal operation cause determination model of the target application trained by the method described in the second aspect above, and obtain the abnormal operation data of the target application to be analyzed;

[0014] Determine the value of the data indicator to be analyzed under the operation data indicator item for the abnormal operation data to be analyzed;

[0015] The abnormal operation cause determination model processes the data indicator values ​​to be analyzed, and determines the type of abnormal operation cause of the target application based on the processing results.

[0016] Fourthly, embodiments of this specification provide a sample data generation apparatus, including:

[0017] The cause classification unit is used to determine at least one cause of abnormal operation of the target application through the automated attribution model corresponding to the target application, and classify the at least one cause of abnormal operation to obtain the cause type of abnormal operation.

[0018] The indicator determination unit is used to obtain abnormal operation data corresponding to the abnormal operation cause type from the historical abnormal operation data of the target application, and determine the data indicator value of the abnormal operation data under the preset operation data indicator item.

[0019] The sample generation unit is used to determine the correspondence between the abnormal operation cause type and the data indicator value, and to use the abnormal operation cause type and the data indicator value as model training sample data according to the correspondence.

[0020] Fifthly, embodiments of this specification provide a model training apparatus, including:

[0021] A sample acquisition unit is used to acquire model training sample data generated by the apparatus as described in the fourth aspect above.

[0022] The model training unit is used to train a pre-built neural network structure using the model training sample data until it converges, and to use the converged neural network structure as the model for determining the cause of abnormal operation of the target application.

[0023] The parameters in the neural network structure represent the contribution weights of each data indicator value corresponding to the same abnormal operation cause type to that same abnormal operation cause type.

[0024] Sixthly, embodiments of this specification provide an apparatus for determining the cause of abnormal operation, including:

[0025] The model acquisition unit is used to acquire the abnormal operation cause determination model of the target application trained by the device described in the fifth aspect above, and to acquire the abnormal operation data of the target application to be analyzed.

[0026] A data determination unit is used to determine the data indicator value to be analyzed under the operation data indicator item for the abnormal operation data to be analyzed.

[0027] The cause analysis unit is used to process the data indicator values ​​to be analyzed through the abnormal operation cause determination model, and determine the abnormal operation cause type of the target application based on the processing result.

[0028] In a seventh aspect, embodiments of this specification provide an electronic device, including: a processor; and a memory configured to store computer-executable instructions, which, when executed, cause the processor to perform the steps of the method described in the first aspect, or the steps of the method described in the second aspect, or the steps of the method described in the third aspect.

[0029] Eighthly, embodiments of this specification provide a computer-readable storage medium for storing computer-executable instructions that, when executed by a processor, implement the steps of the method described in the first aspect, or implement the steps of the method described in the second aspect, or implement the steps of the method described in the third aspect.

[0030] Through one or more embodiments of this specification, the types of abnormal application operation causes and the data indicator values ​​of the abnormal application operation data under predetermined operation data indicator items are used as model training sample data. There is a correspondence between the abnormal operation cause types and the data indicator values. An application abnormal operation cause determination model is trained using this model training sample data, and the abnormal operation cause determination model is used to determine the cause of the application's abnormal operation. Because the correlation between the data indicator values ​​of the application's abnormal operation data and the application's abnormal operation cause types is fully considered when training the model using the generated model training sample data and analyzing the cause of the application's abnormal operation, it is not necessary to rely on massive amounts of data on similar abnormalities to determine the cause of the application's abnormal operation for occasional application anomalies, thus improving the accuracy of determining the cause of application abnormal operation. Attached Figure Description

[0031] To more clearly illustrate the technical solutions in one or more embodiments of this specification or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0032] Figure 1 A schematic flowchart illustrating a sample data generation method provided in one embodiment of this specification;

[0033] Figure 2 A schematic flowchart illustrating a model training method provided in one embodiment of this specification;

[0034] Figure 3 A flowchart illustrating a method for determining the cause of abnormal operation provided in one embodiment of this specification;

[0035] Figure 4 This is a schematic diagram of the sample data generation device provided in one embodiment of this specification;

[0036] Figure 5 This is a schematic diagram of the structure of a model training device provided in one embodiment of this specification;

[0037] Figure 6 This is a schematic diagram of the structure of an abnormal operation cause determination device provided in one embodiment of this specification;

[0038] Figure 7 This is a schematic diagram of the structure of an electronic device provided in one embodiment of this specification. Detailed Implementation

[0039] To enable those skilled in the art to better understand the technical solutions in one or more embodiments of this specification, the technical solutions in one or more embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this specification, and not all of the embodiments. Based on one or more embodiments of this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of this document.

[0040] Figure 1 This is a flowchart illustrating a sample data generation method provided in one embodiment of this specification, as shown below. Figure 1 As shown, the method includes the following steps:

[0041] Step S102: Using the automated attribution model corresponding to the target application, determine at least one cause of abnormal operation of the target application, classify the at least one cause of abnormal operation, and obtain the type of abnormal operation cause.

[0042] Step S104: Obtain the abnormal operation data corresponding to the abnormal operation cause type from the historical abnormal operation data of the target application, and determine the data indicator value of the abnormal operation data under the preset operation data indicator item.

[0043] Step S106: Determine the correspondence between the abnormal operation cause type and the data indicator value, and use the abnormal operation cause type and the data indicator value as model training sample data according to the correspondence.

[0044] In this embodiment, the abnormal operation cause type of the application and the data indicator value of the abnormal operation data of the application under the predetermined operation data indicator item are used as model training sample data. There is a corresponding relationship between the abnormal operation cause type and the data indicator value. Therefore, when training the model using the generated model training sample data and analyzing the abnormal operation cause of the application through the model, the correlation between the data indicator value of the abnormal operation data of the application and the abnormal operation cause type of the application is fully considered. Therefore, for occasional application abnormalities, it is not necessary to rely on a large amount of data of similar abnormalities to determine the abnormal operation cause of the application, thereby improving the accuracy of determining the abnormal operation cause of the application.

[0045] The method in this embodiment can be applied to a backend server and executed by the backend server, which can be a server specifically used to generate model training sample data.

[0046] In this embodiment, the target application can be any type of application, and there is no limitation. In step S102 above, the cause of abnormal operation of the target application is determined by the automated attribution model corresponding to the target application, the cause of abnormal operation is classified to obtain the type of abnormal operation cause, and the operation data indicators pre-configured for the target application are obtained. For the target application, there can be multiple causes of abnormal operation, one or more types of causes of abnormal operation, and one or more operation data indicators.

