Data analysis method, apparatus, medium, device, and program product

Abstract

Embodiments of the present application disclose a data analysis method, device, medium, equipment and program product. A plurality of fields and corresponding field values of online output data of the same data source are obtained, the field names are input into a trained random number noise prediction model, and a random number noise prediction result is obtained. The field values are subjected to difference quantity analysis, and a random number noise analysis result is determined. The online output data and offline test services are used for noise analysis, and an environmental noise analysis result of each field is determined. The random number noise prediction, analysis result and environmental noise analysis result of each field are comprehensively determined, a noise determination result is determined, and playback testing is performed accordingly. Whether a field is a random number or environmental noise field is analyzed from multiple dimensions, the noise determination accuracy is improved, the accuracy of data analysis is improved, and the labor cost is saved.

Description

Technical Field

[0001] This application relates to the field of data detection technology, specifically to a data analysis method, apparatus, medium, equipment, and program product. Background Technology

[0002] With the development of computer and internet technology, the types of software services are increasing, and these services require various upgrades and modifications based on actual application needs. Generally, software services need to be tested before going live or after version upgrades. Only after passing the tests can the service be released to ensure its reliability. Traffic replay testing is a consistency testing method. It involves inputting the same online input data as the online version into the software service in the offline test environment, obtaining the offline output data, and comparing the online output data with the offline output data to determine if there are any anomalies in the software service in the test environment.

[0003] In related technologies, due to the interference of random number noise and environmental noise factors, there may be some inconsistencies between online and offline output data. To address this, certain noisy fields are usually marked from the output data by manual annotation, so that the comparison of noisy fields can be ignored. However, manual annotation may mark non-random number noise fields that are susceptible to environmental noise interference as non-noise fields, causing such fields to be included in the playback test, resulting in low accuracy of the playback test. Summary of the Invention

[0004] This application provides a data analysis method, apparatus, medium, device, and program product that can improve the accuracy of noise field determination, thereby enhancing the precision of data analysis, avoiding manual annotation, and saving labor costs.

[0005] To address the aforementioned technical problems, the embodiments of this application provide the following technical solutions:

[0006] A data analysis method, comprising:

[0007] Obtain multiple fields included in online output data from the same data source, and the field value corresponding to each field in each online output data;

[0008] The field name of each field is input into the trained random number noise prediction model to predict the random number noise prediction result for each field.

[0009] Perform a field value difference quantity analysis on multiple field values ​​corresponding to each field to determine the random number noise analysis result for each field;

[0010] Noise analysis is performed using the online output data and offline testing services to determine the environmental noise analysis results for each field.

[0011] Based on the random number noise prediction results, random number noise analysis results, and environmental noise analysis results for each field, a noise determination result for each field is determined, so as to perform playback tests based on the noise determination result for each field.

[0012] A data analysis device, comprising:

[0013] The first acquisition unit is used to acquire multiple fields included in online output data from the same data source, and the field value corresponding to each field in each online output data.

[0014] The prediction unit is used to input the field name of each field into the trained random number noise prediction model and predict the random number noise prediction result for each field.

[0015] The first determining unit is used to perform field value difference quantity analysis on multiple field values ​​corresponding to each field, and determine the random number noise analysis result of each field.

[0016] The second determining unit is used to perform noise analysis using the online output data and offline testing services to determine the environmental noise analysis result for each field.

[0017] The third determining unit is used to determine the noise judgment result of each field based on the random number noise prediction result, random number noise analysis result, and environmental noise analysis result of each field, so as to perform playback test according to the noise judgment result of each field.

[0018] In some embodiments, the first determining unit includes:

[0019] The first acquisition subunit is used to acquire the number of field value differences for each of the fields;

[0020] The first determining subunit is used to determine the ratio of the number of field value differences for each field to the total number of online output data, thereby obtaining the percentage of differences for each field.

[0021] The second determining subunit is used to determine the target difference quantity percentage interval of each field from multiple difference quantity percentage intervals corresponding to each field.

[0022] The third determining subunit is used to determine the target random number noise analysis result corresponding to the target difference quantity ratio interval of each of the fields as the corresponding random number noise analysis result.

[0023] In some implementations, the first acquisition subunit is configured to:

[0024] For each of the aforementioned fields, obtain the values ​​of multiple fields to be analyzed;

[0025] Select a current field value from multiple field values ​​in sequence, and compare the current field value with each field value in the historical analysis field value set;

[0026] When the current field value differs from the field value in the historical analysis field value set, increment the field value difference count by one;

[0027] Assign the current field value to the set of historical analysis field values;

[0028] When there are unselected field values, return to the step of selecting a current field value from multiple field values ​​in sequence until there are no unselected field values, and obtain the number of field value differences for each field.

[0029] In some embodiments, the second determining unit includes:

[0030] The input subunit is used to input the online input data corresponding to each of the online output data to the offline testing service to obtain the offline output data corresponding to each of the online input data.

[0031] The comparison sub-unit is used to compare the field values ​​of the online output data corresponding to the input data on the same line and the corresponding offline output data in each field to determine the environmental noise analysis result of each field.

[0032] In some embodiments, the comparison subunit includes:

[0033] The first comparison subunit is used to compare the field values ​​of the online output data corresponding to the online input data and the corresponding offline output data in each field.

[0034] The fourth determining subunit is used to determine the environmental noise analysis result of the first field as the environmental noise field when the field value of the online output data of the first field is different from the field value of the corresponding offline output data.

[0035] The fifth determining subunit is used to determine the environmental noise analysis result of the second field in the online output data other than the first field as a non-environmental noise field.

[0036] In some embodiments, the third determining unit includes:

[0037] The second comparison subunit is used to compare the random number noise prediction results and the corresponding random number noise analysis results of each field to determine the random number noise comparison result of each field.

[0038] The sixth determining subunit is used to determine the noise judgment result of each field based on the random number noise comparison result of each field and the corresponding environmental noise analysis result.

[0039] In some implementations, the second comparison subunit is used for:

[0040] When there is a difference between the corresponding random number noise prediction result and the corresponding random number noise analysis result, and the corresponding random number noise analysis result is not the undetermined third field, the random number noise analysis result is determined as the random number noise comparison result of the third field.

[0041] When there is a difference between the corresponding random number noise prediction result and the corresponding random number noise analysis result, and the corresponding random number noise analysis result is a fourth field to be determined, the random number noise prediction result is determined as the random number noise comparison result of the fourth field.

[0042] In some implementations, the sixth determining subunit is used for:

[0043] When there is a corresponding random number noise comparison result that is the fifth field of the random number noise field, the random number noise field is determined as the noise determination result of the fifth field;

[0044] When there is a corresponding random number noise comparison result that is a non-random number noise field, and the corresponding environmental noise analysis result is the sixth field of the environmental noise field, the environmental noise field is determined as the noise judgment result of the sixth field.

[0045] When there is a corresponding random number noise comparison result that is a non-random number noise field, and the corresponding environmental noise analysis result is the seventh field of the non-environmental noise field, the non-random number noise field and the non-environmental noise field are determined as the noise judgment result of the seventh field.

[0046] In some embodiments, the apparatus further includes:

[0047] The matching unit is used to match the field values ​​of multiple fields in the output data on the same line;

[0048] The first generation unit is used to generate matching conditions when at least one field value in the online output data matches the field value in the ninth field.

[0049] The second acquisition unit is used to acquire the number of online output data that meet the matching conditions;

[0050] The calculation unit is used to determine the ratio of the number of conforming data to the total number of online output data, and to obtain the proportion of conforming data.

[0051] The second generation unit is used to generate matching relationship information when the proportion of the number of matching items reaches a preset proportion threshold. The matching relationship information is used to indicate that the field value of the eighth field matches the field value of the ninth field.

[0052] The third determining unit includes:

[0053] The playback test subunit is used to perform playback tests based on the noise determination results of each field and the matching relationship information.

[0054] In some implementations, the first generating unit includes:

[0055] The first generation subunit is used to generate a matching condition that the field value of the eighth field is the same as the field value of the ninth field when at least one online output data contains a field value of the eighth field that is the same as the field value of the ninth field.

[0056] The first acquisition subunit is used to acquire the position of the part of the field value of the ninth field that is the same as the field value of the eighth field when at least one field value in the online output data is the same as part of the field value of the ninth field.

[0057] The second generation subunit is used to generate matching conditions where the field value of the eighth field is the same as the field value of the ninth field in the same position of the part.

[0058] In some embodiments, the first acquisition unit includes:

[0059] The second acquisition subunit is used to acquire multiple online output data from the same data source;

[0060] The seventh determination sub-unit is used to determine multiple fields with the same field type in multiple online output data;

[0061] Extract sub-units, used to extract the field values ​​corresponding to each field from multiple online output data according to the dimensions of each field.

[0062] A computer-readable storage medium storing a plurality of instructions adapted for loading by a processor to execute the data analysis method described above.

[0063] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-described data analysis method.

[0064] A computer program product or computer program includes computer instructions stored in a storage medium. A processor of a computer device reads the computer instructions from the storage medium and executes the computer instructions to implement the aforementioned data analysis method.

[0065] This application embodiment obtains multiple fields from online output data of the same data source, and the field value corresponding to each field in each online output data; inputs the field name of each field into a trained random number noise prediction model to predict the random number noise prediction result of each field; performs field value difference quantity analysis on the multiple field values ​​corresponding to each field to determine the random number noise analysis result of each field; performs noise analysis using the online output data and offline testing services to determine the environmental noise analysis result of each field; and determines the noise judgment result of each field based on the random number noise prediction result, random number noise analysis result, and environmental noise analysis result of each field, so as to perform playback testing based on the noise judgment result of each field.

