Method, device, storage medium and electronic device for cascading test automation execution

Abstract

The application provides a method and device for cascading test automation execution, a storage medium and an electronic device. A service scenario description text obtained in a historical time period is converted into a full-service scenario constraint vector; a request message graph corresponding to the service scenario description text is compressed into a feature vector; the feature vector and the full-service scenario constraint vector arranged in time are input into a time sequence self-attention module to obtain an updated service scenario constraint vector; a similarity between a predicted feature vector and the feature vector at the same time is obtained, and a text graph mapping model is updated; a constraint vector corresponding to a to-be-tested service is obtained, the updated text graph mapping model is processed, a test request message graph is obtained, and a test message is decoded to be used for testing. The knowledge graph and the time sequence self-attention mechanism are used to realize cascading test automation, and the workload is reduced.

Description

Technical Field

[0001] This application relates to the field of automated testing, and more specifically, to a method, apparatus, storage medium, and electronic device for automating cascade testing. Background Technology

[0002] In the integration testing process of software testing, completing a single business function test requires executing multiple unit tests. The workflow can be divided into five stages: identifying target unit cases, matching cascaded unit cases, assembling functional entity components, generating test messages, and executing tests.

[0003] Currently, testers primarily identify cascading unit cases within functional entities based on business documents and requirements meetings, and then build test case libraries to achieve test case reuse. However, reviewing business documents and conducting requirements communication meetings increases the workload of software testing. Building a test case library requires a keyword-matching retrieval model for case searching and reuse, but this method still lacks flexibility and reusability. Summary of the Invention

[0004] The main objective of this application is to provide a method for automating cascading test execution based on knowledge graphs, an apparatus for automating cascading test execution, a computer-readable storage medium, a processor, and an electronic device, so as to at least solve the problem that current software testing solutions require a large amount of work.

[0005] To achieve the above objectives, according to one aspect of this application, a method for automated execution of cascading tests is provided, comprising: converting business scenario description text acquired within a historical time period into a full business scenario constraint vector; compressing a request message graph corresponding to the business scenario description text into a feature vector, wherein the request message graph uses interface encoding nodes as information aggregation centers, and data item nodes and value nodes as a graph form describing interface functions; inputting the feature vector and the full business scenario constraint vector arranged chronologically into a temporal self-attention module for learning to obtain an updated business scenario constraint vector that updates over time; and obtaining an updated business scenario constraint vector based on the feature vector from the first time to the Tth time and the updated business scenario constraint vector obtained at the Tth time. The predicted feature vector at time T+1 is obtained, and the similarity between the predicted feature vector at time T+1 and the feature vector at time T+1 is obtained. The original text graph mapping model is updated based on the similarity to obtain an updated text graph mapping model, wherein the similarity obtained by processing with the updated text graph mapping model is greater than the similarity threshold. The test full business scenario constraint vector corresponding to the description text of the business scenario to be tested is obtained, and the test full business scenario constraint vector is input into the updated text graph mapping model for processing to obtain the test request message graph corresponding to the description text of the business scenario to be tested. The test request message graph is decoded to obtain the test message, and the test message is used to test the business scenario to be tested.

[0006] Optionally, converting the business scenario description text obtained within the historical time period into a full business scenario constraint vector includes: obtaining a pre-trained language model BERT based on knowledge representation learning (KRL); and using the BERT to convert the business scenario description text obtained within the historical time period into the full business scenario constraint vector.

[0007] Optionally, compressing the request message graph corresponding to the business scenario description text into a feature vector includes: obtaining an EvolveGCN graph convolutional network model based on a temporal knowledge graph (TKG); and using the EvolveGCN to compress the request message graph corresponding to the business scenario description text into the feature vector.

[0008] Optionally, before converting the business scenario description text obtained within the historical time period into a full business scenario constraint vector, the method further includes: obtaining historical transaction logs within the historical time period; performing keyword extraction processing on the historical transaction logs to obtain the business scenario description text.

[0009] Optionally, obtaining the predicted feature vector at time T+1 based on the feature vectors from the first time to time T and the updated business scenario constraint vector obtained at time T includes: using the concat function to concatenate the feature vectors from the first time to time T with the updated business scenario constraint vector obtained at time T, then processing them through a fully connected layer (FC) and a softAMX layer, and finally transforming the variable dimension of the hidden layer to the dimension of the prediction target space to obtain the predicted feature vector at time T+1.

[0010] Optionally, updating the original text graph mapping model based on the similarity to obtain an updated text graph mapping model includes: updating at least the weight values, bias values, and convolution kernel sizes in the original text graph mapping model based on the similarity to obtain the updated text graph mapping model.

