A test data processing method and apparatus based on semantic analysis

By performing semantic analysis and automated processing on the test scenario description text, the problem of complex and time-consuming test data processing in existing technologies is solved, and fast and accurate test data results are output.

CN116383050BActive Publication Date: 2026-11-10CHINA CITIC BANK CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202310316190.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-03-28
Publication Date
2026-11-10
Estimated Expiration
2043-03-28

AI Technical Summary

Technical Problem

The existing technology for obtaining statistical and analytical results of test data involves complex and time-consuming procedures, and non-professional users are prone to errors in configuration leading to biased results.

Method used

The method uses semantic analysis to segment the user-input test scenario description text into three categories: data body, constraint condition, and data identifier. It automatically filters and processes the basic data of the test scenario, eliminating the need for users to manually configure search parameters.

Benefits of technology

It enables fast and accurate test data processing, avoids result deviations caused by configuration errors by non-professional users, and improves the level of intelligence.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116383050B_ABST
    Figure CN116383050B_ABST
Patent Text Reader

Abstract

This article relates to the field of fintech, and more particularly to a test data processing method and apparatus based on semantic analysis. It includes performing semantic analysis on user-input test scenario descriptions, and automatically processing the basic test scenario data based on the results of the semantic analysis. This quickly generates the processing results for the basic test scenario data, achieving "one-click data output." This eliminates the need for users to analyze the test scenario manually and configure search parameters. It solves the problems of complex and time-consuming steps in obtaining statistical and analytical results of test data in existing technologies, and the frequent errors in configuration leading to biased statistical and analytical results for non-professional users.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This article relates to the field of information technology, and in particular to a test data processing method and apparatus based on semantic analysis. Background Technology

[0002] Test data is an indispensable outcome of the software testing lifecycle. By performing scenario-based statistical analysis of test data, testers can gain a comprehensive understanding of the project's testing progress and process. In practice, obtaining statistical and analytical results for test data in a specific scenario often requires manual analysis by professionals, or offline analysis of the test scenario by users followed by complex system configurations to retrieve the aggregated results. This process is complex and time-consuming, and for non-professional users, configuration errors can easily lead to inaccurate statistical and analytical results.

[0003] There is an urgent need for a test data processing method based on semantic analysis to solve the problems of complex and time-consuming operation steps in obtaining statistical and analytical results of test data in existing technologies. For non-professional users, configuration errors often lead to deviations in the statistical and analytical results of the obtained test data. Summary of the Invention

[0004] To address the problems in existing technologies, this paper provides an automated data acquisition method and apparatus based on semantic analysis of test scenarios. This method performs semantic analysis on user-input test scenario descriptions and automatically processes basic test scenario data based on the results of the semantic analysis, quickly generating processing results for the basic test scenario data with a single click. This eliminates the need for users to analyze the test scenario themselves and manually configure retrieval parameters. It solves the problems of complex and time-consuming steps in obtaining statistical and analytical results of test data in existing technologies, where configuration errors often lead to deviations in the obtained statistical and analytical results for non-professional users.

[0005] To solve the above-mentioned technical problems, the specific technical solution presented in this paper is as follows: On the one hand, the embodiments of this paper provide a test data processing method based on semantic analysis. The method includes performing semantic word segmentation on the test scenario description text input by the user, and classifying the word segmentation results to obtain the category of each word element. The categories include data body category, constraint condition category, and data identifier category. Initial data is obtained by using lexical units belonging to the data body class to perform preliminary screening of the basic data of the test scenario; The constraint range is determined based on the lexical of the constraint condition class, and the constraint condition field is determined from multiple fields of the initial data. The initial data whose values ​​of the constraint condition field conform to the constraint range are then obtained to obtain the target data. The data processing method is determined based on the lexical representation of the data identifier class, and the data processing field is determined from multiple fields of the target data. The value of the data processing field is processed according to the data processing method to obtain the processing result. The processing result is provided to the user.

[0006] Furthermore, semantic segmentation is performed on the user-input test scenario description text, and the segmentation results are classified to obtain the category of each segmented word, which further includes: The test scenario description text is segmented to obtain multiple word units; The word units are vectorized to obtain the word vector for each word segment; The word vectors are then subjected to dimensionality reduction processing; The category of the word element is obtained by calculating the dimensionality-reduced word vectors using a pre-trained KNN model.

[0007] Furthermore, before vectorizing the lexical units, the method further includes: The stop words in the word units are filtered according to a preset stop word list so that the word units after filtering the stop words can be vectorized.

[0008] Furthermore, the initial screening of the basic data of the test scenario using lexical units belonging to the aforementioned data body class further includes: The first score of the initial data is calculated based on the lexical units belonging to the data body class for each of the basic data of the test scenario. The test scenario data whose first score exceeds the first preset threshold is selected as the initial data.

[0009] Furthermore, the formula for calculating the first score of each test scenario's basic data as the initial data based on the data body class is as follows: P mj = similarity(w i M j ) + α j + β j ; Among them, P mj This represents the first score, similarity() represents the similarity measurement algorithm, and w i M represents the i-th word element in the data body class. jα represents the name of the basic data for the j-th test scenario. j This indicates that the system uses the basic data M of the test scenario within the specified time period. j frequency coefficient, β j This indicates that the user has been using the basic data M of the test scenario since its creation. j The frequency coefficient.

[0010] Furthermore, determining the constraint fields from the multiple fields of the initial data further includes: The second score of the initial data field as the constraint field is calculated based on the historical number of times the initial data field has been used as a constraint field. Select the field with the highest second score as the constraint field.

[0011] Furthermore, the formula for calculating the second score of the initial data as a constraint field based on the historical number of times the initial data's fields have been used as constraint fields is as follows: P fi = Num 系统使用 (fi field is used as a constraint field) + Num 该用户使用 (fi field as a constraint field) × γ; Where fi represents the i-th field in the initial data, P fi The field fi represents the second score of the constraint field, Num 系统使用 (The fi field as a constraint field) represents the total number of times the fi field has been used as a constraint field in the system, Num 该用户使用 (fi field as constraint field) indicates the total number of times the user uses field fi as a constraint field, and γ represents the coefficient.