[0066] Therefore, by inputting field names into a random number noise prediction model to obtain the random number noise prediction results for a field, performing a difference quantity analysis on field values ​​to obtain the random number noise analysis results for the field, and using online output data and offline testing services to perform noise analysis, the environmental noise analysis results for each field are determined. In this way, a comprehensive evaluation is conducted from multiple dimensions to determine whether a field is a noise field and the specific noise type of the noise field, making the determination of noise fields more accurate. Compared with related technologies, which rely on manual annotation to directly label inconsistent fields as noise fields, this application embodiment can comprehensively analyze fields from multiple dimensions to accurately determine whether each field is a random number noise field and whether it is an environmental noise field. This avoids labeling non-random number noise fields that are easily affected by environmental noise as non-noise fields, improves the accuracy of noise field determination, and thus improves the precision of data analysis, avoids manual annotation, and saves labor costs.

[0067] Other features and advantages of this disclosure will be set forth in the following description and will be apparent in part from the description or may be learned by practicing the disclosure. The objectives and other advantages of this disclosure may be realized and obtained by means of the structures particularly pointed out in the description, claims and drawings. Attached Figure Description

[0068] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0069] Figure 1 This is a schematic diagram of a data analysis system provided in an embodiment of this application.

[0070] Figure 2 This is a flowchart illustrating the data analysis method provided in the embodiments of this application.

[0071] Figure 3 This is a schematic diagram of the data acquisition process provided in an embodiment of this application.

[0072] Figure 4 This is a schematic diagram of the random number noise prediction process provided in an embodiment of this application.

[0073] Figure 5 This is a schematic diagram illustrating the process of analyzing the number of field value differences provided in an embodiment of this application.

[0074] Figure 6 This is a schematic diagram of the environmental noise analysis process provided in an embodiment of this application.

[0075] Figure 7 This is a schematic diagram illustrating the determination of noise determination results provided in an embodiment of this application.

[0076] Figure 8 This is a schematic diagram illustrating the matching relationship analysis provided in an embodiment of this application.

[0077] Figure 9 This is a schematic diagram showing the noise determination results and matching conditions for each field provided in the embodiments of this application.

[0078] Figure 10 This is a schematic diagram of a replay test of a field comparing the version of the business service under development with the online business service version in the online environment, provided as an embodiment of this application.

[0079] Figure 11 This is a schematic diagram of the system framework of the data analysis system provided in the embodiments of this application.

[0080] Figure 12 This is a schematic diagram of the structure of the data analysis device provided in the embodiments of this application.

[0081] Figure 13 This is a schematic diagram of the structure of the computer device provided in the embodiments of this application. Detailed Implementation

[0082] To enable those skilled in the art to better understand the solutions of this application, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0083] It is understood that in the specific implementation of this application, data such as online output data, online input data and offline output data are involved. When the above embodiments of this application are applied to specific products or technologies, permission or consent from the subject is required, and the collection, use and processing of related data must comply with relevant laws, regulations and standards.

[0084] Furthermore, when this application embodiment needs to obtain data such as online output data, online input data, and offline output data, it will obtain separate permission or separate consent for the data such as online output data, online input data, and offline output data through pop-up windows or redirection to a confirmation page. After clearly obtaining separate permission or separate consent for the data such as the task, it will then obtain the necessary task-related data for the normal operation of this application embodiment.

[0085] It should be noted that while some processes described in the specification, claims, and accompanying drawings contain multiple steps that appear in a specific order, it should be clearly understood that these steps may not be performed in the order they appear herein, or may be performed in parallel. The step numbers are merely used to distinguish different steps and do not represent any particular order of execution. Furthermore, descriptions such as "first," "second," or "objective" in this document are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0086] Before providing a further detailed description of the embodiments of this disclosure, the terms and concepts used in these embodiments are explained, and they are subject to the following interpretations:

[0087] Replay testing: Generally used for traffic replay, it is a software testing method that plays a crucial role in the software testing process. By simulating real online business scenarios, it conducts comprehensive functional verification and performance evaluation of the development version of the business system.

[0088] Traffic replay testing refers to collecting real traffic data generated during the online operation of a business system during software testing. This data includes input requests and corresponding output responses from the business system. In subsequent testing phases, the collected online input data is resent to the development version of the business system, simulating the execution process of the online business in the test environment. The output generated by the development version is then compared with the original online output to determine whether the functionality of the development version is consistent with the online version.

[0089] After upgrading business systems, fixing vulnerabilities, or refactoring code, replay testing can quickly verify whether existing functions have been affected. Similarly, after an e-commerce platform upgrades its product recommendation algorithm, replay testing can verify the accuracy of recommendation results and the proper functioning of product detail pages.

[0090] Online output data refers to the data generated and output by the business service system in its online operation version. Taking an e-commerce system as an example, the order details and product inventory update information returned by the system after a successful order placement are all online output data.

[0091] Fields: These are the basic building blocks of online output data, used to describe a specific characteristic of the data. In order data, "order number," "product name," and "quantity purchased" are all fields.

[0092] Field value: refers to the data content corresponding to the field. For example, the value of the "Order Number" field is "202410100001" and the value of the "Quantity Purchased" field is "2".

[0093] Random number noise field: This field is used in traffic playback test result comparison because it contains random identifier field values ​​and / or time identifier field values, which interfere with the comparison and cause inconsistencies in the data comparison.

[0094] In business systems, random numbers are often used to meet requirements such as unique identifier generation, encryption security, and simulating uncertainty in real-world scenarios. For example, e-commerce platforms add random numbers to order numbers to ensure uniqueness and a certain level of security. These fields containing random numbers can easily become sources of interference during traffic replay testing.

[0095] In traffic replay testing, the random numbers generated in the online and test environments often differ. For example, the order number generated online might be "20241010001RAND1", while the one generated in the test environment might be "20241010001RAND2", where "RAND1" and "RAND2" are the random number components. When comparing the output data from the online and test environments, the difference in the random numbers within the order numbers leads to inconsistencies in this field. Without special handling, this inconsistency can affect the assessment of the integrity of the business system's functionality, misleading testers into believing that there is a problem with the business function.

[0096] Environmental noise fields refer to discrepancies in certain fields between online and offline test output data caused by differences between the test environment and the actual online operating environment during traffic replay testing.

[0097] Online servers may employ high-performance server clusters with ample memory, multi-core CPUs, and high-speed storage. Test environments, however, may be limited by cost and use lower-configuration servers. For example, an online server might have 64GB of memory, while a test server might only have 16GB. This difference in hardware configuration can lead to variations in system performance when processing data, thus affecting the output of certain fields.

[0098] Furthermore, online environments typically have stable, high-speed network connections to ensure smooth business operations. However, test environments may suffer from bandwidth limitations and higher latency. For example, an online network might have 1000Mbps bandwidth and 10ms latency, while a test environment might have only 100Mbps bandwidth and 50ms latency. In business scenarios involving data transmission and real-time interaction, these differences in network environment can affect data transmission speed and integrity, leading to inconsistencies in some fields.

[0099] Environmental noise fields can interfere with the assessment of the functional integrity of a business system during traffic replay testing. Because of these fields, testers find it difficult to accurately determine whether they are inherently random noise fields that do not participate in the replay test, or whether they are non-random noise fields that are only affected by environmental noise.

[0100] To address the aforementioned problems, this application proposes a data analysis method. This method involves obtaining multiple fields from online output data originating from the same data source, and the corresponding field value for each field in each online output data. The field name of each field is input into a trained random number noise prediction model to predict the random number noise prediction result for each field. A field value difference quantity analysis is performed on the multiple field values ​​corresponding to each field to determine the random number noise analysis result for each field. Noise analysis is conducted using the online output data and an offline testing service to determine the environmental noise analysis result for each field. Based on the random number noise prediction result, random number noise analysis result, and environmental noise analysis result for each field, a noise judgment result for each field is determined, and playback testing is performed based on the noise judgment result for each field.

[0101] In this way, a comprehensive evaluation is conducted from multiple dimensions to determine whether a field is a noise field and the specific noise type of the noise field. This determines whether each field is a random number noise field or an environmental noise field, avoiding the labeling of non-random number noise fields affected by environmental noise as noise fields. This improves the accuracy of noise identification, avoids manual labeling, and saves manual workload and time. Please refer to the specific implementation examples below for the detailed implementation process.

[0102] Please see Figure 1 , Figure 1 This is a schematic diagram of a data analysis system provided in an embodiment of this application. It includes an online server 140, an internet connection 130, a gateway 120, and offline servers 110, etc.

[0103] Online server 140 is configured with a business service system of an already launched business service version, used to respond to business requests initiated by clients and execute corresponding processing functions. Online server 140 refers to a computer system that can provide certain services to offline server 110. Online server 140 can communicate with Internet 130 via wired or wireless means to exchange data.

[0104] For different replay test scenarios, the offline server 110 can be configured with a business service system in development that is yet to be launched, or with a business service system that is the same as the business service system already launched on the online server 140. The offline server 110 and the online server 140 can be a single high-performance computer, a cluster of multiple high-performance computers, a portion of a single high-performance computer (e.g., a virtual machine), or a combination of portions of multiple high-performance computers (e.g., virtual machines) in a network platform.