[0011] Optionally, compressing the request message graph corresponding to the business scenario description text into a feature vector includes: clustering the request message graph around message nodes; classifying the request message graph according to its distance from the message nodes, and integrating and compressing the information of adjacent nodes and connecting edges into the feature vector; wherein, the formula is used:

[0012]

[0013]

[0014]

[0015]

[0016] Determine the feature vector, where z INF-V σ represents the aggregated feature vector of the message graph; σ represents the activation function, which adopts the Leaky ReLU form to increase the nonlinearity of the model; a k This indicates that the weighting coefficient for the aggregation of data item name nodes adjacent to the interface node INF is less than 1; W k This represents the weight matrix used to calculate the aggregation information of the data item name nodes; x k The feature vector representing the name node of adjacent data items; a v This indicates that the weighting coefficient for the aggregation of adjacent data item value nodes is less than 1; W V The weight matrix represents the aggregation information of adjacent data item value nodes; x v e represents the feature vector of the adjacent data item value node; k,INF This represents the information passed from a specific data item name node to the center; e W,INF This represents the information passed from any data item name node to the center; e v,kThis represents the information passed from a data item value node to the center; N k N represents the set of data item name nodes adjacent to the interface node; v This represents the set of data item value nodes adjacent to the interface node; e u,v This represents the information transmitted from any adjacent node u to the central node v; A u，v This represents the adjacency matrix between nodes in the graph. The value of row i and column j of the adjacency matrix is ​​0 or 1, indicating whether there is a connection between nodes i and j; a T x represents the attention weight matrix; u Let P(distance(u,v)) represent the feature vector of any u node, and E represent the combination of all types of nodes in a certain message graph; where P(distance(u,v)) is used as a penalty term to distinguish between interface vertices and data item nodes, and between data item vertices and value vertices.

[0017] According to another aspect of this application, an apparatus for automating cascading test execution is provided, comprising: a conversion unit for converting business scenario description text acquired within a historical time period into a full business scenario constraint vector; a compression unit for compressing a request message graph corresponding to the business scenario description text into a feature vector, wherein the request message graph uses interface encoding nodes as information aggregation centers, and data item nodes and value nodes as a graph form describing the interface function; a learning unit for inputting the feature vector and the full business scenario constraint vector arranged chronologically into a temporal self-attention module for learning, thereby obtaining an updated business scenario constraint vector that updates over time; and a first processing unit for processing the feature vector from a first time to a time T, and the updated business scenario constraint vector obtained at time T. The first processing unit obtains the predicted feature vector at time T+1 by using the scene constraint vector, and obtains the similarity between the predicted feature vector at time T+1 and the feature vector at time T+1. Based on the similarity, the original text graph mapping model is updated to obtain the updated text graph mapping model, wherein the similarity obtained by processing with the updated text graph mapping model is greater than the similarity threshold. The second processing unit is used to obtain the test full business scene constraint vector corresponding to the description text of the business scene to be tested, input the test full business scene constraint vector into the updated text graph mapping model for processing, obtain the test request message graph corresponding to the description text of the business scene to be tested, decode the test request message graph to obtain the test message, and use the test message to test the business scene to be tested.

[0018] According to another aspect of this application, a computer-readable storage medium is provided, the computer-readable storage medium including a stored program, wherein, when the program is executed, a method for controlling the device where the computer-readable storage medium is located to perform automated execution of cascading tests is provided.

[0019] According to another aspect of this application, an electronic device is provided, comprising: one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including a method for automating cascading test execution.

[0020] By applying the technical solution of this application, a knowledge graph-based matching model is constructed, and target unit cases are obtained through semantic recognition of NLP. These cases are then used to match cascaded unit cases, reducing the workload of assembling functional point entities. The model can also automatically reuse or modify highly similar messages to generate test messages, thereby shortening the software development cycle and improving work efficiency. Attached Figure Description

[0021] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0022] Figure 1 A hardware structure block diagram of a mobile terminal for an automated execution method of cascading tests based on a knowledge graph, provided in an embodiment of this application, is shown.

[0023] Figure 2 A flowchart illustrating an automated execution method for cascading tests based on a knowledge graph, according to an embodiment of this application, is shown.

[0024] Figure 3 The diagram illustrates the process of using the pre-trained language model BERT as a semantic encoder to extract business constraint features and obtain the constraint vector for the entire business scenario.

[0025] Figure 4 The diagram shows the input content of the DynGraphEvo model and the flowchart of G-TKG.

[0026] Figure 5 A schematic diagram illustrating the workflow of software integration testing is shown.

[0027] Figure 6 A structural block diagram of a knowledge graph-based cascaded test automation execution device is shown according to an embodiment of this application. Detailed Implementation

[0028] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0029] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0030] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application 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 for the embodiments of this application described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus 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 apparatus.