[0047] In one embodiment, the cause of abnormal operation of the target application is determined by an automated attribution model corresponding to the target application, specifically as follows:

[0048] (1) By attributing the abnormal operation of the target application to each attribution node in the automated attribution model, the cause of the abnormal operation of the target application can be obtained.

[0049] Each attribution node corresponds to a predetermined cause of abnormal operation; the causes of abnormal operation of the target application include the predetermined causes of abnormal operation.

[0050] Specifically, the automated attribution model corresponding to the target application is used to locate anomalies in the target application. This automated attribution model includes multiple sub-models, each containing an attribution node, thus comprising multiple attribution nodes. Each attribution node corresponds to a pre-set, manually configured reason for abnormal operation. Abnormal operation data of the target application can be input into this automated attribution model. The attribution node can determine whether the abnormal operation reason corresponding to the abnormal operation data is the pre-set reason for its own operation, thereby attributing the abnormal operation of the target application. In this embodiment, abnormal operation data of the target application over a period of time, such as one month, is obtained and input into the automated attribution model. The attribution nodes in the automated attribution model then perform abnormal operation attribution on the target application, obtaining various reasons for abnormal operation. Since each attribution node corresponds to a pre-set reason for abnormal operation, the reasons for abnormal operation of the target application should be among the pre-set reasons; that is, the reasons for abnormal operation of the target application include the pre-set reasons for abnormal operation.

[0051] In one embodiment, the automated attribution model corresponding to the target application originally includes multiple attribution nodes, each corresponding to a pre-defined, manually set reason for the target application's abnormal operation. By analyzing the abnormal operation data of the target application over a period of time using the automated attribution model, it can be determined that the cause of the target application's abnormality is located at multiple specific attribution nodes within each attribution node. Nodes other than these specific attribution nodes are not the cause of the target application's abnormality. Therefore, in this embodiment, the predetermined reasons for abnormal operation corresponding to each specific attribution node in the automated attribution model can also be obtained as the cause of the target application's abnormal operation.

[0052] In this embodiment, there can be multiple reasons for the abnormal operation of the target application. In step S102 above, the various reasons for abnormal operation are further classified to obtain at least one type of abnormal operation reason. Each type of abnormal operation reason includes at least one abnormal operation reason.

[0053] Specifically, the various reasons for abnormal operation of the target application are categorized and statistically analyzed to obtain at least one type of abnormal operation reason. Each type of abnormal operation reason contains at least one abnormal operation reason. These reasons can be output to the developers, who can then categorize and statistically analyze them to obtain at least one type of abnormal operation reason. Alternatively, the categorization and statistical analysis of the various abnormal operation reasons can be automated. For example, abnormal operation reasons containing the same keywords can be grouped into one type, such as classifying both "insufficient memory" and "memory read exception" into a memory-related abnormal operation reason type.

[0054] In one embodiment, the abnormal operation cause types include at least one of the following types: the target application's running memory is less than a preset minimum value; the target application's file descriptors exceed a preset maximum value; the target application has more threads than a preset maximum number; the target application has a null pointer exception; the target application has a dangling pointer causing abnormal memory access; the target application has a double free causing abnormal memory access; and the target application has a memory out-of-bounds exception causing abnormal memory access. These seven abnormal operation cause types can cover various abnormal operation causes of the target application, thereby achieving the effect of summarizing and generalizing the various abnormal operation causes of the target application.

[0055] In one embodiment, the method in this embodiment further includes the following actions to determine the runtime data metrics of the target application before obtaining the runtime data metrics pre-configured for the target application:

[0056] (1) Obtain at least one of the following: the identifier of the target application, the version number of the target application, the identifier of the operating system to which the target application is adapted, and the version number of the operating system to which the target application is adapted.

[0057] (2) Based on the information obtained, configure runtime data metrics for the target application.

[0058] First, in action (1), at least one of the following is obtained: the identifier of the target application, the version number of the target application, the identifier of the operating system to which the target application is adapted, and the version number of the operating system to which the target application is adapted. The identifier of the target application can be the name of the target application, and the identifier of the operating system to which the target application is adapted can be the name of the operating system, such as Android or iOS.

[0059] Considering that different applications may experience different runtime anomalies under different operating systems, the runtime data metrics that different applications need to focus on under different operating systems also differ. In action (2), runtime data metrics are configured for the target application based on the acquired information. In one case, the acquired information is output to the developers, who then configure runtime data metrics for the target application based on the acquired information. These runtime data metrics are the runtime data metrics that the target application needs to focus on under the current version and the adapted operating system.

[0060] In another scenario, the developers have pre-built a project library that records the correspondence between four items: the application's identifier, the application's version number, the identifier of the operating system the application is adapted to, and the operating system version number the application is adapted to, and the corresponding runtime data metrics of the application. For example, it records: Application A - Application Version 1.0 - Android System - Android 2.0 - Metric 1, Metric 2, Metric 3. Therefore, after obtaining at least one of the above information, the project library can be searched based on the obtained information to obtain the various runtime data metrics corresponding to the target application. If the information obtained in action (1) is less than four items, then when searching the project library based on the obtained information, if a record includes all the obtained information, then the metrics in that record are determined to be the various runtime data metrics corresponding to the target application.

[0061] In this embodiment, considering that different applications may experience different runtime anomalies under different operating systems, the runtime data metrics that different applications need to focus on under different operating systems are also different. Based on this, at least one of the following is obtained: the identifier of the target application, the version number of the target application, the identifier of the operating system to which the target application is adapted, and the version number of the operating system to which the target application is adapted. Based on the obtained information, runtime data metrics are configured for the target application, so that the configured runtime data metrics match the target application and its operating system.

[0062] After determining the runtime data metrics of the target application through the above process, the runtime data metrics pre-configured for the target application can be obtained. Then, step S104 is executed to obtain the abnormal runtime data corresponding to the abnormal runtime cause type from the historical abnormal runtime data of the target application, and determine the data metric value of the abnormal runtime data under the runtime data metrics.

[0063] First, obtain historical abnormal operation data of the target application. This historical abnormal operation data can be user-reported abnormal operation data of the target application within a historical period, such as one month. When the target application experiences abnormalities on the user side, it can manifest in various ways, such as program crashes or program freezes. Next, from the historical abnormal operation data of the target application, obtain the abnormal operation data corresponding to the abnormal operation reason type. If there are multiple abnormal operation reason types, obtain the abnormal operation data corresponding to each abnormal operation reason type separately.