[0105] Gateway 120, also known as an internetwork connector or protocol converter, is a computer system or device that acts as a translator, enabling network interconnection at the transport layer. It bridges the gap between two systems using different communication protocols, data formats, languages, or even completely different architectures. Gateways can also provide filtering and security functions. Data sent from online server 140 to offline server 110 must pass through gateway 120 to the corresponding offline server 110.

[0106] The data analysis method of this disclosure embodiment can be implemented on an offline server 110.

[0107] When the data analysis method is implemented on the offline server 110, the offline server 110 obtains multiple fields included in the online output data from the same data source, as well as the field value corresponding to each field in each of the online output data; the field name of each field is input into the trained random number noise prediction model to predict the random number noise prediction result of each field; the field value difference quantity analysis is performed on the multiple field values ​​corresponding to each field to determine the random number noise analysis result of each field; noise analysis is performed using the online output data and the offline test service to determine the environmental noise analysis result of each field; based on the random number noise prediction result, random number noise analysis result, and environmental noise analysis result of each field, the noise judgment result of each field is determined, so as to perform replay testing according to the noise judgment result of each field.

[0108] It should be noted that, Figure 1 The schematic diagram of the data analysis system shown is merely an example. The data analysis system and scenario described in this application are intended to more clearly illustrate the technical solutions of this application and do not constitute a limitation on the technical solutions provided in this application. As those skilled in the art will know, with the evolution of data analysis and the emergence of new business scenarios, the technical solutions provided in this application are also applicable to similar technical problems.

[0109] In this embodiment, the description will be from the perspective of a data analysis device. Specifically, the data analysis device can be integrated into a computer device that has a storage unit and a microprocessor installed, thus having computing power. The computer device can be a server, and in this embodiment, the computer device will be described as a server.

[0110] Please see Figure 2 , Figure 2 This is a flowchart illustrating the data analysis method provided in an embodiment of this application. The data analysis method is applied to an offline server and includes:

[0111] In step 201, multiple fields included in the online output data from the same data source are obtained, as well as the field value corresponding to each field in each online output data.

[0112] Among them, such as Figure 3 As shown, Figure 3 This is a schematic diagram illustrating the data collection process provided in this embodiment. The offline server collects online output data from the online server using a data collector and stores it in local storage. Specifically, business-side technical personnel analyze the output of the business service system. The data collector continuously samples and pulls the output of the business system according to time periods, marking each data entry with a data source identifier. This identifier ensures that only fields from online output data with the same data source can be compared during subsequent playback testing. Therefore, each stored online output data entry contains a data source identifier and the field value of the specific field.

[0113] For example, consider a business scenario where a request is sent over the network to an already deployed version of a business service system. After processing, the data is written to the database, and a response is returned over the network. The request represents online input data, while the data written to the database and the returned response constitute online output data. Therefore, it's necessary to collect data from the database and the response to obtain the online output data. The database table names are used as data source identifiers for the online output data collected from the database.

[0114] Specifically, the sampling frequency is determined based on the amount of online business data and the statistical quantities that can be aggregated during subsequent analysis. For example, if a data source generates one million data entries per day, ten thousand of them will be collected. The collector stores these online output data into the collection storage (i.e., local storage) of the traffic replay test system.

[0115] For example, the online business service outputs MySQL data. The collector collects this data and outputs it to Elasticsearch (es). Each es record has two fields: a data source identifier, which records the source database table name of the MySQL data, and the specific online output data, which records the JSON value of each field in the MySQL data.

[0116] Specifically, the embodiments of this application can periodically analyze the stored online output data. First, based on the data source identifier of each stored online output data, online output data with the same data source is obtained, thereby parsing out the various fields contained therein and their corresponding field values.

[0117] In some implementations, obtaining multiple fields included in online output data from the same data source, and the field value corresponding to each field in each online output data, includes:

[0118] (1) Obtain multiple online output data from the same data source;

[0119] (2) Identify multiple fields with the same field type in multiple online output data;

[0120] (3) Extract the field value corresponding to each field from multiple online output data according to the dimension of each field.

[0121] Specifically, based on the data source identifier of each online output data stored, multiple online output data from the same data source are obtained, and each online output data from the same data source is parsed to determine multiple fields with the same field type.

[0122] Specifically, online output data consists of fields of at least one field type. For example, online output data collected from a certain data source identifier A includes a field of unique identifier (id), a field of name, and a field of time. Similarly, online output data from other data sources with the same identifier A also includes a field of id, a field of name, and a field of time.

[0123] After obtaining multiple fields from multiple online output data sources with the same data source, the field values ​​are extracted from each field in each online output data source, using the field as the dimension, to obtain multiple field values ​​for each field.

[0124] For example, the online output data with data source identifier A includes three data points: online output data 1, online output data 2, and online output data 3. Online output data 1 includes field a with a value of 5 and field b with a value of 4; online output data 2 includes field a with a value of 6 and field b with a value of 3; and online output data 3 includes field a with a value of 5 and field b with a value of 7. Therefore, the field values ​​for field a are 5, 6, and 5, and the field values ​​for field b are 4, 3, and 7.

[0125] Therefore, by filtering online output data based on data source identifiers, it is ensured that the acquired online output data comes from the same source, thus maintaining consistency in business meaning and data structure. Furthermore, by extracting multiple field values ​​corresponding to each field as a dimension, the data can be effectively organized according to fields, facilitating individual statistical analysis of each field.

[0126] In step 202, the field name of each field is input into the trained random number noise prediction model to predict the random number noise prediction result for each field.

[0127] The random number noise prediction model is a trained machine learning model used to predict whether a field is a random number noise field based on its name. During training, the model uses a large number of field names known to be random number noise fields or non-random number noise fields as a training set to learn the relationship between the characteristics of the field names and whether they are random number noise fields, thus enabling inference about whether a field name is a random number noise field after training. The random number noise prediction result is the prediction conclusion output by the trained random number noise prediction model based on the input field name, typically represented as "random number noise field," "non-random number noise field," or "to be determined."

[0128] During the random number noise prediction process using the trained random number noise prediction model, the following command can be used to trigger the prediction: "Please determine whether the field name 'aaa' is a random number noise field, and output 'Yes' or 'No'."

[0129] Specifically, to enable the model to make more accurate predictions, additional pre-defined knowledge can be provided. This knowledge includes the field definitions of the input field names (such as database table definitions) and business service description information (such as textual descriptions of business service functions, processing flows, inputs and outputs, etc.), which are not limited here.

[0130] Please see Figure 4 , Figure 4 This is a schematic diagram illustrating the random number noise prediction process provided in this application embodiment. Taking the field list of this MySQL database table as an example, for fields of random identifier type, including fields such as auto-incrementing id and order number, common field names include "id", "list_id", and "order_id"; for fields of time type, including fields such as creation time and transaction date, common field names include "create_time" and "order_date". During the random number noise prediction process using the trained random number noise prediction model, prediction can also be triggered by the following prediction instruction carrying a preset indication: "Please determine whether the field name 'ccc' is a random number noise field. Random number noise fields include random identifier type or time type. For example, random identifier type is bbb, and time type is ccc. Please output yes or no." "Yes" means "random number noise field", "no" means "non-random number noise field", and when the model cannot predict the result of a field corresponding to a certain field name, it is "pending".

[0131] The trained random number noise prediction model can simultaneously process online output data from different sources (e.g.) Figure 4The following data were predicted using data from a line output from Source 1 and a line output from Source 2: The predicted random number noise results for the following fields are: "id" ("random number noise field"), "name" ("pending"), "type" ("pending"), "list_id" ("random number noise field"), "time" ("random number noise field"), "date" ("random number noise field"), and "expand" ("pending").

[0132] In step 203, a field value difference quantity analysis is performed on multiple field values ​​corresponding to each field to determine the random number noise analysis result for each field.

[0133] The field value difference quantity analysis (i.e., random number noise analysis) involves statistically analyzing the differences in field values ​​across different online output data for each field. By counting the number of different field values, it determines whether a field is a random number noise field. The random number noise analysis result is a conclusion drawn from the field value difference quantity analysis and is used to determine whether a field is a random number noise field. It has the same result type as the random number noise prediction result, including three types of results: "random number noise field", "non-random number noise field", or "pending".

[0134] In some implementations, the step of performing field value difference quantity analysis on multiple field values ​​corresponding to each field to determine the random number noise analysis result for each field includes:

[0135] (1) Obtain the number of field value differences for each of the fields;

[0136] (2) Determine the ratio of the number of field value differences for each field to the total number of online output data, and obtain the percentage of differences for each field;

[0137] (3) Determine the target difference quantity percentage interval for each field from the multiple difference quantity percentage intervals corresponding to each field;

[0138] (4) The target random number noise analysis results corresponding to the target difference quantity ratio interval of each field are determined as the corresponding random number noise analysis results.

[0139] The number of field value differences refers to the number of different field values ​​for a given field across multiple online output data sets. The total number of online output data sets is the total number of all online output data sets from the same data source. The percentage of differences is the ratio of the number of field value differences to the total number of online output data sets.

[0140] Specifically, each field is iterated through, and the number of differences in field values ​​for each field across all online output data is collected. For each field, its number of differences is divided by the total number of data points to obtain the percentage of differences for that field. Different percentage ranges for different fields are pre-defined, and each percentage range corresponds to a different random number noise analysis result. The percentage range containing the percentage of differences is the target percentage range, and the target random number noise analysis result corresponding to the target percentage range is determined as the random number noise analysis result for the corresponding field.