[0031] For ease of description, the following explains some of the nouns or terms used in the embodiments of this application:

[0032] Unit example: The smallest executable unit of a test message file at the code level;

[0033] Function point entity: Composed of one or more unit cases in a specific order, it is a collection of test message files that implement the smallest unit function of the business layer;

[0034] Target unit case: In functional point entities, the unit case that plays a semantic identification role or performs the main function;

[0035] Cascading unit case: A unit case that is not the target unit case within the same functional point entity;

[0036] Test messages: Convert unit test cases into files that can be used for testing software, including but not limited to XML, JSON and other formats;

[0037] NLP semantic recognition (Natural Language Processing) is an artificial intelligence technology that extracts the corresponding meaning and significance from text through in-depth analysis and understanding, thereby achieving automated processing and application of text;

[0038] Vertex: Interface name (interface-vertex), data item name (key-vertex), data item value (value-vertex);

[0039] Edges: 1) Contain-Edge: The interface requests the data item, and there is a connection between the interface vertex and the data item vertex; 2) Value-Edge: The connection between the data item and the value.

[0040] As described in the background section, existing software testing solutions require a large amount of work. To address this issue, embodiments of this application provide a method for automated execution of cascading tests based on knowledge graphs, an apparatus for automated execution of cascading tests, a computer-readable storage medium, a processor, and an electronic device.

[0041] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0042] The methods and embodiments provided in this application can be executed on a mobile terminal, computer terminal, or similar computing device. Taking running on a mobile terminal as an example, Figure 1 This is a hardware structure block diagram of a mobile terminal for a method of automating cascading tests based on knowledge graphs, according to an embodiment of the present invention. Figure 1 As shown, a mobile terminal may include one or more ( Figure 1 Only one is shown in the diagram. A processor 102 (which may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA, etc.) and a memory 104 for storing data are also shown. The mobile terminal may further include a transmission device 106 for communication functions and an input / output device 108. Those skilled in the art will understand that... Figure 1 The structure shown is for illustrative purposes only and does not limit the structure of the mobile terminal described above. For example, the mobile terminal may also include components that are more... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown.

[0043] The memory 104 can be used to store computer programs, such as application software programs and modules, like the computer program corresponding to the cascaded test automation execution method in this embodiment of the invention. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, thereby implementing the above-described method. The memory 104 may include high-speed random access memory and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include memory remotely located relative to the processor 102, and these remote memories can be connected to the mobile terminal via a network. Examples of the aforementioned networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof. The transmission device 106 is used to receive or send data via a network. Specific examples of the aforementioned networks may include wireless networks provided by the mobile terminal's communication provider. In one example, the transmission device 106 includes a network interface controller (NIC), which can be connected to other network devices via a base station to communicate with the Internet. In one example, the transmission device 106 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0044] This embodiment provides a method for automating the execution of cascaded tests based on a knowledge graph, which runs on a mobile terminal, computer terminal, or similar computing device. 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. Furthermore, 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.

[0045] Figure 2 This is a flowchart illustrating a method for automating cascading tests based on a knowledge graph, according to an embodiment of this application. Figure 2 As shown, the method includes the following steps:

[0046] Step S201: Convert the business scenario description text obtained within the historical time period into a full business scenario constraint vector;

[0047] The historical time period can be the previous day, the previous month, the previous year, etc.

[0048] For example, business scenarios can include deposit scenarios, withdrawal scenarios, transfer scenarios, and other business scenarios.

[0049] The business scenario description text can be other business scenario description texts such as "Mobile banking, recharge 100 yuan of mobile phone credit to someone else's mobile phone number through a secondary card".

[0050] Step S202: Compress the request message graph corresponding to the business scenario description text into a feature vector. The request message graph uses the interface encoding node as the information aggregation center, and the data item node and value node are a graph form that describes the interface function.

[0051] Specifically, the request message graph belongs to the knowledge graph. The knowledge graph is composed of entities and relations, usually represented in the form of triples: [head (head entity), relation (relation of entities), tail (tail entity)], abbreviated as (h, r, t). In the message knowledge graph sequence, the interface vertices show the order of business requirements calling the backend service interface from the front end, the code logic, and the changing patterns of functional dependencies.

[0052] Step S203: Input the feature vectors arranged in chronological order and the full business scenario constraint vectors into the temporal self-attention module for learning, and obtain the updated business scenario constraint vectors that are updated over time.

[0053] Specifically, by using a pre-designed temporal self-attention module, the evolutionary pattern of the vertex set is modeled for each time period.