[0012] Furthermore, determining the data processing method based on the lexical units of the data identifier class further includes: Semantic analysis is performed on the lexical units of the data identifier class to determine the data processing method.

[0013] Furthermore, the data processing method includes data statistics; Processing the value of the data processing field according to the aforementioned data processing method further includes: The values ​​of the data processing field are statistically analyzed to obtain statistical results, which are then provided to the user.

[0014] Furthermore, determining the data processing fields from the multiple fields of the target data further includes: Based on the order of the words in the test scenario description text, the word preceding the word of the data identifier class is used as the reference field; The third score of the target data field is calculated based on the historical number of times the target data field has been used as a data processing field under this data processing method and the similarity between the field name of the target data and the name of the reference field. The field with the highest third score is selected as the data processing field.

[0015] Furthermore, the formula for calculating the third score of the target data field as the data processing field, based on the historical number of times the target data field has been used as a data processing field under this data processing method and the similarity between the field name of the target data and the name of the reference field, is as follows: P ei = Num 系统使用 (ei field used as a data processing field) + Num 该用户使用 (The ei field is used as a data processing field) × ζ + similarity(w1, ei) × η; Where ei represents the i-th field of the target data, P ei The field ei represents the third score of the data processing field, Num. 系统使用 (ei field as a data processing field) indicates the total number of times the system uses the ei field as a data processing field under this data processing method. Num 该用户使用 (ei field as data processing field) indicates the total number of times the user uses field ei as the data processing field under this data processing method, w1 represents the reference field, similarity(w1, ei) represents the similarity between the name of the reference field and the name of field ei, and ζ and η represent coefficients.

[0016] Furthermore, the category also includes dimension classes; Before providing the processing result to the user, the method further includes: Semantic analysis is performed on the lexical units of the aforementioned dimension class to determine the basis for dimension determination; Determine the dimension fields from multiple fields of the target data; The processing results are grouped according to the values ​​of the dimension fields so that the grouped results can be provided to the user.

[0017] Furthermore, determining the dimension fields from the multiple fields of the target data further includes: The fourth score of the target data field is calculated based on the historical number of times the target data field has been used as a dimension field and the similarity between the field name of the target data and the name on which the dimension is based. The field with the highest fourth score is selected as the dimension field.

[0018] Furthermore, the formula for calculating the fourth score of a field in the target data as a dimension field, based on the historical frequency of the target data's field being used as a dimension field and the similarity between the field name of the target data and the name on which the dimension is based, is as follows: P gi =Num 系统使用 (gi field as dimension field) + Num 该用户使用 (The gi field is used as a dimension field) ×μ +similarity(w2, gi) ×ν; Where gi represents the i-th field of the target data, P ei This indicates that the field gi is the fourth score of the dimension field, Num 系统使用 (The gi field as a data processing field) indicates the total number of times the system uses the gi field as a dimension field, Num 该用户使用 (gi field as data processing field) indicates the total number of times the user uses field gi as the dimension field, w2 represents the dimension basis, similarity(w2, gi) represents the similarity between the name of the dimension basis and the name of field ei, and μ and ν represent coefficients.

[0019] Furthermore, grouping the processing results based on the values ​​of the dimension fields further includes: The processing results are grouped according to the values ​​of the dimension fields using Elasticsearch query syntax.

[0020] On the other hand, this embodiment also provides a test data processing device based on semantic analysis, including, The test scenario description text processing unit is used to perform semantic word segmentation on the test scenario description text input by the user, and to classify the segmentation results to obtain the category of each word element. The categories include data body category, constraint condition category, and data identifier category. The test scenario basic data screening unit is used to screen the test scenario basic data using lexical units belonging to the data body class to obtain initial data; The constraint processing unit is used to determine the constraint range based on the lexical of the constraint condition class, and to determine the constraint condition field from multiple fields of the initial data, and to determine the initial data whose value of the constraint condition field conforms to the constraint range, thereby obtaining the target data; The data processing unit is used to determine the data processing method based on the lexical units of the data identifier class, determine the data processing field from multiple fields of the target data, calculate the value of the data processing field according to the data processing method, and obtain the processing result. The result providing unit is used to provide the processing result to the user.

[0021] On the other hand, this embodiment also provides a computer device, including 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 method described above.

[0022] On the other hand, embodiments of the present invention also provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0023] Finally, this embodiment of the invention also provides a computer program product, which includes a computer program that, when executed by a processor, implements the above-described method.

[0024] Using the embodiments described in this paper, semantic analysis methods are used to perform semantic word segmentation on the user-input test scenario description text. The segmentation results are divided into three categories: data body, constraint, and data identifier. The data body category represents the type of data subject the user wants to process, the constraint category represents the time range specified by the user, and the data identifier category represents the processing method the user wants to apply to the test data of a specified data body type within the specified time range. This achieves automatic processing of test data based on the test scenario description text, eliminating the need for users to analyze the test scenario manually and to configure search parameters manually. This solves the problem that the existing technology for obtaining statistical and analytical results of test data is complex, time-consuming, and often leads to deviations in the statistical and analytical results of test data due to configuration errors by non-professional users.