[0141] For each field, the number of distinct values ​​is counted. For example, in the "order number" field, there might be 70,000 different order numbers out of 70,000 data entries; in the "status" field, due to the limited number of states, there are only a few distinct values. By setting different thresholds for different fields, the proportion of distinct values ​​to the total number of values ​​is used to determine whether a field is noise. A threshold of 0.03% (more than 21 distinct values) is considered a noise field; a threshold of 0.003% (less than 3 distinct values) is considered a non-random noise field; and values ​​between 3 and 21 are not considered. Therefore, for the "status" field, the corresponding intervals for the proportion of differences are (0,3], (3,21), and [21,∞). Among them, (0,3] corresponds to a non-random noise field, (3,21) corresponds to an undetermined random noise field, and [21,∞) corresponds to a random noise field.

[0142] Please see Figure 5 , Figure 5This is a schematic diagram illustrating the field value difference quantity analysis process provided in this application embodiment. By analyzing the number of field value differences for each field in online output data from the same data source, the following results are obtained: the field name "id" has a field value difference quantity of "70000", the field name "name" has a field value difference quantity of "7000", the field name "type" has a field value difference quantity of "3", the field name "list_id" has a field value difference quantity of "65000", the field name "time" has a field value difference quantity of "70000", the field name "date" has a field value difference quantity of "7", and the field name "expand" has a field value difference quantity of "1". Furthermore, by combining the random number noise analysis results of each field with the corresponding interval of the proportion of differences, it can be determined that the random number noise analysis results for the field named "id" are "random number noise field", the random number noise analysis results for the field named "name" are "random number noise field", the random number noise analysis results for the field named "type" are "non-random number noise field", the random number noise analysis results for the field named "list_id" are "random number noise field", the random number noise analysis results for the field named "time" are "random number noise field", the random number noise analysis results for the field named "date" are "pending", and the random number noise analysis results for the field named "expand" are "non-random number noise field".

[0143] Field value difference quantity analysis is a supplementary field analysis to the random number noise prediction process. It can help identify fields named "extended fields" that cannot be analyzed by field name and can also prevent some fields with fixed IDs or fixed times from being misjudged as noise fields. The thresholds for different fields can be adjusted according to the actual business situation, that is, the difference quantity ratio range of different fields can be adjusted according to the situation.

[0144] Therefore, by calculating the percentage of differences in field values ​​and comparing it with a pre-defined range of percentage differences, it is possible to more objectively and accurately determine whether a field is a random number noise field. Different fields have different characteristics, and judging directly based on field names may lead to misjudgment. This data statistics-based analysis method considers the actual distribution of field values, reducing the possibility of misjudgment. Fields with non-standard naming but little actual value variation will not be misjudged as random number noise fields due to their names; fields with large value differences can also be more accurately identified as random number noise fields. By calculating the percentage of differences, the random number noise analysis results of the field are quantified, making the random number noise analysis process more scientific and interpretable. Without relying on subjective judgment, this quantitative analysis method helps to more accurately handle noisy fields in subsequent data processing and analysis, improving the quality and efficiency of data analysis.

[0145] In some implementations, obtaining the number of field value differences for each of the fields includes:

[0146] (1.1) For each of the fields, obtain the values ​​of multiple fields to be analyzed;

[0147] (1.2) Select a current field value from multiple field values ​​in sequence, and compare the current field value with each field value in the historical analysis field value set;

[0148] (1.3) When the current field value is different from the field value in the historical analysis field value set, increment the field value difference count by one;

[0149] (1.4) Assign the current field value to the set of historical analysis field values;

[0150] (1.5) When there are unselected field values, return to the step of selecting a current field value from multiple field values ​​in sequence until there are no unselected field values, and obtain the number of field value differences for each field.

[0151] For each field, multiple field values ​​to be analyzed are retrieved from the online output data, and a current field value is selected sequentially from these values. The historical analysis field value set stores previously analyzed field values; this set continuously adds new field values ​​as the analysis progresses. The current field value is compared with each field value in the historical analysis field value set. If the current field value differs from a value in the historical analysis field value set, the difference count is incremented, and the current field value is then assigned to the historical analysis field value set.

[0152] Specifically, when there are unselected field values ​​among the multiple field values ​​to be analyzed, the process returns to the step of selecting a current field value from the multiple field values ​​in sequence, and this process is repeated until there are no unselected field values, thus obtaining the number of field value differences for each field.

[0153] For example, when analyzing the "Age" field, the set is initially empty. After analyzing the first field value 25, 25 is assigned to the historical analysis field value set, which now contains 25, and the number of field value differences is 0 + 1 = 1. When analyzing the second field value 28, since 28 is different from 25 in the set, the number of field value differences for the "Age" field is incremented by one, becoming 1 + 1 = 2, and 25 is assigned to the historical analysis field value set, which now contains both 25 and 28. The third field value is 25. Since 25 is the same as 25 in the historical analysis field value set, the number of field value differences and the historical analysis field value set remain unchanged; the number of field value differences is still 2, and the historical analysis field value set still includes both 25 and 28.

[0154] Therefore, by sequentially selecting field values ​​and comparing them with the historical analysis set of field values, the number of different values ​​for each field can be accurately counted, i.e., the number of field value differences. This avoids double counting, ensures the accuracy of statistical results, clearly distinguishes between identical and different values, and accurately calculates the number of field value differences.

[0155] In step 204, noise analysis is performed using the online output data and offline testing services to determine the environmental noise analysis results for each field.

[0156] After analyzing whether each field is a random number noise type, it is also necessary to analyze whether it is an environmental noise type. This is to avoid using environmental noise fields that are easily affected by environmental noise during subsequent playback tests.

[0157] Specifically, due to the susceptibility of the environmental noise field to interference from environmental noise, noise analysis can be performed in the offline testing service by combining the actual online output data generated online with the offline testing service data, thereby determining the environmental noise analysis results for each field. The environmental noise analysis results include three types: "Environmental Noise Field", "Non-Environmental Noise Field", and "To be determined".

[0158] In some implementations, the step of using the online output data and offline testing services to perform noise analysis and determine the environmental noise analysis result for each field includes:

[0159] (1) Input the online input data corresponding to each online output data into the offline test service to obtain the offline output data corresponding to each online input data;

[0160] (2) Compare the field values ​​of the online output data and the offline output data corresponding to the same online input data in each field to determine the environmental noise analysis results of each field.

[0161] Online input data refers to external input information received during the runtime of the online application service. Online output data is generated based on this online input data. The service version of the offline testing service is the same as the service version of the online application service corresponding to the online output data. Offline output data refers to the output results generated by the offline testing service after inputting the same online input data as the online business service in the offline testing service with the same service version. Therefore, there is corresponding offline output data for each online output data. Since noise fields of the environmental noise type are affected by the environment, it is necessary to input the online input data corresponding to each online output data into the offline testing service. The offline testing service then outputs the corresponding offline output data of the online input data in the offline testing service to ensure that the input data and the business service environment are consistent, thereby determining whether the field values ​​are affected by the environment.

[0162] Among them, the environmental noise analysis results are obtained by comparing the field values ​​of each field in the online output data and the offline output data to determine which field values ​​are inconsistent due to environmental influences. These fields are the environmental noise fields, and the results obtained from the analysis are the environmental noise analysis results.

[0163] In some implementations, the comparison of the online output data corresponding to the same online input data and the corresponding offline output data for each field to determine the environmental noise analysis result for each field includes:

[0164] (1) Compare the field values ​​of the online output data and the offline output data corresponding to the same online input data in each field;

[0165] (2) When the field value of the first field in the online output data is different from the field value of the corresponding offline output data, the environmental noise analysis result of the first field is determined to be the environmental noise field;

[0166] (3) The environmental noise analysis results of the second field other than the first field in the online output data are determined as non-environmental noise fields.

[0167] Specifically, for each field, the field value in the online output data corresponding to the input data on the same line, and the field value in the corresponding offline output data are obtained. The field values ​​of the same field in the obtained online and offline output data are compared. The first field is the field whose field value in the online output data corresponding to the input data on the same line differs from the field value in the offline output data during the comparison process. The environmental noise analysis result of the first field is determined as the environmental noise field. The second field in the online output data, excluding the first field, is the field whose field value in the online output data corresponding to the input data on the same line does not differ from the field value in the offline output data during the comparison process. The environmental noise analysis result of the second field is determined as the non-environmental noise field.

[0168] For details, please refer to Figure 6 , Figure 6 This is a schematic diagram illustrating the environmental noise analysis process provided in an embodiment of this application. The offline testing service and the online application service have the same business service version, both being online versions. In the offline testing service, online input data is input, and corresponding offline output data is obtained. By comparing the field values ​​of each field with the corresponding online input data and the offline output data, it is determined whether it is an environmental noise field. Figure 6 In the analysis, the field named "expand" has a field value of 3 for the input data on the same line and a field value of 4 for the output data on the corresponding line. Since the field values ​​are different, the environmental noise analysis result for the field named "expand" is "Environmental Noise Field".

[0169] In step 205, based on the random number noise prediction results, random number noise analysis results, and environmental noise analysis results for each field, the noise judgment result for each field is determined so that a playback test can be performed based on the noise judgment result for each field.

[0170] For random number noise, each field has corresponding random number noise prediction and analysis results. However, due to different result determination methods, the random number noise prediction and analysis results for each field may differ. Therefore, by further analyzing the random number noise prediction and analysis results for each field, the accuracy of the analysis results for the random number noise dimension of each field can be improved.

[0171] By combining the environmental noise analysis results for each field, the noise judgment results for each field in both the random number noise dimension and the environmental noise dimension are determined. Based on the comprehensively determined noise judgment results, playback tests are conducted to improve the test effect of playback tests.