[0064] In one embodiment, abnormal operation data corresponding to the abnormal operation reason type is obtained from the historical abnormal operation data of the target application, specifically as follows:

[0065] (1) By using each attribution node in the automated attribution model to perform abnormal operation attribution on historical abnormal operation data, the data corresponding to the predetermined abnormal operation cause corresponding to each attribution node is determined in the historical abnormal operation data; wherein, each attribution node corresponds to a predetermined abnormal operation cause.

[0066] (2) Based on the predetermined abnormal operation reason corresponding to each attribution node, determine the target attribution node corresponding to the abnormal operation reason type in each attribution node, and take the data in the historical abnormal operation data that corresponds to the predetermined abnormal operation reason corresponding to the target attribution node as the abnormal operation data corresponding to the abnormal operation reason type.

[0067] As described above, the automated attribution model corresponding to the target application includes multiple attribution nodes, each corresponding to a pre-set, manually configured reason for abnormal operation. Abnormal operation data of the target application can be input into this automated attribution model. Attribution nodes can determine whether the abnormal operation reason corresponding to the abnormal operation data is the pre-set reason for their own abnormal operation, thereby attributing the abnormal operation of the target application. Based on this principle, in this embodiment, historical abnormal operation data of the target application is obtained. This historical abnormal operation data consists of abnormal data reported multiple times by the user, therefore including multiple batches of sub-data, each batch being the abnormal data reported by the user once. This historical abnormal operation data is input into the automated attribution model. Each attribution node in the automated attribution model determines whether the abnormal reason of each batch of sub-data is the pre-set reason for its own abnormal operation. Therefore, each attribution node can analyze the historical abnormal operation data to obtain data corresponding to the pre-set reason for its own abnormal operation, thus determining the data corresponding to the pre-set reason for each attribution node within the historical abnormal operation data.

[0068] For example, historical abnormal operation data includes 9 batches of sub-data with three attribution nodes. Through analysis, it was found that the abnormality cause of sub-data batches 1-3 is the predetermined abnormal operation cause corresponding to the first attribution node. Therefore, the data corresponding to the predetermined abnormal operation cause corresponding to the first attribution node is sub-data batches 1-3. Similarly, the abnormality cause of sub-data batches 4-6 is the predetermined abnormal operation cause corresponding to the second attribution node. Therefore, the data corresponding to the predetermined abnormal operation cause corresponding to the second attribution node is sub-data batches 4-6. The abnormality cause of sub-data batches 7-9 is the predetermined abnormal operation cause corresponding to the third attribution node. Therefore, the data corresponding to the predetermined abnormal operation cause corresponding to the third attribution node is sub-data batches 7-9.

[0069] Next, based on the predetermined abnormal operation reason corresponding to each attribution node, the target attribution node corresponding to the abnormal operation reason type is determined among each attribution node. As described above, the abnormal operation reason type includes at least one abnormal operation reason of the target application, and the abnormal operation reason of the target application is located among each predetermined abnormal operation reason. Therefore, the abnormal operation reason corresponding to the abnormal operation reason type can be determined here, and the abnormal operation reason is searched among each predetermined abnormal operation reason. The attribution node corresponding to the found abnormal operation reason is determined as the target attribution node.

[0070] Finally, the data in the historical abnormal operation data that corresponds to the predetermined abnormal operation cause corresponding to the target attribution node is taken as the abnormal operation data corresponding to the abnormal operation cause type. For example, if the target attribution node is the third attribution node mentioned above, then the abnormal operation data corresponding to the abnormal operation cause type is the sub-data of batches 7-9 mentioned above.

[0071] If there are multiple types of abnormal operation reasons, the abnormal operation data corresponding to each abnormal operation reason type is obtained using the methods described above. After obtaining the abnormal operation data corresponding to each abnormal operation reason type, the data indicator value of that abnormal operation data under the above-mentioned operation data indicator items is determined.

[0072] In one embodiment, the runtime data metrics include at least one of the following: the virtual memory level of the user terminal where the target application resides; the operating system version information of the user terminal where the target application resides; the exception signals generated and recorded in the runtime data of the target application according to the preset exception signal generation rules; the number of threads in the target application; the number of preset exceptions that occurred before the target application crashed; and the exception stack type of the target application.

[0073] In one embodiment, determining the data indicator value of the abnormal operation data under the aforementioned operation data indicator items specifically involves automatically determining the data indicator value of the abnormal operation data under the aforementioned operation data indicator items using an automated script. It is understood that if there are multiple types of abnormal operation causes, it is necessary to determine the data indicator value of the abnormal operation data corresponding to each type of abnormal operation cause under the aforementioned operation data indicator items.

[0074] In one embodiment, the data indicator values ​​of abnormal running data under the running data indicator item are automatically determined by an automated script, including at least one of the following methods:

[0075] (1) Extract the virtual memory size of the user terminal from the abnormal operation data, and determine the virtual memory level based on the virtual memory size;

[0076] (2) Extract operating system version information from abnormal operation data;

[0077] (3) Extract abnormal signals from abnormal operation data;

[0078] (4) Extract the number of threads of the target application from the abnormal operation data, and determine the thread count level based on the number of threads;

[0079] (5) Determine the number of times the target application crashed before the default exception occurred based on the abnormal operation log of the target application recorded in the abnormal operation data;

[0080] (6) Extract the exception stack key of the target application from the abnormal operation data, and determine the exception stack type based on the exception stack key.

[0081] In method (1), the virtual memory size of the user terminal where the target application resides is automatically extracted from the abnormal operation data by a script. The virtual memory level is determined based on the virtual memory size and the pre-established mapping relationship between virtual memory size and virtual memory level. In method (2), if the abnormal operation data records the operating system version information of the user terminal where the target application resides, the operating system version information, such as the version number, is directly extracted by a script. In method (3), the abnormal operation data of the target application will generate and record abnormal signals according to preset abnormal signal generation rules. For example, according to the rules of the LUNIX system, abnormal signals with the content of natural numbers 1-12 are generated and recorded. The abnormal signal is directly extracted from the abnormal operation data by an automated script.