[0025] Furthermore, during constraint processing, the system automatically determines constraint fields from multiple fields in the initial data. Based on the values ​​of these constraint fields and the constraint range analyzed from the test scenario description text, it automatically filters out data that meets the constraint range. Additionally, it automatically determines data processing fields from multiple fields in the target data. The system then automatically processes the values ​​of these determined data processing fields according to the data processing method obtained from the test scenario description text analysis. Compared to traditional methods that require staff to configure retrieval parameters, this invention eliminates the need for staff to specify constraint fields and data processing fields. This avoids deviations in the statistical and analytical results of test data obtained due to configuration errors by non-professional users, thus improving the intelligence level of test data processing. Attached Figure Description

[0026] To more clearly illustrate the technical solutions in the embodiments or prior art described herein, the accompanying drawings used in the description of the embodiments or prior art will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this article. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0027] Figure 1 The figure shown is a schematic diagram of an implementation system for a test data processing method based on semantic analysis, according to an embodiment of this paper. Figure 2 The diagram shown is a flowchart of a test data processing method based on semantic analysis, as described in this embodiment. Figure 3 The following steps are shown in the embodiment of this paper: performing semantic word segmentation on the test scenario description text input by the user, and classifying the segmentation results to obtain the category of each segmented word; Figure 4 The figure shows the steps of using lexical units belonging to the data body class to perform initial screening of basic data for the test scenario in an embodiment of this article; Figure 5 The figure shows the steps for determining the constraint fields from multiple fields of the initial data in an embodiment of this paper; Figure 6 The diagram illustrates the steps involved in determining data processing fields from multiple fields of the target data in an embodiment of this paper. Figure 7 The figure shows the steps in this embodiment to group the processing results according to the lexical units of the dimension class; Figure 8 The diagram illustrates the steps for determining dimension fields from multiple fields of the target data in this embodiment. Figure 9 The diagram shown is a schematic representation of a test data processing device based on semantic analysis, as described in this embodiment. Figure 10 The diagram shown is a structural schematic of the computer device in the embodiment of this article.

[0028] [Explanation of Figure Markers]: 101. Terminal; 102. Server; 901. Test Scenario Description Text Processing Unit; 902. Initial screening unit for basic data of test scenarios; 903. Constraint Processing Unit; 904. Data Processing Unit; 905. Result Providing Unit; 1002. Computer equipment; 1004. Processing equipment; 1006. Storage resources; 1008. Drive mechanism; 1010. Input / Output Module; 1012. Input devices; 1014. Output devices; 1016. Presentation device; 1018. Graphical User Interface; 1020. Network interface; 1022. Communication link; 1024. Communication bus. Detailed Implementation

[0029] The technical solutions in the embodiments described below will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments described herein, and not all of the embodiments. Based on the embodiments described herein, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this document.

[0030] It should be noted that the terms "first," "second," etc., used in the specification, claims, and accompanying drawings herein are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, apparatus, product, or device that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0031] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0032] like Figure 1 The diagram shown is a schematic of an implementation system for a test data processing method based on semantic analysis according to an embodiment of this paper. It may include: a terminal 101 and a server 102. The terminal 101 and the server 102 communicate with each other through a network. The network may include a local area network (LAN), a wide area network (WAN), the Internet, or a combination thereof, and is connected to a website, user equipment (e.g., computing devices), and a backend system.

[0033] Test analysts can input test scenario description text (i.e., processing requirement text) on terminal 101. Terminal 101 sends the test scenario description text to server 102. Server 102 can connect to a database storing basic test scenario data (i.e., test result data obtained during or after testing) and retrieve the basic test scenario data from the database. Server 102 can analyze the test scenario description text, search the database based on the index conditions obtained from the analysis, obtain test data that meets the requirements of the test analysts, and then automatically process the obtained test data according to the data processing methods obtained after analyzing the test scenario description text. Finally, the processing results are sent to terminal 101 so that terminal 101 can provide the processing results to the test analysts. Optionally, server 102 can be a node of a cloud computing system (not shown in the figure), or each server can be a separate cloud computing system, including multiple computers interconnected by a network and operating as a distributed processing system.

[0034] To address the problems existing in the prior art, this embodiment provides a test data processing method based on semantic analysis. It realizes semantic analysis of the test scenario description input by the user, and automatically processes the basic data of the test scenario based on the results of semantic analysis, quickly generating the processing results of the basic data of the test scenario, achieving "one-click data output", omitting the process of user analysis of the test scenario, and eliminating the need for users to manually configure search parameters. Figure 2The diagram shows a flowchart of a test data processing method based on semantic analysis, as described in this embodiment. This diagram illustrates the process of automatically processing basic test scenario data based on test scenario description text from user data. However, this process can involve more or fewer steps, based on conventional or non-creative work. The order of steps listed in the embodiment is merely one possible execution order among many and does not represent the only possible order. In actual system or device products, the methods shown in the embodiment or the accompanying drawings can be executed sequentially or in parallel. Specifically, as shown... Figure 2 As shown, the method can be executed by server 102 and may include: Step 201: Perform semantic word segmentation on the test scenario description text input by the user, and classify the segmentation results to obtain the category of each word element. The categories include data body category, constraint condition category, and data identifier category. Step 202: Use the lexical units belonging to the data body class to perform initial screening of the basic data of the test scenario to obtain initial data; Step 203: Determine the constraint range based on the lexical of the constraint condition class, and determine the constraint condition field from multiple fields of the initial data. Determine the initial data whose values ​​of the constraint condition field conform to the constraint range to obtain the target data. Step 204: Determine the data processing method based on the lexical of the data identifier class, and determine the data processing field from multiple fields of the target data. Process the value of the data processing field according to the data processing method to obtain the processing result. Step 205: Provide the processing result to the user.

[0035] The method provided in this embodiment performs semantic segmentation on the user-input test scenario description text using semantic analysis. The segmentation results are categorized into three classes: data body, constraint, and data identifier. The data body class represents the type of data subject the user wants to process; the constraint class represents the time range specified by the user; and the data identifier class represents the processing the user wants to perform on the test data of a specified data body type within the specified time range. This method analyzes the user-input test scenario description text into multiple data types and filters or processes the basic test scenario data according to these data types, ultimately obtaining the processing result. It achieves automatic processing of test data based on the test scenario description text, eliminating the need for user analysis of the test scenario and manual configuration of search parameters. This solves the problem in existing technologies where obtaining statistical and analytical results of test data involves complex and time-consuming steps, and for non-professional users, configuration errors often lead to deviations in the obtained statistical and analytical results.