[0172] In some implementations, determining the noise determination result for each field based on the random number noise prediction result, random number noise analysis result, and environmental noise analysis result for each field includes:

[0173] (1) Compare the random number noise prediction results and the corresponding random number noise analysis results of each field to determine the random number noise comparison results of each field;

[0174] (2) Based on the random number noise comparison results of each field and the corresponding environmental noise analysis results, determine the noise judgment result of each field.

[0175] Both the random number noise prediction process and the random number noise analysis process predict whether a field is a random number noise field. However, there may be situations where the random number noise prediction result obtained for the same field is inconsistent with the random number noise analysis result obtained in the random number noise analysis process. In this case, the random number noise prediction result and the corresponding random number noise analysis result of each field can be compared first to determine the random number noise comparison result of each field. Then, the random number noise comparison result of each field and the corresponding environmental noise analysis result are combined to determine the final noise judgment result of each field.

[0176] Therefore, by comprehensively considering the results of random number noise prediction, random number noise analysis, and environmental noise analysis, the noise attributes of a field can be evaluated from multiple perspectives, avoiding the one-sidedness that may result from a single method. The random number noise prediction model makes predictions based on information such as field names, while random number noise analysis starts from actual data characteristics such as the number of differences in field values. Combining the two can complement and verify each other. When the results of the two are inconsistent, further analysis through comparison can more accurately determine whether a field is a random number noise field. Combined with the environmental noise analysis results, the final noise determination result is more reliable.

[0177] In some implementations, the step of comparing the random number noise prediction results and the corresponding random number noise analysis results for each field to determine the random number noise comparison result for each field includes:

[0178] (1.1) When there is a corresponding random number noise prediction result that is different from the corresponding random number noise analysis result, and the corresponding random number noise analysis result is not the undetermined third field, the random number noise analysis result is determined as the random number noise comparison result of the third field;

[0179] (1.2) When there is a corresponding random number noise prediction result that is different from the corresponding random number noise analysis result, and the corresponding random number noise analysis result is the fourth field to be determined, the random number noise prediction result is determined as the random number noise comparison result of the fourth field.

[0180] Since the random number noise analysis results are obtained by specifically analyzing multiple field values ​​of each field, rather than by predicting random number noise through a simpler model as in random number noise prediction results, the accuracy of the random number noise analysis process is higher than that of the random number noise prediction process. Therefore, the random number noise analysis results are given priority as the random number noise comparison results.

[0181] Based on this, if the random number noise prediction result of the third field is different from the corresponding random number noise analysis result, and the corresponding random number noise analysis result is not "pending", then the random number noise analysis result will be used as the random number noise comparison result of the third field.

[0182] For example, when the random number noise analysis result of the third field is "random number noise field", regardless of whether the corresponding noise prediction result is "random number noise field", "non-random number noise field" or "pending", "random number noise field" is determined as the random number noise comparison result of the third field; when the random number noise analysis result of the third field is "non-random number noise field", regardless of whether the corresponding noise prediction result is "random number noise field", "non-random number noise field" or "pending", "non-random number noise field" is determined as the random number noise comparison result of the third field.

[0183] Specifically, if the random number noise prediction result of the fourth field is different from the corresponding random number noise analysis result, and the corresponding random number noise analysis result is "pending", since the random number noise analysis result cannot determine whether the fourth field is a random number noise field, the random number noise prediction result will be used as the random number noise comparison result of the fourth field.

[0184] For example, when the random number noise analysis result of the fourth field is "pending", the corresponding noise prediction result is "random number noise field", and the "random number noise field" is used as the random number noise comparison result of the fourth field; when the random number noise analysis result of the fourth field is "pending", the corresponding noise prediction result is "non-random number noise field", and the "non-random number noise field" is used as the random number noise comparison result of the fourth field.

[0185] Furthermore, if the random number noise analysis result of a field is the same as the corresponding random number noise prediction result, then either result will be determined as the random number noise comparison result of that field.

[0186] Therefore, when the random number noise analysis result differs from the random number noise prediction result, and the random number noise analysis result is not "pending," the random number analysis result should be prioritized as the random number noise comparison result. This ensures that the final judgment more closely reflects the true characteristics of the data and reduces misjudgments caused by the limitations of model prediction. When the random number noise analysis result is "pending," the random number noise prediction result should be used as the random number noise comparison result to avoid missing judgments due to unclear analysis results. In this case, the random number noise prediction result can serve as a supplementary reference, ensuring that each field receives a corresponding judgment and guaranteeing the completeness of the judgment.

[0187] In some implementations, determining the noise determination result for each field based on the random number noise comparison result and the corresponding environmental noise analysis result includes:

[0188] (1.1) When there is a corresponding random number noise comparison result that is the fifth field of the random number noise field, the random number noise field is determined as the noise judgment result of the fifth field;

[0189] (1.2) When there is a corresponding random number noise comparison result that is a non-random number noise field, and the corresponding environmental noise analysis result is the sixth field of the environmental noise field, the environmental noise field is determined as the noise judgment result of the sixth field;

[0190] (1.3) When there is a corresponding random number noise comparison result that is a non-random number noise field, and the corresponding environmental noise analysis result is the seventh field of the non-environmental noise field, the non-random number noise field and the non-environmental noise field are determined as the noise judgment result of the seventh field.

[0191] If the random number noise comparison result of the fifth field is a random number noise field, it means that the fifth field will definitely not participate in the subsequent playback test. Therefore, regardless of whether the environmental noise analysis result of the fifth field is an environmental noise field or a non-environmental noise field, the "random number noise field" is directly determined as the noise judgment result of the fifth field.

[0192] For a random number noise comparison result that is a non-random number noise field, and the corresponding environmental noise analysis result that is the sixth field of the environmental noise field, due to the interference of environmental noise, although the sixth field is a non-random number noise field, the environmental noise field still needs to be used as the noise judgment result of the sixth field to avoid the sixth field from participating in subsequent playback tests.

[0193] For the seventh field, where the random number noise comparison result is a non-random number noise field and the corresponding environmental noise analysis result is a non-environmental noise field, it indicates that the seventh field is neither a random number noise field nor a field that is affected by environmental noise. Therefore, the non-random number noise field and the non-environmental noise field are determined as the noise judgment result of the seventh field, meaning that the seventh field needs to participate in the subsequent playback test.

[0194] Therefore, by comprehensively considering the results of random number noise comparison and environmental noise analysis, the fields were reasonably classified. The noise determination results of the fields were divided into three categories: random number noise fields, environmental noise fields, and fields that are neither random number noise fields nor environmental noise fields. This classification method clearly defines the noise attributes of different fields, which helps to determine whether each field should participate in subsequent playback tests and improves processing efficiency.

[0195] For details, please refer to Figure 7 , Figure 7This is a schematic diagram illustrating the determination of noise judgment results provided in this application embodiment. For fields named "id", "list_id", and "time", since their random number noise prediction result is "random number noise field" and their random number noise analysis result is "random number noise field", their random number noise comparison result is "random number noise field". Therefore, there is no need to consider their environmental noise analysis result, and the noise judgment result is directly obtained as "random number noise field". For a field named "name", since its random number noise prediction result is "pending" and its random number noise analysis result is "random number noise field", its random number noise comparison result is "random number noise field". Therefore, there is no need to consider its environmental noise analysis result, and the noise judgment result is directly obtained as "random number noise field". For a field named "type", since its random number noise prediction result is "pending" and its random number noise analysis result is "non-random number noise field", its... The random number noise comparison result is "non-random number noise field", and since its environmental noise analysis result is "non-environmental noise field", its noise determination result is "non-random number noise field and non-environmental noise field". For the field named "date", since its random number noise prediction result is "random number noise field" and its random number noise analysis result is "pending", its random number noise comparison result is "random number noise field". Therefore, there is no need to pay attention to its environmental noise analysis result, and its noise determination result is directly "random number noise field". For the field named "expand", since its random number noise prediction result is "pending" and its random number noise analysis result is "non-random number noise field", its random number noise comparison result is "non-random number noise field", but since its environmental noise analysis result is "environmental noise field", its noise determination result is "environmental noise field".

[0196] In some implementations, prior to performing the playback test based on the noise determination result for each of the fields, the method further includes:

[0197] (1) Match the field values ​​of multiple fields in the output data on the same line;

[0198] (2) When at least one online output data contains a field value of the eighth field that matches the field value of the ninth field, generate a matching condition;

[0199] (3) Obtain the number of online output data that meet the matching conditions;

[0200] (4) Determine the ratio of the number of conforming data to the total number of online output data to obtain the proportion of conforming data;

[0201] (5) When the proportion of the number of matching items reaches a preset proportion threshold, matching relationship information is generated. The matching relationship information is used to indicate that the field value of the eighth field matches the field value of the ninth field.

[0202] The playback test based on the noise determination result of each field includes:

[0203] (6) Perform a playback test based on the noise determination result of each field and the matching relationship information.

[0204] In this process, the values ​​of multiple fields in each online output data are matched to determine whether there is a correlation between the values ​​of different fields. This correlation can also be used to determine whether there are any anomalies in the business service during subsequent replay testing.

[0205] For example, this business service is configured to record the value of the field named "name" along with the value of the field named "list_id" when generating the field value, thus ensuring that the field value of "list_id" matches the field value of "name". For instance, in online output data, the field value "abc123456" of "list_id" shares a partial value "abc" with the field value "abc" of "name". Therefore, during subsequent replay testing, the matching of the field value of "name" with the field value of "list_id" in the test output data can be used to determine if there are any anomalies in the business service version used for the replay test.