[0082] In method (4), the number of threads of the target application is extracted from the abnormal running data by a script, and the thread level is determined according to the number of threads and the pre-established mapping relationship between the number of threads and the thread level. In method (5), during the operation of the target application, a running log is generated. The running log is part of the running data. If the application runs abnormally, an abnormal running log is generated. The abnormal running log records the exceptions that occur. Each time an exception occurs, it is recorded once. For example, the exception "file open error" is recorded and recorded three times. In this method, the abnormal running log is obtained from the abnormal running data by an automated script. Based on the preset exceptions that occurred before the target application crashed, the number of times the preset exceptions occurred before the target application crashed is determined. The preset exceptions can be various exceptions set by the developers. In method (6), the exception stack key of the target application is extracted from the abnormal running data. The exception stack type is determined according to the matching rules between the exception stack key and the pre-defined exception stack key and exception stack type.

[0083] As can be seen, this embodiment can automatically determine the data indicator values ​​of abnormal operation data under the operation data indicator items of the target application, improving the efficiency of sample data generation. Furthermore, if there are multiple types of abnormal operation causes, the data indicator values ​​of the abnormal operation data corresponding to each type of abnormal operation cause can be determined under the aforementioned operation data indicator items.

[0084] Next, in step S106 above, the correspondence between the types of abnormal operation causes and the data indicator values ​​is determined. Based on this correspondence, the types of abnormal operation causes and the data indicator values ​​are used as training sample data for the model. Specifically, determining the correspondence between the types of abnormal operation causes and the data indicator values ​​involves:

[0085] (1) Obtain the correspondence between the types of abnormal operation causes and abnormal operation data, and the subordinate relationship between abnormal operation data and data indicator values;

[0086] (2) Based on the correspondence between the abnormal operation cause type and the abnormal operation data, and the subordinate relationship between the abnormal operation data and the data indicator value, determine the correspondence between the abnormal operation cause type and the data indicator value.

[0087] Specifically, when there are multiple types of abnormal operation reasons, each type has corresponding abnormal operation data, and each type of abnormal operation data has a data indicator value under the target application's operation data indicator item. Therefore, each type of abnormal operation data has mutually dependent data indicator values. Thus, in action (1), the correspondence between each abnormal operation reason type and each type of abnormal operation data, and the dependency relationship between each type of abnormal operation data and each type of abnormal operation data's data indicator value are obtained. In action (2), based on the obtained correspondence and dependency relationships, it can be determined that the abnormal operation reason type and the data indicator value of that type of abnormal operation data correspond to each other. Thus, when there are multiple abnormal operation reason types, the correspondence between the abnormal operation reason type and the data indicator value is determined, and mutually corresponding abnormal operation reason types and data indicator values ​​are obtained.

[0088] After determining the corresponding abnormal operation cause types and data indicator values, these types and values ​​are used as training sample data for the model. Specifically, for an abnormal operation cause type, if the corresponding abnormal operation data comes from multiple users, each user's abnormal operation data will correspond to a set of data indicator values. Therefore, this abnormal operation cause type can correspond to multiple sets of data indicator values. Each set of data indicator values ​​includes the numerical values ​​of the multiple operation data indicator items obtained through the automated script. An abnormal operation cause type and its corresponding set or multiple sets of data indicator values ​​constitute a training sample.

[0089] In summary, the method in this embodiment uses the abnormal operation cause type of the application and the data indicator value of the abnormal operation data under a predetermined operation data indicator item as model training sample data. There is a corresponding relationship between the abnormal operation cause type and the data indicator value. Therefore, when training the model using the generated model training sample data and analyzing the abnormal operation cause of the application, the correlation between the data indicator value of the abnormal operation data of the application and the abnormal operation cause type of the application is fully considered. Thus, for occasional application abnormalities, it is not necessary to rely on a massive amount of data on similar abnormalities to determine the abnormal operation cause of the application, thereby improving the accuracy of determining the abnormal operation cause of the application.

[0090] Furthermore, one embodiment of this specification also provides a model training method. Figure 2 This is a flowchart illustrating a model training method provided in one embodiment of this specification, as shown below. Figure 2 As shown, the process includes:

[0091] Step S202, obtain the above Figure 1 The method described above generates model training sample data;

[0092] Step S204: Use the model training sample data to train the pre-built neural network structure until it converges, and use the converged neural network structure as the model to determine the cause of abnormal operation of the target application.

[0093] The parameters in the neural network structure represent the contribution weights of each data indicator value corresponding to the same abnormal operation cause type.

[0094] In this embodiment, when training the model using the generated model training sample data and analyzing the causes of abnormal application operation through the model, the correlation between the data indicator values ​​of the abnormal application operation data and the types of abnormal application operation causes is fully considered. Therefore, for occasional application abnormalities, it is not necessary to rely on a large amount of data of similar abnormalities to determine the cause of the abnormal application operation, thus improving the accuracy of determining the cause of the abnormal application operation.

[0095] In step S202 above, the above-mentioned... Figure 1 The model training sample data generated by the method in the article includes corresponding abnormal operation cause types and data indicator values.

[0096] In step S204 above, the pre-built neural network structure is trained using model training sample data until it converges. The converged neural network structure is then used as the model to determine the cause of abnormal operation in the target application. This neural network structure can be a single-layer CNN network structure.

[0097] In one embodiment, the model training sample data is used to train a pre-built neural network structure until it converges. Specifically, each abnormal operation cause type and the corresponding data index value for each abnormal operation cause type are input into the neural network structure for training until the neural network structure converges.

[0098] As mentioned earlier, for a given type of abnormal operation, if the corresponding abnormal operation data comes from multiple users, then each user's abnormal operation data will correspond to a set of data indicator values. Therefore, this type of abnormal operation can correspond to multiple sets of data indicator values. Each set of data indicator values ​​includes the numerical values ​​of the multiple operation data indicator items obtained through the automated script. An abnormal operation cause type and its corresponding set or multiple sets of data indicator values ​​constitute a training sample. Multiple training samples are used to train a CNN neural network structure. The parameters in the neural network structure represent the contribution weights of each data indicator value corresponding to the same abnormal operation cause type for that same abnormal operation cause type.

[0099] For example, if there are 3 types of abnormal operation reasons and 6 operation data indicators, then for one type of abnormal operation reason, if it has abnormal operation data from 2 users, it corresponds to 2 sets of data indicator values, each set of data indicator values ​​has 6 values, corresponding to each operation data indicator item. The parameters in the neural network structure are used to represent the contribution weight of the data indicator value under each operation data indicator item to that abnormal operation reason type. The trained and converged neural network structure is the abnormal operation reason determination model for the target application.