[0036] Furthermore, during constraint processing, the system automatically determines constraint fields from multiple fields in the initial data. Based on the values ​​of these constraint fields and the constraint range analyzed from the test scenario description text, it automatically filters out data that meets the constraint range. Additionally, it automatically determines data processing fields from multiple fields in the target data. The system then automatically processes the values ​​of these determined data processing fields according to the data processing method obtained from the test scenario description text analysis. Compared to traditional methods that require staff to configure retrieval parameters, this invention eliminates the need for staff to specify constraint fields and data processing fields. This avoids deviations in the statistical and analytical results of test data obtained due to configuration errors by non-professional users, thus improving the intelligence level of test data processing.

[0037] In existing technologies, users need to manually analyze the test scenario description text they provide to obtain the test scenario data they require. To eliminate this manual analysis step, this invention uses natural language processing technology to segment the test scenario description text. It identifies the meaning of each word in the description text and categorizes them accordingly.

[0038] In this embodiment of the invention, two input methods are provided: First, the user writes and inputs a test scenario description text according to the format requirements of the test scenario description text. Second, the user selects test scenario elements provided by the system, such as defect type, case type, etc., and the system automatically assembles and generates a test scenario description text based on the user's selected elements.

[0039] According to one embodiment of the present invention, such as Figure 3 As shown, semantic word segmentation is performed on the user-input test scenario description text, and the segmentation results are classified to obtain the category of each segmented word, which further includes: Step 301: Perform word segmentation on the test scenario description text to obtain multiple word units; Step 302: Vectorize the word units to obtain the word vector for each word segment; Step 303: Perform dimensionality reduction processing on the word vectors; Step 304: Calculate the category of the word element by using the pre-trained KNN model to reduce the dimensionality of the word vector.

[0040] In this embodiment of the invention, the system captures the test scenario description text input by the user and treats it as text. The analyzer selects ik_smart for word segmentation to obtain multiple tokens. At the same time, a custom dictionary is configured on Elasticsearch. Based on the key test data indicator names defined in the test domain and combined with the system's own usage, the number of test cases, number of signatures, admission criteria, defect rate, etc., are selected to establish a professional test dictionary. Nginx is used to remotely extend the dictionary. When the content input by the user is in the custom dictionary, it will be returned as an independent token, thereby making the word segmentation results more accurate and reliable.

[0041] According to an embodiment of the present invention, in order to improve the classification speed and accuracy, before vectorizing the word units, the method further includes: filtering stop words in the word units according to a preset stop word list, so as to facilitate vectorization of the word units after filtering the stop words.

[0042] For example: word segmentation result set P W = {W1, W2, W3 ... W n}; via the system's preset stop word list L s The result set after processing: P w = {w1, w2, w3, ... w m}; For result set P w Each word w i (1<=i<=m) are all vectorized using the word2vec model, with size=30 as the dimension of the feature vector, finally obtaining the word vector V of the segmented word. i = (x i1 ,x i2 , x i3 , ... x i30 ), where m represents the total number of word segments; For each word vector V i Dimensionality reduction is performed by using the PCA algorithm to extract the main feature components of each word vector, resulting in the dimensionality-reduced word vector v. i = (y i1 , y i2 , y i3 , y i4 , y i5 ); Using a trained KNN model, input word vector v i = (y i1 , y i2 , y i3 , yi4 , y i5 This allows us to predict the category K of each word segmentation. i .

[0043] In this embodiment of the invention, the test scenario description text "number of test cases created by the system in the past week" is analyzed to obtain lexical units belonging to the data body class, constraint class, and data identifier class. The data body class indicates the type of data the user needs to process, i.e., the initial data required by the scenario, such as "number of cases." "Number of cases" can be added to a pre-set professional test lexicon, allowing comparison with the professional test lexicon when determining the lexical units for the data body class. The constraint class serves as the conditional constraint on the initial data in the scenario description, such as a time constraint, e.g., "in the past week." The data identifier class represents the processing type of the initial data, i.e., what processing the user needs to perform on the retrieved data, such as detail display, mean calculation, summation, etc. When training the KNN algorithm, training samples of these categories in the training set can be pre-labeled, and training can be performed based on the labeling results.

[0044] In this embodiment of the invention, the basic data for the test scenario can be data generated and stored in Elasticsearch using SQL, Excel, or other methods, or large wide table data queried in real time from DB2, MySQL, or Oracle using SQL statements. Figure 4 As shown, the initial screening of basic test scenario data using lexical units belonging to the data body class further includes: Step 401: Calculate the first score of the basic data of each test scenario based on the lexical units belonging to the data body class; Step 402: Select the test scenario basic data whose first score exceeds the first preset threshold as the initial data.

[0045] Furthermore, the formula for calculating the first score of each test scenario's basic data as the initial data based on the data body class is: P mj = similarity(w i M j ) + α j + β j .

[0046] Among them, P mj This represents the first score, similarity() represents the similarity measurement algorithm, and w i M represents the i-th word element in the data body class. j α represents the name of the basic data for the j-th test scenario. jThis indicates that the system uses the basic data M of the test scenario within the specified time period. j frequency coefficient, β j This indicates that the user has been using the basic data M of the test scenario since its creation. j The frequency coefficient.