[0206] Based on this, if at least one online output data contains a field value in the eighth field that matches the field value in the ninth field, a matching condition is generated. The matching condition is used to describe the matching relationship between the field values ​​in the eighth field and the ninth field. The number of matching conditions in each online output data is traversed. The ratio of the number of matching conditions to the total number of online output data is calculated to obtain the proportion of matching conditions. If the proportion of matching conditions reaches a preset threshold, it indicates that the field values ​​in the eighth field and the ninth field meet the matching condition. Therefore, matching relationship information is generated to indicate that the field values ​​in the eighth field and the ninth field match.

[0207] The preset percentage threshold is a pre-set percentage threshold, such as 99%. When the percentage of the number of online output data that meet the matching conditions reaches the preset percentage threshold, it means that the field value of the eighth field in most of the online output data matches the field value of the ninth field. Therefore, in the subsequent playback test, the noise judgment result and matching relationship information of each field can be referred to for playback test.

[0208] Therefore, by matching the values ​​of multiple fields in the same online output data, the relationships between fields can be discovered. During playback testing, combining these relationships with the noise assessment results of the fields allows for a more accurate determination of whether business services are experiencing anomalies. This prompts in-depth analysis of the values ​​of different fields in the online output data to uncover potential relationships. These relationships may be implicit in the business logic, but they are crucial for understanding business processes and data flow. Discovering these relationships helps to gain a more comprehensive understanding of the operational mechanisms of business services, providing strong support for business optimization and troubleshooting.

[0209] In some implementations, generating a matching condition when at least one online output data field has a value that matches the value of the eighth field includes:

[0210] (1.1) When at least one online output data contains a field value of the eighth field that is the same as the field value of the ninth field, generate a matching condition that the field value of the eighth field is the same as the field value of the ninth field;

[0211] (1.2) When at least one online output data contains a field value of the eighth field that is the same as a part of the field value of the ninth field, obtain the part of the same position of the field value of the ninth field that is the same as the field value of the eighth field;

[0212] (1.3) Generate a matching condition where the field value of the eighth field is the same as the field value of the ninth field in the same position of the part.

[0213] The matching of field values ​​includes two cases: all identical or partially identical. If at least one online output data contains a field value of the eighth field that is identical to a field value of the ninth field, then all identical, and the generated matching condition is that the field value of the eighth field is identical to the field value of the ninth field. If at least one online output data contains a field value of the eighth field that is identical to a partial field value of the ninth field, then the partial identical position of the part of the field value of the ninth field that is identical to the field value of the eighth field is obtained. This partial identical position is used to indicate which part of the field value of the ninth field is identical to the field value of the eighth field, and the generated matching condition is that the field value of the eighth field is identical to the field value of the ninth field that is at the partial identical position.

[0214] For example, if the field value of the field named "time" is the same as the field value of the field named "date", then the generated matching condition is "the field value of the field named "time" is the same as the field value of the field named "date""; if the field value "abc" of the field named "name" has a part of the field value "abc" that is the same as the first three digits of the field value "abc123456" of the field named "list_id", then the part of the same position is the first three digits, and the generated matching condition is "the field value of the field named "name" is the same as the first three digits of the field value of the field named "list_id"" or simply "field name = field list_id[0:3]".

[0215] Please see Figure 8 and Figure 9 , Figure 8 This is a schematic diagram of the matching relationship analysis provided in an embodiment of this application. Figure 9 This is a schematic diagram illustrating the noise determination results and matching relationship information for each field provided in the embodiments of this application. For online output data from the same data source, the field values ​​of fields within the same online output data are matched. Taking online output data from source 1 as an example, it is determined that for the field named "name", the matching relationship information "name = list_id[0:5]" indicates that the first five digits of the field value of the field named "name" are the same as those of the field value of the field named "list_id". For the field named "date", the matching relationship information "date = time[0:8]" indicates that the first eight digits of the field value of the field named "date" are the same as those of the field value of the field name of the field named "time".

[0216] Therefore, by distinguishing between cases where all field values ​​are identical and cases where some values ​​are identical, and generating corresponding matching conditions for each case, the relationships between different fields can be described more accurately. When all field values ​​are identical, a simple and clear matching condition is generated directly; when some values ​​are identical, the location information of the identical parts is obtained and matching conditions are generated accordingly, ensuring that the matching conditions accurately reflect the partial matching relationships between field values. In replay testing, accurate matching conditions are crucial for determining whether business services have encountered anomalies. The matching conditions generated by this method can capture the matching situation between field values ​​more meticulously, enabling more rigorous verification of the correctness of business logic during testing. If, during replay testing, field values ​​that should have met the matching conditions no longer match, it can be more accurately determined that the business service version used in the replay test may have encountered anomalies, thereby improving the effectiveness and reliability of the test.

[0217] In some implementations, when replay testing is used to compare fields of test output data of the under-development business service version with fields of online output data of the online business service version in the online environment, fields with noise determination results of "random number noise field" or "environmental noise field" are not included in the comparison, while fields with noise determination results of "non-random number noise field and non-environmental noise field" are included in the comparison. When replay testing is used to compare fields of test output data of the under-development business service version with fields of test output data of the online business service version in the test environment, fields with noise determination results of "random number noise field" are not included in the comparison, while fields with noise determination results of "environmental noise field" and "non-random number noise field and non-environmental noise field" can be included in the comparison.

[0218] For details, please refer to Figure 10 , Figure 10 This is a schematic diagram illustrating a replay test of fields comparing the under-development business service version with the online business service version in the online environment, provided as an embodiment of this application. Figure 8 The noise determination result of the field named "type" was determined to be "non-random number noise field and non-environmental noise field". Therefore, during the comparison process, the field value of the field named "type" in the online output data and the offline output data are compared to see if they are the same. If the field value is the same, the test passes; if the field value is different, the test fails.

[0219] In addition, for fields with matching conditions, the number of test outputs that meet the matching conditions and the total number of outputs are obtained during the replay test; the ratio of the number of outputs to the total number of outputs is calculated to obtain the test output percentage; when the test output percentage reaches the set output percentage threshold, it means that the test has passed; when the test output percentage does not reach the set output percentage threshold, it means that the test has failed and there is an anomaly in the business service of the business service version under development.

[0220] Please see Figure 11 , Figure 11 This is a schematic diagram of the system framework of the data analysis system provided in this application embodiment. It includes a collector, an analysis subsystem, storage, a model, a playback test subsystem, and a network (web). Online output data is first collected by the collector, which is the starting point of data processing and is responsible for collecting online output data generated online to prepare for subsequent analysis. The analysis subsystem includes three functions: random number noise prediction, random number noise analysis, and environmental noise analysis. All three functions require retrieving the online output data collected by the collector from storage during their implementation. The model is the trained random number noise prediction model used to implement the random number noise prediction function. Random number noise analysis determines the final result by determining the proportion of differences in the values ​​of each field. Environmental noise analysis simulates a business server version identical to the online business service version through the playback test subsystem, thereby determining the environmental noise analysis results of fields by comparing online and offline output data; and determining the matching relationships between fields by analyzing the matching relationships between field values. Finally, based on a comprehensive analysis of the random number noise prediction results, random number noise analysis results, and environmental noise analysis results for each field, the noise judgment result for each field is determined, and the matching relationship information obtained from the matching relationship analysis is combined. The analysis results are then displayed online for manual confirmation. After confirmation, the comparison subsystem is configured as a business service version under development. The field values ​​in the online output data and the test output data are compared to each other. If they differ, the test fails; if they match, the test passes, thus obtaining the final comparison result.

[0221] As described above, this application embodiment obtains multiple fields from online output data of the same data source, and the field value corresponding to each field in each online output data; inputs the field name of each field into a trained random number noise prediction model to predict the random number noise prediction result of each field; performs field value difference quantity analysis on multiple field values ​​corresponding to each field to determine the random number noise analysis result of each field; performs noise analysis using the online output data and offline testing services to determine the environmental noise analysis result of each field; and determines the noise judgment result of each field based on the random number noise prediction result, random number noise analysis result, and environmental noise analysis result of each field, so as to perform playback testing based on the noise judgment result of each field.

[0222] Therefore, by inputting the field name into the random number noise prediction model to obtain the random number noise prediction result of the field, performing difference quantity analysis on the field value to obtain the random number noise analysis result of the field, and using the online output data and offline testing services to perform noise analysis, the environmental noise analysis result of each field is determined. In this way, a comprehensive evaluation is performed from multiple dimensions to determine whether a field is a noise field and the specific noise type of the noise field, making the determination of noise fields more accurate. Compared with the related technology, which relies on manual annotation to directly label inconsistent fields as noise fields, the embodiments of this application can perform comprehensive analysis on fields from multiple dimensions to accurately determine whether each field is a random number noise field and whether it is an environmental noise field. This avoids labeling non-random number noise fields that are easily affected by environmental noise as non-noise fields, improves the accuracy of noise field determination, and thus improves the precision of data analysis, avoids manual annotation, and saves labor costs.

[0223] To facilitate better implementation of the data analysis method provided in the embodiments of this application, the embodiments of this application also provide an apparatus based on the above-described data analysis method. The meanings of the terms used are the same as in the data analysis method described above, and specific implementation details can be found in the descriptions within the method embodiments.

[0224] Please see Figure 12 , Figure 12 This is a schematic diagram of the structure of a data analysis device provided in an embodiment of this application. The data analysis device is applied to a computer device and may include a first acquisition unit 301, a prediction unit 302, a first determination unit 303, a second determination unit 304, and a third determination unit 305.