[0054] Step S204: Based on the feature vectors from the first time to the Tth time and the updated business scenario constraint vector obtained at the Tth time, the predicted feature vector at the T+1th time is obtained, and the similarity between the predicted feature vector at the T+1th time and the feature vector at the T+1th time is obtained. The original text graph mapping model is updated based on the similarity to obtain the updated text graph mapping model. The similarity obtained by processing with the updated text graph mapping model is greater than the similarity threshold.

[0055] That is, by using the first T feature vectors and the updated business scenario constraint vector obtained at time T, the predicted feature vector at time T+1 can be predicted; the feature vector at time T+1 is the true value of the feature vector at time T+1, and then the predicted value and the true value are compared to obtain the similarity.

[0056] Step S205: Obtain the test full business scenario constraint vector corresponding to the description text of the business scenario to be tested, input the test full business scenario constraint vector into the updated text graph mapping model for processing, obtain the test request message graph corresponding to the description text of the business scenario to be tested, decode the test request message graph to obtain the test message, and use the test message to test the business scenario to be tested.

[0057] For example, the description text of the business scenario to be tested is: Customer A transfers 100 yuan to Customer B; then the test full business scenario constraint vector AAAA corresponding to the description text of the business scenario to be tested is obtained. Then, AAAA is input into the updated text graph mapping model to obtain the test request message graph corresponding to AAAA.

[0058] Specifically, the function point entity files are converted into test messages and the tests are executed; the accuracy of test message generation can be optimized based on the test history in the server, thereby improving the matching degree of cascaded unit cases.

[0059] The cascading test automation method of this application converts the business scenario description text into a full business scenario constraint vector; compresses the request message graph corresponding to the business scenario description text into feature vectors. Then, a temporal self-attention learning mechanism is used for learning; then, using the first T feature vectors and the updated business scenario constraint vector obtained at time T, the predicted feature vector at time T+1 can be predicted; next, the text graph mapping model is updated based on similarity, and then, given the business scenario description text to be tested, the model is used to obtain the test request message graph corresponding to the business scenario description text to be tested, and then the test message is obtained by decoding, and the test message is used to test the business scenario to be tested. This solution uses knowledge graphs and temporal self-attention mechanisms to automate cascading tests, requiring relatively little workload.

[0060] Furthermore, while some existing solutions automate the integration testing process, they rely on the software's high-level design and its corresponding target source code. That is, it is necessary to maintain the high-level design documents for both the "target unit cases" and "cascaded unit cases" to determine the test sequence for each "unit case" within the functional entity. Maintaining these high-level design documents increases the workload of integration testing automation and reduces its flexibility. However, the solution proposed in this application avoids these drawbacks.

[0061] By constructing a knowledge graph-based matching model and obtaining target unit cases through NLP semantic recognition, cascade unit cases can be matched, reducing the workload of assembling functional point entities. The model can also automatically reuse or modify highly similar messages to generate test messages, thereby shortening the software development cycle and improving work efficiency.

[0062] In the embodiments of this application, step S201 converts the business scenario description text obtained within the historical time period into a full business scenario constraint vector, including the following steps:

[0063] Step S2011: Obtain the pre-trained language model BERT based on Knowledge Representation Learning (KRL).

[0064] Step S2012: Use BERT to convert the business scenario description text obtained within the historical time period into a full business scenario constraint vector;

[0065] Specifically, this method uses the pre-trained language model BERT as the semantic encoder to extract constraint vectors. The specific process is as follows: Figure 3 In other words, the BERT model, as a representative of pre-trained language models, has powerful language representation and feature extraction capabilities. Furthermore, by stacking attention-based transformer blocks and having a bidirectional language model structure, it can better supplement the contextual language features to make the transformation constraint vector more accurate.

[0066] Further, step S202 compresses the request message graph corresponding to the business scenario description text into a feature vector, including the following steps:

[0067] Step S2021: Obtain the graph convolutional network model EvolveGCN based on temporal knowledge graph (TKG);

[0068] Step S2022: Use EvolveGCN to compress the request message graph corresponding to the business scenario description text into a feature vector.

[0069] In this method, the temporal knowledge graph is a multi-relational knowledge graph. Compared to conventional knowledge graphs, the relations in a temporal knowledge graph are associated with time periods. Current knowledge graph research mostly focuses on static KGs, which do not change over time, while time-varying KGs are less explored. However, temporal information is very important because much structured knowledge is only valid within a specific time frame, and the changes in facts follow a time series. Recent research has begun to incorporate temporal information into KRL and KGC, termed temporal knowledge graphs. In other words, through use and subsequent upgrades, the EvolveGCN model can more effectively compress the message graph into feature vectors, resulting in more accurate data item nodes and value nodes for the graph's functional interfaces.

[0070] The method may include, before converting the business scenario description text obtained within the historical time period into a full business scenario constraint vector, obtaining historical transaction logs within the historical time period; and performing keyword extraction processing on the historical transaction logs to obtain the business scenario description text.