[0047] This invention can be understood as calculating the similarity between the names of basic data and the lexical units in the data body class, thereby selecting basic data with high similarity as initial data. In the above formula, the similarity algorithm uses the word2vec model to calculate the similarity of the i-th lexical unit w in the data body class. i With the j-th basic data M j Similarity, α j This can provide basic data (queries, subscriptions, etc.) used in the system over the past month. j frequency coefficient, β j This user has been using the basic data M since its creation. j The frequency coefficients, and satisfy the following example conditions: When the system's basic data for the past month is M j When the number of times 'a' is used, α is 0.05 when it is in the interval (0,5); α is 0.10 when it is in the closed interval [6,10]; α is 0.15 when a>10; and α is 0 in other cases. Since the system created this user's basic data M j When the number of times b is used is in the closed interval (0,5], β is 0.05; when b is in the closed interval [6,10], β is 0.15; when a>10, β is 0.20; otherwise, β is 0.

[0048] Alternatively, the initial data can be selected from the test scenario data whose first score exceeds a first preset threshold. Alternatively, the initial data can be selected from the top-ranked, predetermined number of data points sorted in descending order of the first score. The first preset threshold or predetermined number can be set empirically, and this embodiment does not impose any limitations.

[0049] Then, the initial data is processed to apply constraints. In this embodiment of the invention, the constraint range is determined based on the lexical units of the constraint class, for example, by using a natural language recognition algorithm to analyze the time range represented by the lexical units of the constraint class. Then it is necessary to determine which field in the initial data matches the constraint range (i.e., determine the constraint field). Therefore, according to one embodiment of the invention, as... Figure 5 As shown, determining the constraint fields from multiple fields of the initial data further includes: Step 501: Calculate the second score of the initial data field as the constraint field based on the historical number of times the initial data field has been used as a constraint field; Step 502: Select the field with the highest second score as the constraint field.

[0050] In this embodiment of the invention, the system can record the number of times each field of the basic data is used as a constraint field in the historical processing. For example, the system analyzes and determines that a certain field is a constraint field or the user specifies a certain field as a constraint field. Then, based on the number of times the field in the initial data is used as a constraint field in this automatic processing, the score of that field is calculated. Then, the field with the highest score is selected as the initial field.

[0051] Specifically, the formula for calculating the second score of the initial data as a constraint field based on the historical number of times the initial data's field has been used as a constraint field is: P fi = Num 系统使用 (fi field is used as a constraint field) + Num 该用户使用 (fi field is used as a constraint field) × γ.

[0052] Where fi represents the i-th field in the initial data, P fi The field fi represents the second score of the constraint field, Num 系统使用 (The fi field as a constraint field) represents the total number of times the fi field has been used as a constraint field in the system, Num 该用户使用 (The fi field is used as a constraint field) indicates the total number of times the user uses the fi field as a constraint field, and γ represents a coefficient. γ can be determined experimentally; preferably, in this embodiment of the invention, γ is set to 2.

[0053] In some other embodiments of the present invention, a semantic recognition algorithm can also be used to calculate the matching degree between the name of each field of the initial data and the tokens of the constraint class, thereby selecting the constraint field based on the matching degree.

[0054] In some other embodiments of the present invention, the constraint class can also be in the form of constraints other than time constraints, for example, determined according to the field format of the underlying data. For other forms of constraints, commonly used indicator constraint fields and constraint values ​​defined in the testing domain can be pre-stored in Elasticsearch, and the constraint fields can be obtained by assembling query conditions using Elasticsearch's query syntax.

[0055] According to one embodiment of the invention, determining the data processing method based on the lexical representation of the data identifier class further includes: Semantic analysis is performed on the lexical units of the data identifier class to determine the data processing method.

[0056] In this embodiment of the invention, a semantic analysis model can be pre-trained. Test data belonging to the category of cardinality data in the test scenario is used as the training set, and the training set is labeled. The semantic analysis model is then trained using the labeled training set. Semantic analysis is then performed on the lexical units of the data identifier class to obtain the corresponding data processing method.

[0057] According to one embodiment of the present invention, the data processing method may include data statistics and data detailing. Data statistics involves aggregating the values ​​of the data processing fields of the target data. Data detailing involves displaying the values ​​of the data processing fields of the target data in detail.

[0058] When performing data statistics, the aggregation type is first determined based on the content of the data identifier words. For example, if the word content contains "average" or "mean value", the aggregation type is "average value (avg)". If the word content contains "maximum value", the aggregation type is "maximum value (max)". If the word content contains "sum" or "summing up", the aggregation type is "summing up (sum)". If the word content contains "number" or "quantity", the aggregation type is "count".

[0059] When it is data details, the data processing is determined based on the content of the data identifier class terminology, such as the terminology content containing names, etc.

[0060] In this embodiment of the invention, in addition to determining the data processing type, it is also necessary to determine which field of the target data should be used as the data processing field. For example, if the data processing method is to calculate the total, it is necessary to determine which field of the target data should be used to calculate the total. Therefore, according to one embodiment of the invention, such as Figure 6 As shown, determining the data processing fields from multiple fields of the target data further includes: Step 601: Based on the order of the lexical units in the test scenario description text, take the preceding lexical unit of the data identifier class as the reference field; Step 602: Calculate the third score of the target data field as the data processing field based on the historical number of times the target data field has been used as a data processing field under this data processing method and the similarity between the field name of the target data and the name of the reference field; Step 603: Select the field with the highest third score as the data processing field.

[0061] In this embodiment of the invention, due to the difficulty in selecting the aggregation field, the selection of which field to use as the aggregation field cannot rely solely on the existing data in the system, but also needs to be determined based on the test scenario described by the user. Therefore, in this embodiment of the invention, after semantic analysis of the test scenario description text to obtain lexical units, it is necessary to save the order of the lexical units. After removing stop words, the preceding lexical unit of the data identifier class is found and used as the reference field. For example, after semantic analysis of the test scenario description text "number of test cases created by the system in the past week", the resulting word segments are: "system", "past", "one", "week", "nearly week", "create", "test", "case", "number", "number of cases", "number of test cases". The lexical unit of the data identifier class is "number", and its preceding lexical unit (prefix word) is "case". Therefore, when selecting the aggregation field, "case" is used as the reference field, and the aggregation field "case id" is selected according to "case".