[0100] Furthermore, one embodiment of this specification also provides a method for determining the cause of abnormal operation. Figure 3 This is a flowchart illustrating a method for determining the cause of abnormal operation provided in one embodiment of this specification, as shown below. Figure 3 As shown, the process includes:

[0101] Step S302, obtain as follows Figure 2 The method in the article trains a model to determine the cause of abnormal operation of the target application, and obtains the abnormal operation data of the target application to be analyzed.

[0102] Step S304: Determine the value of the data indicator to be analyzed under the running data indicator item for the abnormal running data to be analyzed;

[0103] Step S306: The abnormal operation cause determination model processes the indicator values ​​of the data to be analyzed, and the abnormal operation cause type of the target application is determined based on the processing result.

[0104] In this embodiment, when analyzing the causes of abnormal application operation using the generated model, the correlation between the data indicator values ​​of the abnormal operation data of the application and the types of abnormal operation causes of the application is fully considered. Therefore, for occasional application abnormalities, it is not necessary to rely on a large amount of data of similar abnormalities to determine the cause of the abnormal operation of the application, thereby improving the accuracy of determining the cause of the abnormal operation of the application.

[0105] In step S302 above, the model generated by the aforementioned method is obtained, and the abnormal operation data to be analyzed of the target application is obtained. The abnormal operation data to be analyzed can be the abnormal operation data reported by the user this time.

[0106] In step S304 above, the values ​​of the data indicators to be analyzed under the running data indicator items are determined using the methods described above. The specific process is described in the previous section and will not be repeated here.

[0107] In step S306 above, the data index value to be analyzed under the running data index item of the abnormal running data to be analyzed is input into the abnormal running cause determination model. The abnormal running cause determination model processes the data index value to be analyzed and outputs the processing result. The processing result can be the code of the abnormal running cause type corresponding to the abnormal running data to be analyzed. Based on the code, the mapping relationship between the code and the abnormal running cause type is found to determine the abnormal running cause type of the target application.

[0108] When the abnormal operation data to be analyzed comes from a single user, there is a corresponding set of data indicator values ​​to be analyzed. When the abnormal operation data to be analyzed comes from multiple users, each user corresponds to a set of data indicator values ​​to be analyzed, resulting in multiple sets of data indicator values ​​to be analyzed. One or more sets of data indicator values ​​to be analyzed can be input into the abnormal operation cause determination model for processing.

[0109] Furthermore, Figure 3 The process may also include: in the abnormal operation data to be analyzed, obtaining abnormal operation data associated with the indicator values ​​of the data to be analyzed, and outputting the associated abnormal operation data and the abnormal operation reason type of the target application.

[0110] Specifically, the data metrics to be analyzed are simple numerical values, such as memory level and exception stack type. From the exception runtime data to be analyzed, the exception runtime data associated with the data metrics is obtained. This associated exception runtime data is the data that generated the data metrics, or the raw data of the data metrics, such as specific memory size and exception stack keywords. This associated exception runtime data and the exception cause type of the target application are then output to the developers for analysis to further optimize the target application.

[0111] By employing the above methods for generating sample data, training models, and determining the causes of abnormal operation, the causes of occasional application anomalies can be determined without relying on massive amounts of data on similar anomalies, thus improving the accuracy of determining the causes of application anomalies.

[0112] This specification also provides a sample data generation apparatus in one embodiment. Figure 4 This is a schematic diagram of the structure of a sample data generation device provided in one embodiment of this specification, as shown below. Figure 4 As shown, the device includes:

[0113] The cause classification unit 41 is used to determine at least one cause of abnormal operation of the target application through the automated attribution model corresponding to the target application, classify the at least one cause of abnormal operation, and obtain the type of abnormal operation cause.

[0114] The indicator determination unit 42 is used to obtain abnormal operation data corresponding to the abnormal operation cause type from the historical abnormal operation data of the target application, and determine the data indicator value of the abnormal operation data under the preset operation data indicator item.

[0115] The sample generation unit 43 is used to determine the correspondence between the abnormal operation cause type and the data indicator value, and to use the abnormal operation cause type and the data indicator value as model training sample data according to the correspondence.

[0116] Optionally, the cause classification unit 41 is specifically used to: perform abnormal operation attribution on the target application through each attribution node in the automated attribution model to obtain the abnormal operation cause of the target application; wherein, each attribution node corresponds to a predetermined abnormal operation cause; the abnormal operation cause of the target application includes the predetermined abnormal operation cause.

[0117] Optionally, it further includes an indicator configuration unit, configured to: obtain at least one of the identifier of the target application, the version number of the target application, the identifier of the operating system to which the target application is adapted, and the version number of the operating system to which the target application is adapted; and configure the runtime indicator for the target application based on the obtained information.

[0118] Optionally, the indicator determination unit 42 is specifically used to: perform abnormal operation attribution on the historical abnormal operation data through each attribution node in the automated attribution model, so as to determine the data in the historical abnormal operation data corresponding to the predetermined abnormal operation reason corresponding to each attribution node; wherein, each attribution node corresponds to a predetermined abnormal operation reason; based on the predetermined abnormal operation reason corresponding to each attribution node, determine the target attribution node corresponding to the abnormal operation reason type in each attribution node, and take the data in the historical abnormal operation data corresponding to the predetermined abnormal operation reason corresponding to the target attribution node as the abnormal operation data corresponding to the abnormal operation reason type.

[0119] Optionally, the runtime data metrics include at least one of the following: the virtual memory level of the user terminal where the target application resides; the operating system version information of the user terminal where the target application resides; the exception signals generated and recorded in the runtime data of the target application according to the preset exception signal generation rules; the number of threads of the target application; the number of preset exceptions that occurred before the target application crashed; and the exception stack type of the target application.

[0120] Optionally, the indicator determination unit 42 is specifically used to: automatically determine the data indicator value of the abnormal running data under the running data indicator item through an automated script.

[0121] Optionally, the indicator determination unit 42 is further configured to: extract the virtual memory size of the user terminal from the abnormal operation data, and determine the virtual memory level based on the virtual memory size; extract the operating system version information from the abnormal operation data; extract the abnormal signal from the abnormal operation data; extract the number of threads of the target application from the abnormal operation data, and determine the thread count level based on the number of threads; determine the number of times a preset abnormality occurred before the target application crashed based on the abnormal operation log of the target application recorded in the abnormal operation data; extract the abnormal stack key of the target application from the abnormal operation data, and determine the abnormal stack type based on the abnormal stack key.