[0071] In other words, such as Figure 4 As shown, by selecting the historical transaction logs of the account under the business scenario description, the sequence of request messages for successful transactions can be quickly and accurately filtered to form a combination of request sub-graph sequence with request timestamp and business scenario description text, such as: mobile banking, recharging 100 yuan of phone credit to someone else's mobile phone number through a secondary card.

[0072] In this embodiment, the predicted feature vector at time T+1 can be obtained based on the feature vectors from the first time to time T and the new business scenario constraint vector obtained at time T. This includes: using the concat function to link the feature vectors from the first time to time T with the updated business scenario constraint vector obtained at time T, and then processing them through a fully connected layer (FC) and a softAMX layer to transform the variable dimension of the hidden layer to the dimension of the prediction target space, thereby obtaining the predicted feature vector at time T+1.

[0073] Specifically, analogous to the language generation process, the graph at time T+1 is only affected by the constraint vector Cdes and the message graph sequence before time T. The information of the vertex set V and edge set E that make up the graph is encoded and compressed into low-dimensional feature vectors. A mapping relationship is learned, mapping the features of the graph sequence before time T to the vertex set V and edge set E of the message graph at time T+1 under the control of the semantic constraint feature vector. This mapping relationship can be expressed mathematically:

[0074] GT+1=VT+1,ET+1=f(Cdesc;V1,…,VT;E1,…,ET)

[0075] Where VT+1, ET+1 represent the vertex set and edge set at time T+1, Cdesc represents the constraint vector, V1,…,VT represent the vertex set from time 1 to time T, and E1,…,ET represent the edge set from time 1 to time T. Further explanation of this mapping function's function is that at each time step, the mapping function uses the historical dynamic information of the graph sequence evolution to predict the vertex set and edge set in the message graph at the next time step.

[0076] In other words, the feature vectors from the first time step to the Tth time step, and the new business scenario constraint vector obtained at the Tth time step, are used to obtain the predicted feature vector at the T+1th time step. The concat function is used to link the historical semantics of the message sequence and the constraint vector. After processing by the fully connected layer FC and the softamx layer, the predicted feature vector obtained is more accurate.

[0077] Furthermore, the original text graph mapping model can be updated based on similarity to obtain an updated text graph mapping model, including: updating at least the weight values, bias values, and convolution kernel sizes in the original text graph mapping model based on similarity to obtain an updated text graph mapping model.

[0078] Specifically, the constraint feature vectors describing the business scenario, and the graph sequence, gradually complete the target constraints described by the constraint vectors as time steps progress. With the progression of positional encoding, a formula is used to measure the completion similarity between the historical vertex feature set and the constraint vectors; based on this similarity, the influence of the constraint vectors on the generation of the next time step message is gradually reduced. In other words, the weights, biases, and convolution kernels in the graph mapping model corresponding to historical test records with high similarity are more accurate, resulting in a more precise updated text graph mapping model.

[0079] Furthermore, the request message graph corresponding to the business scenario description text is compressed into a feature vector, including: clustering the request message graph around message nodes; classifying the request message graph according to its distance from message nodes, and integrating and compressing the information of adjacent nodes and connecting edges into a feature vector; wherein, the formula is used:

[0080]

[0081]

[0082]

[0083]

[0084] Determine the eigenvector, where z INF-V σ represents the aggregated feature vector of the message graph; σ represents the activation function, which adopts the Leaky ReLU form to increase the nonlinearity of the model; a k This indicates that the weighting coefficient for the aggregation of data item name nodes adjacent to the interface node INF is less than 1; W k This represents the weight matrix used to calculate the aggregation information of the data item name nodes; x k The feature vector representing the name node of adjacent data items; a v This indicates that the weighting coefficient for the aggregation of adjacent data item value nodes is less than 1; W V The weight matrix represents the aggregation information of adjacent data item value nodes; x v e represents the feature vector of the adjacent data item value node; k,INF This represents the information passed from a specific data item name node to the center; e W,INF This represents the information passed from any data item name node to the center; e v,k This represents the information passed from a data item value node to the center; N k N represents the set of data item name nodes adjacent to the interface node; v This represents the set of data item value nodes adjacent to the interface node; e u,v This represents the information transmitted from any adjacent node u to the central node v; A u，vThis represents the adjacency matrix between nodes in the graph. The value of row i and column j of the adjacency matrix is ​​0 or 1, indicating whether there is a connection between nodes i and j; a T x represents the attention weight matrix; u Let P(distance(u,v)) represent the feature vector of any u node, and E represent the combination of all types of nodes in a certain message graph; where P(distance(u,v)) is used as a penalty term to distinguish between interface vertices and data item nodes, and between data item vertices and value vertices.