[0225] The first acquisition unit 301 is used to acquire multiple fields included in online output data from the same data source, and the field value corresponding to each field in each online output data.

[0226] The prediction unit 302 is used to input the field name of each field into the trained random number noise prediction model and predict the random number noise prediction result of each field.

[0227] The first determining unit 303 is used to perform field value difference quantity analysis on multiple field values ​​corresponding to each field, and determine the random number noise analysis result of each field.

[0228] The second determining unit 304 is used to perform noise analysis using the online output data and offline testing services to determine the environmental noise analysis result for each field.

[0229] The third determining unit 305 is used to determine the noise judgment result of each field based on the random number noise prediction result, random number noise analysis result, and environmental noise analysis result of each field, so as to perform playback test according to the noise judgment result of each field.

[0230] In some embodiments, the first determining unit 303 includes:

[0231] The first acquisition subunit is used to acquire the number of field value differences for each of the fields;

[0232] The first determining subunit is used to determine the ratio of the number of field value differences for each field to the total number of online output data, thereby obtaining the percentage of differences for each field.

[0233] The second determining subunit is used to determine the target difference quantity percentage interval of each field from multiple difference quantity percentage intervals corresponding to each field.

[0234] The third determining subunit is used to determine the target random number noise analysis result corresponding to the target difference quantity ratio interval of each of the fields as the corresponding random number noise analysis result.

[0235] In some implementations, the first acquisition subunit is configured to:

[0236] For each of the aforementioned fields, obtain the values ​​of multiple fields to be analyzed;

[0237] Select a current field value from multiple field values ​​in sequence, and compare the current field value with each field value in the historical analysis field value set;

[0238] When the current field value differs from the field value in the historical analysis field value set, increment the field value difference count by one;

[0239] Assign the current field value to the set of historical analysis field values;

[0240] When there are unselected field values, return to the step of selecting a current field value from multiple field values ​​in sequence until there are no unselected field values, and obtain the number of field value differences for each field.

[0241] In some embodiments, the second determining unit 304 includes:

[0242] The input subunit is used to input the online input data corresponding to each of the online output data to the offline testing service to obtain the offline output data corresponding to each of the online input data.

[0243] The comparison sub-unit is used to compare the field values ​​of the online output data corresponding to the input data on the same line and the corresponding offline output data in each field to determine the environmental noise analysis result of each field.

[0244] In some embodiments, the comparison subunit includes:

[0245] The first comparison subunit is used to compare the field values ​​of the online output data corresponding to the online input data and the corresponding offline output data in each field.

[0246] The fourth determining subunit is used to determine the environmental noise analysis result of the first field as the environmental noise field when the field value of the online output data of the first field is different from the field value of the corresponding offline output data.

[0247] The fifth determining subunit is used to determine the environmental noise analysis result of the second field in the online output data other than the first field as a non-environmental noise field.

[0248] In some embodiments, the third determining unit 305 includes:

[0249] The second comparison subunit is used to compare the random number noise prediction results and the corresponding random number noise analysis results of each field to determine the random number noise comparison result of each field.

[0250] The sixth determining subunit is used to determine the noise judgment result of each field based on the random number noise comparison result of each field and the corresponding environmental noise analysis result.

[0251] In some implementations, the second comparison subunit is used for:

[0252] When there is a difference between the corresponding random number noise prediction result and the corresponding random number noise analysis result, and the corresponding random number noise analysis result is not the undetermined third field, the random number noise analysis result is determined as the random number noise comparison result of the third field.

[0253] When there is a difference between the corresponding random number noise prediction result and the corresponding random number noise analysis result, and the corresponding random number noise analysis result is a fourth field to be determined, the random number noise prediction result is determined as the random number noise comparison result of the fourth field.

[0254] In some implementations, the sixth determining subunit is used for:

[0255] When there is a corresponding random number noise comparison result that is the fifth field of the random number noise field, the random number noise field is determined as the noise determination result of the fifth field;

[0256] When there is a corresponding random number noise comparison result that is a non-random number noise field, and the corresponding environmental noise analysis result is the sixth field of the environmental noise field, the environmental noise field is determined as the noise judgment result of the sixth field.

[0257] When there is a corresponding random number noise comparison result that is a non-random number noise field, and the corresponding environmental noise analysis result is the seventh field of the non-environmental noise field, the non-random number noise field and the non-environmental noise field are determined as the noise judgment result of the seventh field.

[0258] In some embodiments, the apparatus further includes:

[0259] The matching unit is used to match the field values ​​of multiple fields in the output data on the same line;

[0260] The first generation unit is used to generate matching conditions when at least one field value in the online output data matches the field value in the ninth field.

[0261] The second acquisition unit is used to acquire the number of online output data that meet the matching conditions;

[0262] The calculation unit is used to determine the ratio of the number of conforming data to the total number of online output data, and to obtain the proportion of conforming data.

[0263] The second generation unit is used to generate matching relationship information when the proportion of the number of matching items reaches a preset proportion threshold. The matching relationship information is used to indicate that the field value of the eighth field matches the field value of the ninth field.

[0264] The third determining unit 305 includes:

[0265] The playback test subunit is used to perform playback tests based on the noise determination results of each field and the matching relationship information.

[0266] In some implementations, the first generating unit includes:

[0267] The first generation subunit is used to generate a matching condition that the field value of the eighth field is the same as the field value of the ninth field when at least one online output data contains a field value of the eighth field that is the same as the field value of the ninth field.

[0268] The first acquisition subunit is used to acquire the position of the part of the field value of the ninth field that is the same as the field value of the eighth field when at least one field value in the online output data is the same as part of the field value of the ninth field.

[0269] The second generation subunit is used to generate matching conditions where the field value of the eighth field is the same as the field value of the ninth field in the same position of the part.

[0270] In some embodiments, the first acquisition unit 301 includes:

[0271] The second acquisition subunit is used to acquire multiple online output data from the same data source;

[0272] The seventh determination sub-unit is used to determine multiple fields with the same field type in multiple online output data;

[0273] Extract sub-units, used to extract the field values ​​corresponding to each field from multiple online output data according to the dimensions of each field.

[0274] The specific implementation of each of the above units can be found in the previous embodiments, and will not be repeated here.

[0275] As described above, in this embodiment, the first acquisition unit 301 acquires multiple fields included in online output data from the same data source, and the field value corresponding to each field in each online output data; the prediction unit 302 inputs the field name of each field into the trained random number noise prediction model to predict the random number noise prediction result of each field; the first determination unit 303 performs field value difference quantity analysis on the multiple field values ​​corresponding to each field to determine the random number noise analysis result of each field; the second determination unit 304 performs noise analysis using the online output data and offline testing services to determine the environmental noise analysis result of each field; and the third determination unit 305 determines the noise judgment result of each field based on the random number noise prediction result, random number noise analysis result, and environmental noise analysis result of each field, so as to perform playback testing based on the noise judgment result of each field.

[0276] Therefore, by inputting field names into a random number noise prediction model to obtain the random number noise prediction results for a field, performing a difference quantity analysis on field values ​​to obtain the random number noise analysis results for the field, and using online output data and offline testing services to perform noise analysis, the environmental noise analysis results for each field are determined. In this way, a comprehensive evaluation is conducted from multiple dimensions to determine whether a field is a noise field and the specific noise type of the noise field, making the determination of noise fields more accurate. Compared with related technologies, which rely on manual annotation to directly label inconsistent fields as noise fields, this application embodiment can comprehensively analyze fields from multiple dimensions to accurately determine whether each field is a random number noise field and whether it is an environmental noise field. This avoids labeling non-random number noise fields that are easily affected by environmental noise as non-noise fields, improves the accuracy of noise field determination, and thus improves the precision of data analysis, avoids manual annotation, and saves labor costs.

[0277] The specific implementation of each of the above units can be found in the previous embodiments, and will not be repeated here.

[0278] Figure 13 This is a partial structural block diagram of a computer device 1000 implementing an embodiment of the present disclosure. The computer device 1000 can vary significantly due to different configurations or performance characteristics, and may include one or more central processing units (CPUs) 422 (e.g., one or more processors) and a memory 432, and one or more storage media 430 (e.g., one or more mass storage devices) storing application programs 442 or data 444. The memory 432 and storage media 430 may be temporary or persistent storage. The program stored in the storage media 430 may include one or more modules (not shown in the diagram), each module including a series of instruction operations on the server 500. Furthermore, the CPU 422 may be configured to communicate with the storage media 430 and execute the series of instruction operations in the storage media 430 on the server 500.

[0279] Computer device 1000 may also include one or more power supplies 425, one or more wired or wireless network interfaces 450, one or more input / output interfaces 458, and / or one or more operating systems 441, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™, etc.

[0280] The central processing unit 422 in the computer device 1000 can be used to execute the data analysis method of the present disclosure embodiment, namely: acquiring multiple fields included in online output data from the same data source, and the field value corresponding to each field in each online output data; inputting the field name of each field into a trained random number noise prediction model to predict the random number noise prediction result of each field; performing field value difference quantity analysis on the multiple field values ​​corresponding to each field to determine the random number noise analysis result of each field; performing noise analysis using the online output data and offline testing services to determine the environmental noise analysis result of each field; and determining the noise judgment result of each field based on the random number noise prediction result, the random number noise analysis result, and the environmental noise analysis result of each field, so as to perform playback testing according to the noise judgment result of each field.