[0122] Optionally, the sample generation unit 43 is specifically used to: obtain the correspondence between the abnormal operation cause type and the abnormal operation data, and the subordinate relationship between the abnormal operation data and the data indicator value; and determine the correspondence between the abnormal operation cause type and the data indicator value based on the correspondence between the abnormal operation cause type and the abnormal operation data, and the subordinate relationship between the abnormal operation data and the data indicator value.

[0123] Optionally, the abnormal operation cause type includes at least one of the following types: the running memory of the user terminal where the target application is located is less than a preset minimum value; the file descriptors of the target application exceed a preset maximum value; the number of threads of the target application is more than a preset maximum number; the target application has a null pointer exception; the target application has a wild pointer causing abnormal memory access; the target application has a duplicate free causing abnormal memory access; the target application has a memory out-of-bounds exception causing abnormal memory access.

[0124] The sample data generation device in this embodiment can implement each process of the aforementioned sample data generation method and achieve the same effect and function, which will not be repeated here.

[0125] This specification also provides a model training device in one embodiment. Figure 5 This is a schematic diagram of the structure of a model training device provided in one embodiment of this specification, as shown below. Figure 5 As shown, the device includes:

[0126] Sample acquisition unit 51 is used to acquire model training sample data generated by the device as described in claim 14;

[0127] The model training unit 52 is used to train a pre-built neural network structure using the model training sample data until convergence, and to use the converged neural network structure as the model for determining the cause of abnormal operation of the target application.

[0128] The parameters in the neural network structure represent the contribution weights of each data indicator value corresponding to the same abnormal operation cause type to that same abnormal operation cause type.

[0129] Optionally, the model training unit 52 is specifically used to: input each of the abnormal operation cause types and each of the data indicator values ​​corresponding to each of the abnormal operation cause types into the neural network structure for training until the neural network structure converges.

[0130] The model training device in this embodiment can implement each process of the aforementioned model training method and achieve the same effect and function, which will not be repeated here.

[0131] One embodiment of this specification also provides a device for determining the cause of abnormal operation. Figure 6 This is a schematic diagram of the structure of a device for determining the cause of abnormal operation provided in one embodiment of this specification, as shown below. Figure 6 As shown, the device includes:

[0132] The model acquisition unit 61 is used to acquire the abnormal operation cause determination model of the target application trained by the device as described in claim 15, and to acquire the abnormal operation data of the target application to be analyzed.

[0133] Data determination unit 62 is used to determine the data indicator value to be analyzed under the operation data indicator item of the abnormal operation data to be analyzed;

[0134] The cause analysis unit 63 is used to process the data indicator values ​​to be analyzed through the abnormal operation cause determination model, and determine the abnormal operation cause type of the target application based on the processing result.

[0135] Optionally, it also includes an output unit, configured to: obtain abnormal operation data associated with the indicator value of the data to be analyzed from the abnormal operation data to be analyzed; and output the associated abnormal operation data and the abnormal operation reason type of the target application.

[0136] The device for determining the cause of abnormal operation in this embodiment can implement each process of the aforementioned method for determining the cause of abnormal operation and achieve the same effect and function, which will not be repeated here.

[0137] This specification also provides an electronic device according to one embodiment. Figure 7 This is a schematic diagram of the structure of an electronic device provided in one embodiment of this specification, such as... Figure 7 As shown, electronic devices can vary significantly due to differences in configuration or performance. They may include one or more processors 801 and memories 802, with the memory 802 storing one or more application programs or data. The memory 802 can be temporary or persistent storage. The application programs stored in the memory 802 may include one or more modules (not shown), each module including a series of computer-executable instructions within the electronic device. Furthermore, the processor 801 may be configured to communicate with the memory 802, executing the series of computer-executable instructions stored in the memory 802 on the electronic device. The electronic device may also include one or more power supplies 803, one or more wired or wireless network interfaces 804, one or more input or output interfaces 805, one or more keyboards 806, etc.

[0138] In one specific embodiment, the electronic device is a sample data generation device, specifically a backend server, including a processor; and a memory configured to store computer-executable instructions, which, when executed, cause the processor to perform the following process:

[0139] By using the automated attribution model corresponding to the target application, at least one cause of abnormal operation of the target application is determined, and the at least one cause of abnormal operation is classified to obtain the type of abnormal operation cause.

[0140] From the historical abnormal operation data of the target application, obtain the abnormal operation data corresponding to the abnormal operation reason type, and determine the data indicator value of the abnormal operation data under the preset operation data indicator item;

[0141] Determine the correspondence between the abnormal operation cause type and the data indicator value, and use the abnormal operation cause type and the data indicator value as model training sample data according to the correspondence.

[0142] The sample data generation device in the embodiments of this specification can realize each process of the aforementioned sample data generation method and achieve the same effect and function, which will not be repeated here.

[0143] In another specific embodiment, the electronic device is a model training device, specifically a backend server, including a processor; and a memory configured to store computer-executable instructions, which, when executed, cause the processor to perform the following process:

[0144] Obtain the model training sample data generated by the aforementioned sample data generation method;

[0145] Using the model training sample data, a pre-built neural network structure is trained until convergence. The converged neural network structure is used as the model for determining the cause of abnormal operation of the target application.

[0146] The parameters in the neural network structure represent the contribution weights of each data indicator value corresponding to the same abnormal operation cause type to that same abnormal operation cause type.

[0147] The model training device described in this specification can implement all the processes of the aforementioned model training method and achieve the same effect and function, which will not be repeated here.

[0148] In yet another specific embodiment, the electronic device is an abnormal operation cause determination device, specifically a backend server, including a processor; and a memory configured to store computer-executable instructions, which, when executed, cause the processor to perform the following process:

[0149] Obtain the abnormal operation cause determination model of the target application trained by the model training method described above, and obtain the abnormal operation data of the target application to be analyzed.

[0150] Determine the value of the data indicator to be analyzed under the operation data indicator item for the abnormal operation data to be analyzed;

[0151] The abnormal operation cause determination model processes the data indicator values ​​to be analyzed, and determines the type of abnormal operation cause of the target application based on the processing results.

[0152] The abnormal operation cause determination device in the embodiments of this specification can realize each process of the aforementioned abnormal operation cause determination method and achieve the same effect and function, which will not be repeated here.