[0085] Specifically, the request message graph is clustered around message nodes; the request message graph is hierarchically classified according to its distance from message nodes, and the information of adjacent nodes and connecting edges is integrated and compressed into feature vectors. This mechanism is called the progressive attention mechanism.

[0086] Specifically, position-emedding is introduced, using the absolute sequential position in the time step sequence as the step signal for time step updates to distinguish the differences between vertex feature sets at each time step. In other words, the message graph obtained by using the formula has a penalty term that can effectively distinguish the obtained data, making the subsequently obtained feature vectors more accurate.

[0087] To enable those skilled in the art to better understand the technical solution of this application, the implementation process of a knowledge graph-based cascaded test automation method of this application will be described in detail below with reference to specific embodiments.

[0088] This embodiment relates to a specific method for automating cascading test execution based on knowledge graphs, including the following steps:

[0089] Step S1: Obtain historical transaction logs within the historical time period, and extract keywords from the historical transaction logs to obtain business scenario description text. Convert the business scenario description text obtained within the historical time period into a full business scenario constraint vector.

[0090] Step S2: Compress the request message graph corresponding to the business scenario description text into a feature vector. The request message graph uses the interface encoding node as the information aggregation center, and the data item node and value node serve as a graph form that describes the interface function.

[0091] Step S3: Input the feature vectors arranged in chronological order and the constraint vectors of the entire business scenario into the temporal self-attention module for learning, and obtain the updated business scenario constraint vectors that are updated over time;

[0092] Step S4: Using the concat function, the feature vectors from the first time step to the Tth time step are concatenated with the updated business scenario constraint vectors obtained at the Tth time step. After processing through the fully connected layer (FC) and the softamx layer, the variable dimension of the hidden layer is transformed to the dimension of the prediction target space to obtain the predicted feature vector at the T+1 time step. The similarity between the predicted feature vector at the T+1 time step and the feature vector at the T+1 time step is obtained. The original text graph mapping model is updated based on the similarity to obtain the updated text graph mapping model. The similarity obtained by processing the updated text graph mapping model is greater than the similarity threshold.

[0093] Step S5: Obtain the full test business scenario constraint vector corresponding to the description text of the business scenario to be tested. Input the full test business scenario constraint vector into the updated text graph mapping model for processing to obtain the test request message graph corresponding to the description text of the business scenario to be tested. Decode the test request message graph to obtain the test message and use the test message to test the business scenario to be tested.

[0094] In this embodiment, cascading testing is automated by employing knowledge graphs and temporal self-attention mechanisms, thereby minimizing the amount of work required.

[0095] In the integration testing process of software testing, completing even the smallest unit of business function testing requires executing multiple unit tests. For example... Figure 5 As shown, the target unit case is first determined according to business requirements, then the cascaded unit cases are matched, then the functional point entities are assembled, and finally test messages are generated to execute the test.

[0096] This application also provides a cascading test automation execution device. It should be noted that this cascading test automation execution device can be used to execute the method for cascading test automation provided in this application. This device is used to implement the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0097] The following describes an automated execution device for cascaded testing provided in an embodiment of this application.

[0098] Figure 6 This is a schematic diagram of a cascaded test automated execution device according to an embodiment of this application. Figure 6 As shown, the device includes:

[0099] The conversion unit 61 is used to convert the business scenario description text obtained within a historical time period into a full business scenario constraint vector.

[0100] Compression unit 62 is used to compress the request message graph corresponding to the business scenario description text into a feature vector. The request message graph takes the interface encoding node as the information aggregation center, and the data item node and value node are a graph form that describes the interface function.

[0101] Learning unit 63 is used to input the feature vectors arranged in chronological order and the constraint vectors of the entire business scenario into the temporal self-attention module for learning, so as to obtain the updated business scenario constraint vectors that are updated over time.

[0102] The first processing unit 64 is used to obtain the predicted feature vector at time T+1 based on the feature vector from the first time to time T and the updated business scenario constraint vector obtained at time T, and to obtain the similarity between the predicted feature vector at time T+1 and the feature vector at time T+1, and to update the original text graph mapping model based on the similarity to obtain the updated text graph mapping model, wherein the similarity obtained by processing with the updated text graph mapping model is greater than the similarity threshold.

[0103] The second processing unit 65 is used to obtain the test full business scenario constraint vector corresponding to the description text of the business scenario to be tested, input the test full business scenario constraint vector into the updated text graph mapping model for processing, obtain the test request message graph corresponding to the description text of the business scenario to be tested, decode the test request message graph to obtain the test message, and use the test message to test the business scenario to be tested.