[0281] This disclosure also provides a computer-readable storage medium for storing program code that executes the data analysis methods of the foregoing embodiments, namely: acquiring multiple fields included in online output data from the same data source, and the field value corresponding to each field in each online output data; inputting the field name of each field into a trained random number noise prediction model to predict the random number noise prediction result for each field; performing a field value difference quantity analysis on the multiple field values ​​corresponding to each field to determine the random number noise analysis result for each field; performing noise analysis using the online output data and offline testing services to determine the environmental noise analysis result for each field; and determining the noise judgment result for each field based on the random number noise prediction result, the random number noise analysis result, and the environmental noise analysis result for each field, so as to perform playback testing based on the noise judgment result for each field.

[0282] This disclosure also provides a computer program product comprising a computer program. A processor of a computer device reads and executes the computer program, causing the computer device to perform the data analysis method described above.

[0283] Furthermore, the terms “comprising” and “including”, and any variations thereof, are intended to cover non-exclusive inclusion, such that a process, method, system, product, or apparatus that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to such process, method, product, or apparatus.

[0284] It should be understood that in this disclosure, "at least one item" means one or more, and "more than one" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0285] It should be understood that in the description of the embodiments disclosed herein, "multiple" means two or more, "greater than", "less than", "exceeding" etc. are understood to exclude the number itself, and "above", "below", "within" etc. are understood to include the number itself.

[0286] In the several embodiments provided in this disclosure, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, indirect coupling or communication connection between apparatuses or units, and may be electrical, mechanical, or other forms.

[0287] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0288] Furthermore, the functional units in the various embodiments of this disclosure can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0289] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this disclosure, 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 storage medium 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 disclosure. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0290] It should also be understood that the various implementation methods provided in this disclosure can be combined arbitrarily to achieve different technical effects.

[0291] In this application embodiment, the terms "module" or "unit" refer to a computer program or part of a computer program that has a predetermined function and works with other related parts to achieve a predetermined goal, and can be implemented wholly or partially using software, hardware (such as processing circuitry or memory), or a combination thereof. Similarly, a processor (or multiple processors or memory) can be used to implement one or more modules or units. Furthermore, each module or unit can be part of an overall module or unit that includes the functionality of that module or unit.

[0292] The above is a detailed description of the embodiments of this disclosure. However, this disclosure is not limited to the above embodiments. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of this disclosure. All such equivalent modifications or substitutions are included within the scope defined by the claims of this disclosure.

Claims

1. A data analysis method, characterized in that, include: Obtain multiple fields included in online output data from the same data source, and the field value corresponding to each field in each online output data; The field name of each field is input into the trained random number noise prediction model to predict the random number noise prediction result for each field. Perform a field value difference quantity analysis on multiple field values ​​corresponding to each field to determine the random number noise analysis result for each field; Noise analysis is performed using the online output data and offline testing services to determine the environmental noise analysis results for each field. Based on the random number noise prediction results, random number noise analysis results, and environmental noise analysis results for each field, a noise determination result for each field is determined, so as to perform playback tests based on the noise determination result for each field.

2. The data analysis method according to claim 1, characterized in that, The step of performing field value difference quantity analysis on multiple field values ​​corresponding to each field to determine the random number noise analysis result for each field includes: Obtain the number of differences in field values ​​for each of the aforementioned fields; Determine the ratio of the number of differences in the field values ​​of each field to the total number of data in the online output data to obtain the proportion of differences in each field; From the multiple difference quantity percentage intervals corresponding to each field, determine the target difference quantity percentage interval in which the difference quantity percentage of each field is located; The target random number noise analysis results corresponding to the target difference quantity ratio interval of each of the fields are determined as the corresponding random number noise analysis results.

3. The data analysis method according to claim 2, characterized in that, The step of obtaining the number of field value differences for each of the fields includes: For each of the aforementioned fields, obtain the values ​​of multiple fields to be analyzed; Select a current field value from multiple field values ​​in sequence, and compare the current field value with each field value in the historical analysis field value set; When the current field value differs from the field value in the historical analysis field value set, increment the field value difference count by one; Assign the current field value to the set of historical analysis field values; When there are unselected field values, return to the step of selecting a current field value from multiple field values ​​in sequence until there are no unselected field values, and obtain the number of field value differences for each field.

4. The data analysis method according to any one of claims 1 to 3, characterized in that, The step of using the online output data and offline testing services to perform noise analysis and determine the environmental noise analysis results for each field includes: Input the online input data corresponding to each of the online output data into the offline testing service to obtain the offline output data corresponding to each of the online input data. By comparing the field values ​​of the online output data corresponding to the same online input data and the corresponding offline output data for each field, the environmental noise analysis results for each field are determined.

5. The data analysis method according to claim 4, characterized in that, The comparison of the online output data and the corresponding offline output data corresponding to the same online input data in each field determines the environmental noise analysis result for each field, including: Compare the field values ​​of each field in the online output data and the corresponding offline output data corresponding to the same online input data; When the field value of the first field in the online output data is different from the corresponding field value in the offline output data, the environmental noise analysis result of the first field is determined to be the environmental noise field. The environmental noise analysis result of the second field in the online output data, excluding the first field, is determined to be a non-environmental noise field.

6. The data analysis method according to claim 1, characterized in that, The determination of the noise assessment result for each field based on the random number noise prediction result, random number noise analysis result, and environmental noise analysis result for each field includes: By comparing the random number noise prediction results and the corresponding random number noise analysis results for each field, the random number noise comparison result for each field is determined. Based on the random number noise comparison results of each field and the corresponding environmental noise analysis results, the noise judgment result of each field is determined.

7. The data analysis method according to claim 6, characterized in that, The random number noise prediction result and the corresponding random number noise analysis result of each field are compared to determine the random number noise comparison result of each field, including: When there is a difference between the corresponding random number noise prediction result and the corresponding random number noise analysis result, and the corresponding random number noise analysis result is not the undetermined third field, the random number noise analysis result is determined as the random number noise comparison result of the third field. When there is a difference between the corresponding random number noise prediction result and the corresponding random number noise analysis result, and the corresponding random number noise analysis result is a fourth field to be determined, the random number noise prediction result is determined as the random number noise comparison result of the fourth field.

8. The data analysis method according to claim 6, characterized in that, The determination of the noise assessment result for each field based on the random number noise comparison result and the corresponding environmental noise analysis result includes: When there is a corresponding random number noise comparison result that is the fifth field of the random number noise field, the random number noise field is determined as the noise determination result of the fifth field; When there is a corresponding random number noise comparison result that is a non-random number noise field, and the corresponding environmental noise analysis result is the sixth field of the environmental noise field, the environmental noise field is determined as the noise judgment result of the sixth field. When there is a corresponding random number noise comparison result that is a non-random number noise field, and the corresponding environmental noise analysis result is the seventh field of the non-environmental noise field, the non-random number noise field and the non-environmental noise field are determined as the noise judgment result of the seventh field.

9. The data analysis method according to claim 1, characterized in that, Before performing the playback test based on the noise determination result of each field, the method further includes: Match the values ​​of multiple fields in the output data on the same line; A matching condition is generated when at least one field value in the online output data matches the field value in the ninth field. Obtain the number of online output data that meet the matching conditions; Determine the ratio of the number of conforming data to the total number of online output data to obtain the proportion of conforming data; When the proportion of the number of matching items reaches a preset proportion threshold, matching relationship information is generated. The matching relationship information is used to indicate that the field value of the eighth field matches the field value of the ninth field. The playback test based on the noise determination result of each field includes: Based on the noise determination results of each field and the matching relationship information, a playback test is performed.

10. The data analysis method according to claim 9, characterized in that, When at least one online output data contains a field value in the eighth field that matches the field value in the ninth field, a matching condition is generated, including: When at least one online output data contains a field value in the eighth field that is the same as the field value in the ninth field, a matching condition is generated in which the field value in the eighth field is the same as the field value in the ninth field. When at least one online output data contains a field value of the eighth field that is the same as a partial field value of the ninth field, obtain the partial same position of the part of the field value of the ninth field that is the same as the field value of the eighth field; The matching condition is generated where the field value of the eighth field is the same as the field value of the ninth field in the same position of the specified portion.

11. The data analysis method according to claim 1, characterized in that, The acquisition of multiple fields included in the online output data from the same data source, and the corresponding field value of each field in each of the online output data, includes: Obtain multiple online output data from the same data source; Identify multiple fields with the same data type in multiple online output data sets; Extract the field value corresponding to each field from multiple online output data according to the dimension of each field.

12. A data analysis device, characterized in that, include: The first acquisition unit is used to acquire multiple fields included in online output data from the same data source, and the field value corresponding to each field in each online output data. The prediction unit is used to input the field name of each field into the trained random number noise prediction model and predict the random number noise prediction result for each field. The first determining unit is used to perform field value difference quantity analysis on multiple field values ​​corresponding to each field, and determine the random number noise analysis result of each field. The second determining unit is used to perform noise analysis using the online output data and offline testing services to determine the environmental noise analysis result for each field. The third determining unit is used to determine the noise judgment result of each field based on the random number noise prediction result, random number noise analysis result, and environmental noise analysis result of each field, so as to perform playback test according to the noise judgment result of each field.

13. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a plurality of instructions adapted for loading by a processor to execute the data analysis method according to any one of claims 1 to 11.

14. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the data analysis method according to any one of claims 1 to 11.

15. A computer program product, comprising a computer program or instructions, characterized in that, When the computer program or instructions are executed by a processor, they implement the data analysis method according to any one of claims 1 to 11.

Images