[0062] Then, the data processing fields are determined. The format of the basic data in this embodiment of the invention can be as shown in Table 1: Table 1

[0063] Specifically, the formula for calculating the third score of the target data field as the data processing field, based on the historical number of times the target data field has been used as a data processing field under this data processing method and the similarity between the field name of the target data and the name of the reference field, is as follows: P ei = Num 系统使用 (ei field used as a data processing field) + Num 该用户使用 (The ei field is used as a data processing field) × ζ + similarity(w1, ei) × η Where ei represents the i-th field of the target data, P ei The field ei represents the third score of the data processing field, Num. 系统使用 (ei field as a data processing field) indicates the total number of times the system uses the ei field as a data processing field under this data processing method. Num 该用户使用 (ei field as data processing field) indicates the total number of times the user uses field ei as the data processing field under this data processing method, w1 represents the reference field, similarity(w1, ei) represents the similarity between the name of the reference field and the name of field ei, and ζ and η represent coefficients.

[0064] In this embodiment of the invention, the similarity algorithm is used to calculate the similarity between the reference field w1 and the field name ei of the word segmentation content using the word2vec model. ζ and η can be set according to the experiment. Preferably, ζ is 2 and η is 5.

[0065] Using the formula above, we determine the data processing field as "case id". Then, we use Elasticsearch's query syntax to process the value of this field according to the processing method. For example, if the processing method is "count", then we can count the total number of values ​​for the "case id" field. Finally, we provide the processed data to the user.

[0066] For example, if the test scenario description text is "total number of cases created in December 2022", then the output processing result will be 4.

[0067] In this embodiment of the invention, the method for selecting the preceding term of a data identifier as the reference field is based on user habits regarding test scenario descriptions and extensive experimentation. It should be noted that prefix words do not necessarily have a 100% correlation with the aggregation field. Therefore, when selecting the aggregation field, the selection is not solely based on prefix words, but also on the weighted average of three factors: the number of times the user has used this field as an aggregation field stored in the current system, the total number of times this field has been used as an aggregation field in the system, and so on. Experiments have shown that the accuracy of the aggregation field obtained using this calculation formula is over 90%.

[0068] According to one embodiment of this article, the categories can be further subdivided, specifically, the categories also include dimension categories; like Figure 7 As shown, before providing the processing result to the user, the method further includes: Step 701: Perform semantic analysis on the lexical units of the dimension class to determine the dimension basis; Step 702: Determine the dimension fields from the multiple fields of the target data; Step 703: Group the processing results according to the values ​​of the dimension fields so that the grouping results can be provided to the user.

[0069] In this embodiment of the invention, lexical units of the dimension class are used to group the processing results, that is, to perform grouped queries on the target data. Alternatively, a trained KNN model can be used to determine lexical units belonging to the dimension class in the test scenario description text. For example, in the test scenario description text "number of new test cases in each domain this week", the lexical unit of the dimension class is "domain" after word segmentation. Then, the processing result "total number of new test cases this week" is grouped according to "domain" to obtain the number of new test cases in each "domain" this week.

[0070] Before grouping the processing results, it is necessary to determine which field of the target data will be the dimension field, that is, to determine which field's value will be used for grouping. Therefore, as follows: Figure 8 As shown, determining the dimension fields from multiple fields of the target data further includes: Step 801: Calculate the fourth score of the target data field as the dimension field based on the historical number of times the target data field has been used as a dimension field and the similarity between the field name of the target data and the name on which the dimension is based; Step 802: Select the field with the highest fourth score as the dimension field.

[0071] Furthermore, the formula for calculating the fourth score of a field in the target data as a dimension field, based on the historical frequency of the target data's field being used as a dimension field and the similarity between the field name of the target data and the name on which the dimension is based, is: P gi =Num 系统使用 (gi field as dimension field) + Num 该用户使用 (gi field is used as a dimension field) ×μ + similarity(w2, gi) ×ν.

[0072] Where gi represents the i-th field of the target data, P ei This indicates that the field gi is the fourth score of the dimension field, Num 系统使用 (The gi field as a data processing field) indicates the total number of times the system uses the gi field as a dimension field, Num 该用户使用 (The gi field is used as a data processing field) indicates the total number of times the user uses the gi field as the dimension field, w2 represents the dimension basis, similarity(w2, gi) represents the similarity between the name of the dimension basis and the name of the ei field, and μ and ν represent coefficients. μ and ν can be set according to experiments. Preferably, μ is 2 and ν is 10, which can be adjusted according to the actual situation.

[0073] After determining the dimension fields, the processing results are grouped according to the values ​​of the dimension fields using Elasticsearch query syntax to obtain grouped results, which are then provided to the user. In this embodiment of the invention, various data display formats are supported, including tables, pie charts, line charts, bar charts, scatter plots, and dashboards. Display elements include: time period, dimension name, display field name, and data result content. In addition to data display, the system also provides multiple data services such as export, WeChat subscription, email subscription, customized data dashboards, and data monitoring and alerts.

[0074] Based on the same inventive concept, such as Figure 9 As shown in the embodiments herein, a test data processing device based on semantic analysis is also provided, including: The test scenario description text processing unit 901 is used to perform semantic word segmentation on the test scenario description text input by the user, and to classify the segmentation results to obtain the category of each word element. The categories include data body category, constraint condition category, and data identifier category. The test scenario basic data screening unit 902 is used to screen the test scenario basic data using lexical units belonging to the data body class to obtain initial data; The constraint processing unit 903 is used to determine the constraint range based on the lexical of the constraint condition class, and to determine the constraint condition field from multiple fields of the initial data, and to determine the initial data whose value of the constraint condition field conforms to the constraint range, thereby obtaining the target data; The data processing unit 904 is used to determine the data processing method based on the lexical of the data identifier class, determine the data processing field from multiple fields of the target data, calculate the value of the data processing field according to the data processing method, and obtain the processing result; Result providing unit 905 is used to provide the processing result to the user.