[0104] This application discloses an apparatus for automating cascading test execution. A conversion unit converts a business scenario description text into a full business scenario constraint vector; a compression unit compresses the request message graph corresponding to the business scenario description text into a feature vector. A learning unit then uses a temporal self-attention learning mechanism for learning; a first processing unit uses the first T feature vectors and the updated business scenario constraint vector obtained at time T to predict the predicted feature vector at time T+1; a second processing unit updates the text graph mapping model based on similarity, and then, given the business scenario description text to be tested, uses the model to obtain the test request message graph corresponding to the business scenario description text, and then decodes it to obtain the test message, which is then used to test the business scenario to be tested. This solution uses knowledge graphs and a temporal self-attention mechanism to automate cascading tests, requiring relatively little effort.

[0105] As an optional approach, the conversion unit includes a first acquisition module and a conversion module. The first acquisition module acquires a pre-trained language model BERT based on Knowledge Representation Learning (KRL). The conversion module uses BERT to convert the business scenario description text acquired within a historical time period into a full business scenario constraint vector. This method's BERT model possesses powerful language representation and feature extraction capabilities. Furthermore, by stacking attention-based transformer blocks and a bidirectional language model structure, it can better supplement the contextual language features to make the converted constraint vector more accurate.

[0106] An optional approach involves a compression unit comprising a second acquisition module and a compression module. The second acquisition module acquires an EvolveGCN graph convolutional network model based on a temporal knowledge graph (TKG). The compression module uses EvolveGCN to compress the request message graph corresponding to the business scenario description text into feature vectors. This method's EvolveGCN model is algorithmically compatible with the BERT model. Through use and subsequent upgrades, the EvolveGCN model can more effectively compress the message graph into feature vectors, resulting in more accurate data item nodes and value nodes for the obtained graph's functional interfaces.

[0107] In one optional embodiment, the device further includes a first acquisition unit and a third processing unit. The acquisition unit acquires historical transaction logs within the historical time period before converting the business scenario description text acquired within the historical time period into a full business scenario constraint vector. The third processing unit performs keyword extraction processing on the historical transaction logs to obtain the business scenario description text. This method, by selecting historical transaction logs of the account under the business scenario description, can quickly and accurately filter the sequence of request messages for successfully completed transactions, forming a combination of a request subgraph sequence with request timestamps and the business scenario description text. For example, in mobile banking, a secondary card is used to recharge another person's mobile phone number with 100 yuan.

[0108] In one optional approach, the first processing unit includes a processing module. This module uses the `concat` function to concatenate the feature vectors from time 1 to time T with the updated business scenario constraint vector obtained at time T. After processing through a fully connected layer (FC) and a softAMX layer, the variable dimension of the hidden layer is transformed to the dimension of the prediction target space, resulting in the predicted feature vector at time T+1. This method utilizes the flexible `concat` function to concatenate the message sequence's historical semantics and constraint vectors. The predicted feature vector obtained after processing through the FC and softAMX layers is more accurate.

[0109] In one alternative approach, the first processing unit includes an update module. This update module updates at least the weight values, bias values, and convolution kernel sizes in the original text-to-graph mapping model based on similarity, resulting in an updated text-to-graph mapping model. This method yields more accurate weight values, bias values, and convolution kernels in the graph mapping model corresponding to historical test records with high similarity, resulting in a more precise updated text-to-graph mapping model.

[0110] In an optional scheme, the compression unit further includes clustering the request message graph around message nodes; hierarchically classifying the request message graph according to its distance from message nodes, and integrating and compressing the information of adjacent nodes and connecting edges into a feature vector; wherein, the formula is used:

[0111]

[0112]

[0113]

[0114]

[0115] Determine the eigenvector, where z INF-V σ represents the aggregated feature vector of the message graph; σ represents the activation function, which adopts the Leaky ReLU form to increase the nonlinearity of the model; a k This indicates that the weighting coefficient for the aggregation of data item name nodes adjacent to the interface node INF is less than 1; W k This represents the weight matrix used to calculate the aggregation information of the data item name nodes; x k The feature vector representing the name node of adjacent data items; a v This indicates that the weighting coefficient for the aggregation of adjacent data item value nodes is less than 1; W V The weight matrix represents the aggregation information of adjacent data item value nodes; x v e represents the feature vector of the adjacent data item value node; k,INF This represents the information passed from a specific data item name node to the center; e W,INF This represents the information passed from any data item name node to the center; e v,k This represents the information passed from a data item value node to the center; N k N represents the set of data item name nodes adjacent to the interface node; v This represents the set of data item value nodes adjacent to the interface node; e u,v This represents the information transmitted from any adjacent node u to the central node v; A u，v This represents the adjacency matrix between nodes in the graph. The value of row i and column j of the adjacency matrix is ​​0 or 1, indicating whether there is a connection between nodes i and j; a Tx represents the attention weight matrix; u Let P(distance(u,v)) represent the feature vector of any u node, and E represent the combination of all types of nodes in a certain message graph; where P(distance(u,v)) is used as a penalty term to distinguish between interface vertices and data item nodes, and between data item vertices and value vertices.