[0153] Another embodiment of this specification provides a computer-readable storage medium for storing computer-executable instructions that, when executed by a processor, implement the following process:

[0154] By using the automated attribution model corresponding to the target application, at least one cause of abnormal operation of the target application is determined, and the at least one cause of abnormal operation is classified to obtain the type of abnormal operation cause.

[0155] From the historical abnormal operation data of the target application, obtain the abnormal operation data corresponding to the abnormal operation reason type, and determine the data indicator value of the abnormal operation data under the preset operation data indicator item;

[0156] Determine the correspondence between the abnormal operation cause type and the data indicator value, and use the abnormal operation cause type and the data indicator value as model training sample data according to the correspondence.

[0157] The storage medium in the embodiments of this specification can implement each process of the aforementioned sample data generation method and achieve the same effect and function, which will not be repeated here.

[0158] Another embodiment of this specification provides a computer-readable storage medium for storing computer-executable instructions that, when executed by a processor, implement the following process:

[0159] Obtain the model training sample data generated by the aforementioned sample data generation method;

[0160] Using the model training sample data, a pre-built neural network structure is trained until convergence. The converged neural network structure is used as the model for determining the cause of abnormal operation of the target application.

[0161] The parameters in the neural network structure represent the contribution weights of each data indicator value corresponding to the same abnormal operation cause type to that same abnormal operation cause type.

[0162] The storage medium in the embodiments of this specification can realize each process of the aforementioned model training method and achieve the same effect and function, which will not be repeated here.

[0163] Another embodiment of this specification provides a computer-readable storage medium for storing computer-executable instructions that, when executed by a processor, implement the following process:

[0164] Obtain the abnormal operation cause determination model of the target application trained by the model training method described above, and obtain the abnormal operation data of the target application to be analyzed.

[0165] Determine the value of the data indicator to be analyzed under the operation data indicator item for the abnormal operation data to be analyzed;

[0166] The abnormal operation cause determination model processes the data indicator values ​​to be analyzed, and determines the type of abnormal operation cause of the target application based on the processing results.

[0167] The storage medium in the embodiments of this specification can implement each process of the aforementioned method for determining the cause of abnormal operation, and achieve the same effect and function, which will not be repeated here.

[0168] In various embodiments of this specification, the computer-readable storage medium includes read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk, etc.

[0169] In the 1990s, improvements to a technology could be clearly distinguished as either hardware improvements (e.g., improvements to the circuit structure of diodes, transistors, switches, etc.) or software improvements (improvements to the methodology). However, with technological advancements, many methodological improvements today can be considered direct improvements to the hardware circuit structure. Designers almost always obtain the corresponding hardware circuit structure by programming the improved methodology into the hardware circuit. Therefore, it cannot be said that a methodological improvement cannot be implemented using a hardware physical module. For example, a Programmable Logic Device (PLD) (e.g., a Field Programmable Gate Array (FPGA)) is such an integrated circuit whose logic function is determined by the user programming the device. Designers can program a digital system themselves to "integrate" it onto a PLD, without needing chip manufacturers to design and manufacture dedicated integrated circuit chips. Furthermore, nowadays, instead of manually manufacturing integrated circuit chips, this programming is mostly implemented using "logic compiler" software. Similar to the software compiler used in program development, the original code before compilation must be written in a specific programming language, called a Hardware Description Language (HDL). There are many HDLs, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, and RHDL (Ruby Hardware Description Language). Currently, the most commonly used are VHDL (Very-High-Speed ​​Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should understand that by simply performing some logic programming on the method flow using one of these hardware description languages ​​and programming it into an integrated circuit, the hardware circuit implementing the logical method flow can be easily obtained.

[0170] The controller can be implemented in any suitable manner. For example, it can take the form of a microprocessor or processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) executable by the (micro)processor, logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers, and embedded microcontrollers. Examples of controllers include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicon Labs C8051F320. A memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art will also recognize that, in addition to implementing the controller in purely computer-readable program code form, the same functionality can be achieved by logically programming the method steps to make the controller take the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers. Therefore, such a controller can be considered a hardware component, and the means included therein for implementing various functions can also be considered as structures within the hardware component. Alternatively, the means for implementing various functions can be considered as both software modules implementing the method and structures within the hardware component.

[0171] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, a computer can be, for example, a personal computer, laptop computer, cellular phone, camera phone, smartphone, personal digital assistant, media player, navigation device, email device, game console, tablet computer, wearable device, or any combination of these devices.

[0172] For ease of description, the above apparatus is described by dividing it into various functional units. Of course, when implementing the embodiments of this specification, the functions of each unit can be implemented in one or more software and / or hardware.

[0173] Those skilled in the art will understand that one or more embodiments of this specification can be provided as a method, system, or computer program product. Therefore, one or more embodiments of this specification may take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, one or more embodiments of this specification may take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0174] This specification is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this specification. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create a machine for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0175] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0176] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0177] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0178] One or more embodiments of this specification can be described in the general context of computer-executable instructions, such as program modules, that are executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform a particular task or implement a particular abstract data type. One or more embodiments of this specification can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0179] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0180] The above description is merely an embodiment of this document and is not intended to limit the scope of this document. Various modifications and variations can be made to this document by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this document should be included within the scope of the claims of this document.

Claims

1. A method for generating sample data, characterized in that, include: By using the automated attribution model corresponding to the target application, at least one cause of abnormal operation of the target application is determined, and the at least one cause of abnormal operation is classified to obtain the type of abnormal operation cause. From the historical abnormal operation data of the target application, obtain the abnormal operation data corresponding to the abnormal operation reason type, and determine the data indicator value of the abnormal operation data under the preset operation data indicator item; Determine the correspondence between the abnormal operation cause type and the data indicator value, and use the abnormal operation cause type and the data indicator value as model training sample data according to the correspondence.

2. The method according to claim 1, characterized in that, The cause of the abnormal operation of the target application is determined using the automated attribution model corresponding to the target application, including: The abnormal operation of the target application is attributed to each attribution node in the automated attribution model to obtain the cause of the abnormal operation of the target application. Each of the attribution nodes corresponds to a predetermined abnormal operation reason; the abnormal operation reason of the target application includes the predetermined abnormal operation reason.

3. The method according to claim 1, characterized in that, The method further includes: Obtain at least one of the following: the identifier of the target application, the version number of the target application, the identifier of the operating system to which the target application is adapted, and the version number of the operating system to which the target application is adapted; Based on the acquired information, configure the runtime data metrics for the target application.