[0075] The beneficial effects obtained by the above-described device are the same as those obtained by the above-described method, and will not be described in detail in the embodiments of this specification.

[0076] like Figure 10The diagram illustrates the structure of a computer device according to an embodiment of this document. The apparatus described herein can be the computer device in this embodiment, executing the methods described above. The computer device 1002 may include one or more processing devices 1004, such as one or more central processing units (CPUs), each of which can implement one or more hardware threads. The computer device 1002 may also include any storage resource 1006 for storing information of any kind, such as code, settings, data, etc. Without limitation, for example, the storage resource 1006 may include any one or more combinations of the following: any type of RAM, any type of ROM, flash memory, hard disk, optical disk, etc. More generally, any storage resource can use any technology to store information. Furthermore, any storage resource can provide volatile or non-volatile retention of information. Further, any storage resource may represent a fixed or removable component of the computer device 1002. In one case, when the processing device 1004 executes associated instructions stored in any storage resource or combination of storage resources, the computer device 1002 can perform any operation of the associated instructions. The computer device 1002 also includes one or more drive mechanisms 1008 for interacting with any storage resource, such as hard disk drive mechanism, optical disk drive mechanism, etc.

[0077] Computer device 1002 may further include an input / output module 1010 (I / O) for receiving various inputs (via input device 1012) and providing various outputs (via output device 1014). A specific output mechanism may include a presentation device 1016 and an associated graphical user interface (GUI) 1018. In other embodiments, the input / output module 1010 (I / O), input device 1012, and output device 1014 may be omitted, and the device may function solely as a computer device within a network. Computer device 1002 may also include one or more network interfaces 1020 for exchanging data with other devices via one or more communication links 1022. One or more communication buses 1024 couple the components described above together.

[0078] The communication link 1022 can be implemented in any way, such as via a local area network, a wide area network (e.g., the Internet), a point-to-point connection, or any combination thereof. The communication link 1022 may include any combination of hardwired links, wireless links, routers, gateway functions, name servers, etc., governed by any protocol or combination of protocols.

[0079] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described data classification and retrieval method.

[0080] This invention also provides a computer program product, which includes a computer program that, when executed by a processor, implements the above-described data classification and retrieval method.

[0081] It should be understood that in the various embodiments of this document, the sequence number of each process does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this document.

[0082] It should also be understood that, in the embodiments herein, the term "and / or" is merely a description of the relationship between associated objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this document generally indicates that the preceding and following associated objects have an "or" relationship.

[0083] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this document.

[0084] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0085] In the embodiments provided herein, 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 couplings or direct couplings or communication connections shown or discussed may be indirect couplings or communication connections through some interfaces, devices, or units, or they may be electrical, mechanical, or other forms of connection.

[0086] 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 the embodiments described herein, depending on actual needs.

[0087] Furthermore, the functional units in the various embodiments of this document 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.

[0088] 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 paper, 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 described in the various embodiments of this paper. 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.

[0089] This document uses specific embodiments to illustrate the principles and implementation methods of this document. The descriptions of the embodiments above are only for the purpose of helping to understand the methods and core ideas of this document. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this document. Therefore, the content of this specification should not be construed as a limitation of this document.

Claims

1. A test data processing method based on semantic analysis, characterized in that, The method includes: The test scenario description text input by the user is semantically segmented, and the segmentation results are classified to obtain the category of each word element. The categories include data body category, constraint condition category, and data identifier category. Initial data is obtained by using lexical units belonging to the data body class to perform preliminary screening of the basic data of the test scenario; The constraint range is determined based on the lexical of the constraint condition class, and the constraint condition field is determined from multiple fields of the initial data. The initial data whose values ​​of the constraint condition field conform to the constraint range are then obtained to obtain the target data. The data processing method is determined based on the lexical representation of the data identifier class, and the data processing field is determined from multiple fields of the target data. The value of the data processing field is processed according to the data processing method to obtain the processing result. The processing result is provided to the user; The initial screening of the basic data for the test scenario using lexical units belonging to the aforementioned data body class further includes: The first score of the initial data is calculated based on the lexical units belonging to the data body class for each of the basic data of the test scenario. The test scenario data whose first score exceeds the first preset threshold is selected as the initial data. The formula for calculating the first score of the basic data for each test scenario as the initial data, based on the data body class, is as follows: P mj = similarity(w i , M j ) + a j + b j ; Among them, P mj This represents the first score, similarity() represents the similarity measurement algorithm, and w i M represents the i-th word element in the data body class. j α represents the name of the basic data for the j-th test scenario. j This indicates that the system uses the basic data M of the test scenario within the specified time period. j frequency coefficient, β j This indicates that the user has been using the basic data M of the test scenario since its creation. j The frequency coefficient.

2. The method according to claim 1, characterized in that, The user-input test scenario description text is semantically segmented, and the segmentation results are classified to obtain the category of each segmented word, which further includes: The test scenario description text is segmented to obtain multiple word units; The word units are vectorized to obtain the word vector for each word segment; The word vectors are then subjected to dimensionality reduction processing; The category of the word element is obtained by calculating the dimensionality-reduced word vectors using a pre-trained KNN model.

3. The method according to claim 2, characterized in that, Before vectorizing the lexical units, the method further includes: The stop words in the word units are filtered according to a preset stop word list so that the word units after filtering the stop words can be vectorized.

4. The method according to claim 1, characterized in that, Determining the constraint fields from multiple fields of the initial data further includes: The second score of the initial data field as the constraint field is calculated based on the historical number of times the initial data field has been used as a constraint field. Select the field with the highest second score as the constraint field.