4. The method according to claim 1, characterized in that, From the historical abnormal operation data of the target application, obtain the abnormal operation data corresponding to the abnormal operation reason type, including: The historical abnormal operation data is attributed to abnormal operation through each attribution node in the automated attribution model, so as to determine the data in the historical abnormal operation data corresponding to the predetermined abnormal operation reason corresponding to each attribution node; wherein, each attribution node corresponds to a predetermined abnormal operation reason. Based on the predetermined abnormal operation reason corresponding to each attribution node, a target attribution node corresponding to the abnormal operation reason type is determined among each attribution node, and the data in the historical abnormal operation data corresponding to the predetermined abnormal operation reason corresponding to the target attribution node is taken as the abnormal operation data corresponding to the abnormal operation reason type.

5. The method according to claim 1, characterized in that, The operational data metrics include at least one of the following: The virtual memory level of the user terminal where the target application resides; The operating system version information of the user terminal where the target application is located; The abnormal signals generated and recorded in the runtime data of the target application according to the preset abnormal signal generation rules; The number of threads in the target application; The target application was said to have encountered a preset number of exceptions before crashing. The exception stack type of the target application.

6. The method according to claim 5, characterized in that, Determining the data indicator values ​​of the abnormal operating data under the operating data indicator item includes: The abnormal operating data is automatically determined using automated scripts under the operating data indicator item.

7. The method according to claim 6, characterized in that, The abnormal operating data is automatically determined using automated scripts under the operating data indicator item, including at least one of the following methods: Extract the virtual memory size of the user terminal from the abnormal operation data, and determine the virtual memory level based on the virtual memory size; Extract the operating system version information from the abnormal operation data; Extract the abnormal signal from the abnormal operation data; Extract the number of threads of the target application from the abnormal operation data, and determine the thread count level based on the number of threads; Based on the abnormal operation logs of the target application recorded in the abnormal operation data, determine the number of times a preset abnormality occurred before the target application crashed; Extract the exception stack keywords of the target application from the abnormal operation data, and determine the exception stack type based on the exception stack keywords.

8. The method according to claim 1, characterized in that, Determining the correspondence between the types of abnormal operation causes and the data indicator values ​​includes: Obtain the correspondence between the abnormal operation cause type and the abnormal operation data, and the subordinate relationship between the abnormal operation data and the data indicator value; Based on the correspondence between the abnormal operation cause type and the abnormal operation data, and the subordinate relationship between the abnormal operation data and the data indicator value, the correspondence between the abnormal operation cause type and the data indicator value is determined.

9. The method according to claim 1, characterized in that, The abnormal operation cause type includes at least one of the following types: The user terminal where the target application is located has less than a preset minimum memory limit; The target application's file descriptors exceed the preset maximum value; The target application has more threads than the preset maximum number; The target application contains a null pointer exception; The target application contains dangling pointers that cause abnormal memory access. The target application exhibits abnormal memory access caused by repeated memory releases. The target application is experiencing memory out-of-bounds access, causing abnormal memory access.

10. A model training method, characterized in that, include: Obtain the model training sample data generated by the method described in any one of claims 1-9; Using the model training sample data, a pre-built neural network structure is trained until convergence. The converged neural network structure is used as the model for determining the cause of abnormal operation of the target application. The parameters in the neural network structure represent the contribution weights of each data indicator value corresponding to the same abnormal operation cause type to that same abnormal operation cause type.

11. The method according to claim 10, characterized in that, Using the model to train sample data, the pre-built neural network structure is trained until convergence, including: Each of the abnormal operation cause types and the corresponding data indicator values ​​are input into the neural network structure for training until the neural network structure converges.

12. A method for determining the cause of abnormal operation, characterized in that, include: Obtain the abnormal operation cause determination model of the target application trained by the method as described in claim 10 or 11, and obtain the abnormal operation data of the target application to be analyzed; Determine the value of the data indicator to be analyzed under the operation data indicator item for the abnormal operation data to be analyzed; The abnormal operation cause determination model processes the data indicator values ​​to be analyzed, and determines the type of abnormal operation cause of the target application based on the processing results.

13. The method according to claim 12, characterized in that, Also includes: From the abnormal operation data to be analyzed, obtain the abnormal operation data associated with the indicator values ​​of the data to be analyzed; Output the associated abnormal operation data and the type of abnormal operation reason for the target application.

14. A sample data generation device, characterized in that, include: The cause classification unit is used to determine at least one cause of abnormal operation of the target application through the automated attribution model corresponding to the target application, and classify the at least one cause of abnormal operation to obtain the cause type of abnormal operation. The indicator determination unit is used to obtain abnormal operation data corresponding to the abnormal operation cause type from the historical abnormal operation data of the target application, and determine the data indicator value of the abnormal operation data under the preset operation data indicator item. The sample generation unit is used to determine the correspondence between the abnormal operation cause type and the data indicator value, and to use the abnormal operation cause type and the data indicator value as model training sample data according to the correspondence.

15. A model training device, characterized in that, include: A sample acquisition unit is used to acquire model training sample data generated by the device as described in claim 14; The model training unit is used to train a pre-built neural network structure using the model training sample data until it converges, and to use the converged neural network structure as the model for determining the cause of abnormal operation of the target application. The parameters in the neural network structure represent the contribution weights of each data indicator value corresponding to the same abnormal operation cause type to that same abnormal operation cause type.

16. A device for determining the cause of abnormal operation, characterized in that, include: The model acquisition unit is used to acquire the abnormal operation cause determination model of the target application trained by the device as described in claim 15, and to acquire the abnormal operation data of the target application to be analyzed. A data determination unit is used to determine the data indicator value to be analyzed under the operation data indicator item for the abnormal operation data to be analyzed. The cause analysis unit is used to process the data indicator values ​​to be analyzed through the abnormal operation cause determination model, and determine the abnormal operation cause type of the target application based on the processing result.

17. An electronic device, characterized in that, include: processor; as well as, A memory configured to store computer-executable instructions, which, when executed, cause the processor to perform the steps of the method according to any one of claims 1-9, or the steps of the method according to claim 10 or 11, or the steps of the method according to claim 12 or 13.

18. A computer-readable storage medium, characterized in that, The computer-readable storage medium is used to store computer-executable instructions that, when executed by a processor, implement the steps of the method according to any one of claims 1-9, or the steps of the method according to claim 10 or 11, or the steps of the method according to claim 12 or 13.

Images