5. The method according to claim 4, characterized in that, The formula for calculating the second score of the initial data as a constraint field based on the historical number of times the initial data's fields have been used as constraint fields is as follows: P fi = Num 系统使用 (fi field is used as a constraint field) + Num 该用户使用 (fi field is used as a constraint field) × γ; Where fi represents the i-th field in the initial data, P fi The field fi represents the second score of the constraint field, Num 系统使用 (The fi field as a constraint field) represents the total number of times the fi field has been used as a constraint field in the system, Num 该用户使用 (fi field as constraint field) indicates the total number of times the user uses field fi as a constraint field, and γ represents the coefficient.

6. The method according to claim 1, characterized in that, Determining the data processing method based on the lexical representation of the data identifier class further includes: Semantic analysis is performed on the lexical units of the data identifier class to determine the data processing method.

7. The method according to claim 1, characterized in that, The data processing method includes data statistics; Processing the value of the data processing field according to the aforementioned data processing method further includes: The values ​​of the data processing field are statistically analyzed to obtain statistical results, which are then provided to the user.

8. The method according to claim 1, characterized in that, Determining the data processing fields from multiple fields of the target data further includes: Based on the order of the words in the test scenario description text, the word preceding the word of the data identifier class is used as the reference field; The third score of the target data field is calculated based on the historical number of times the target data field has been used as a data processing field under this data processing method and the similarity between the field name of the target data and the name of the reference field. The field with the highest third score is selected as the data processing field.

9. The method according to claim 8, characterized in that, The formula for calculating the third score of the target data field as the data processing field, based on the historical number of times the target data field has been used as a data processing field under this data processing method and the similarity between the field name of the target data and the name of the reference field, is as follows: P ei = Num 系统使用 (ei field used as a data processing field) + Num 该用户使用 (The ei field is used as a data processing field) × ζ + similarity(w1, ei) × η; Where ei represents the i-th field of the target data, P ei The field ei represents the third score of the data processing field, Num. 系统使用 (ei field as a data processing field) indicates the total number of times the system uses the ei field as a data processing field under this data processing method. Num 该用户使用 (ei field as data processing field) indicates the total number of times the user uses field ei as the data processing field under this data processing method, w1 represents the reference field, similarity(w1, ei) represents the similarity between the name of the reference field and the name of field ei, and ζ and η represent coefficients.

10. The method according to claim 1, characterized in that, The category also includes dimension classes; Before providing the processing result to the user, the method further includes: Semantic analysis is performed on the lexical units of the aforementioned dimension class to determine the basis for dimension determination; Determine the dimension fields from multiple fields of the target data; The processing results are grouped according to the values ​​of the dimension fields so that the grouped results can be provided to the user.

11. The method according to claim 10, characterized in that, Determining dimension fields from multiple fields of the target data further includes: The fourth score of the target data field is calculated based on the historical number of times the target data field has been used as a dimension field and the similarity between the field name of the target data and the name on which the dimension is based. The field with the highest fourth score is selected as the dimension field.

12. The method according to claim 11, characterized in that, The formula for calculating the fourth score of a field in the target data as a dimension field, based on the historical frequency of that field being used as a dimension field and the similarity between the field name in the target data and the name on which the dimension is based, is as follows: P gi =Num 系统使用 (gi field as dimension field) + Num 该用户使用 (The gi field is used as a dimension field) ×μ +similarity(w2, gi) ×ν; Where gi represents the i-th field of the target data, P ei This indicates that the field gi is the fourth score of the dimension field, Num 系统使用 (The gi field as a data processing field) indicates the total number of times the system uses the gi field as a dimension field, Num 该用户使用 (gi field as data processing field) indicates the total number of times the user uses field gi as the dimension field, w2 represents the dimension basis, similarity(w2, gi) represents the similarity between the name of the dimension basis and the name of field ei, and μ and ν represent coefficients.

13. The method according to claim 10, characterized in that, Grouping the processing results based on the values ​​of the dimension field further includes: The processing results are grouped according to the values ​​of the dimension fields using Elasticsearch query syntax.

14. A test data processing device based on semantic analysis, characterized in that, The device includes: The test scenario description text processing unit is used to perform semantic word segmentation on the test scenario description text input by the user, and to classify the segmentation results to obtain the category of each word element. The categories include data body category, constraint condition category, and data identifier category. The test scenario basic data screening unit is used to screen the test scenario basic data using lexical units belonging to the data body class to obtain initial data; The constraint processing unit is used to determine the constraint range based on the lexical of the constraint condition class, and to determine the constraint condition field from multiple fields of the initial data, and to determine the initial data whose value of the constraint condition field conforms to the constraint range, thereby obtaining the target data; The data processing unit is used to determine the data processing method based on the lexical units of the data identifier class, determine the data processing field from multiple fields of the target data, calculate the value of the data processing field according to the data processing method, and obtain the processing result. The result providing unit is used to provide the processing result to the user; The initial screening of the basic data for the test scenario using lexical units belonging to the aforementioned data body class further includes: The first score of the initial data is calculated based on the lexical units belonging to the data body class for each of the basic data of the test scenario. The test scenario data whose first score exceeds the first preset threshold is selected as the initial data. The formula for calculating the first score of the basic data for each test scenario as the initial data, based on the data body class, is as follows: P mj = similarity(w i , M j ) + a j + b j ; Among them, P mj This represents the first score, similarity() represents the similarity measurement algorithm, and w i M represents the i-th word element in the data body class. j α represents the name of the basic data for the j-th test scenario. j This indicates that the system uses the basic data M of the test scenario within the specified time period. j frequency coefficient, β j This indicates that the user has been using the basic data M of the test scenario since its creation. j The frequency coefficient.

15. 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 method according to any one of claims 1 to 13.

16. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method of any one of claims 1 to 13.

17. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the method of any one of claims 1 to 13.

Citation Information

Patent Citations

  • User intention prediction method based on artificial intelligence, electronic device and storage medium

    CN111259209A

  • Voice data processing method based on service scene recognition and related device

    CN115374793A