[0116] An apparatus for automating cascaded testing includes a processor and a memory. The aforementioned conversion unit, compression unit, learning unit, etc., are all stored as program units in the memory. The processor executes these program units stored in the memory to achieve the corresponding functions. All of the above modules reside in the same processor; alternatively, the modules may be located in different processors in any combination.

[0117] The processor contains a kernel that retrieves the corresponding program units from memory. One or more kernels can be configured; adjusting kernel parameters reduces the workload of assembling function point entities and shortens the software development cycle. Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory includes at least one memory chip.

[0118] This invention provides a processor for running a program, wherein the program executes a knowledge graph-based cascading test automation method during runtime.

[0119] Specifically, a method for automating cascading test execution based on knowledge graphs includes:

[0120] Step S201: Convert the business scenario description text obtained within the historical time period into a full business scenario constraint vector;

[0121] Step S202: Compress the request message graph corresponding to the business scenario description text into a feature vector. The request message graph uses the interface encoding node as the information aggregation center, and the data item node and value node are a graph form that describes the interface function.

[0122] Step S203: Input the feature vectors arranged in chronological order and the full business scenario constraint vectors into the temporal self-attention module for learning, and obtain the updated business scenario constraint vectors that are updated over time.

[0123] Step S204: Based on the feature vectors from the first time to the Tth time and the updated business scenario constraint vector obtained at the Tth time, the predicted feature vector at the T+1th time is obtained, and the similarity between the predicted feature vector at the T+1th time and the feature vector at the T+1th time is obtained. The original text graph mapping model is updated based on the similarity to obtain the updated text graph mapping model. The similarity obtained by processing with the updated text graph mapping model is greater than the similarity threshold.

[0124] Step S205: Obtain the full test business scenario constraint vector corresponding to the description text of the business scenario to be tested, input the full test business scenario constraint vector into the updated text graph mapping model for processing, obtain the test request message graph corresponding to the description text of the business scenario to be tested, decode the test request message graph to obtain the test message, and use the test message to test the business scenario to be tested.

[0125] This application also provides an electronic device, including a processor, a memory, and a program stored in the memory and executable on the processor, wherein the processor executes the program to perform at least the following steps:

[0126] Step S201: Convert the business scenario description text obtained within the historical time period into a full business scenario constraint vector;

[0127] Step S202: Compress the request message graph corresponding to the business scenario description text into a feature vector. The request message graph uses the interface encoding node as the information aggregation center, and the data item node and value node are a graph form that describes the interface function.

[0128] Step S203: Input the feature vectors arranged in chronological order and the full business scenario constraint vectors into the temporal self-attention module for learning, and obtain the updated business scenario constraint vectors that are updated over time.

[0129] Step S204: Based on the feature vectors from the first time to the Tth time and the updated business scenario constraint vector obtained at the Tth time, the predicted feature vector at the T+1th time is obtained, and the similarity between the predicted feature vector at the T+1th time and the feature vector at the T+1th time is obtained. The original text graph mapping model is updated based on the similarity to obtain the updated text graph mapping model. The similarity obtained by processing with the updated text graph mapping model is greater than the similarity threshold.

[0130] Step S205: Obtain the full test business scenario constraint vector corresponding to the description text of the business scenario to be tested, input the full test business scenario constraint vector into the updated text graph mapping model for processing, obtain the test request message graph corresponding to the description text of the business scenario to be tested, decode the test request message graph to obtain the test message, and use the test message to test the business scenario to be tested.

[0131] The devices mentioned in this article can be servers, PCs, tablets, mobile phones, etc.

[0132] This application also provides a computer program product, which, when executed on a data processing device, is suitable for executing an initialization program having at least the following method steps:

[0133] Step S201: Convert the business scenario description text obtained within the historical time period into a full business scenario constraint vector;

[0134] Step S202: Compress the request message graph corresponding to the business scenario description text into a feature vector. The request message graph uses the interface encoding node as the information aggregation center, and the data item node and value node are a graph form that describes the interface function.

[0135] Step S203: Input the feature vectors arranged in chronological order and the full business scenario constraint vectors into the temporal self-attention module for learning, and obtain the updated business scenario constraint vectors that are updated over time.

[0136] Step S204: Based on the feature vectors from the first time to the Tth time and the updated business scenario constraint vector obtained at the Tth time, the predicted feature vector at the T+1th time is obtained, and the similarity between the predicted feature vector at the T+1th time and the feature vector at the T+1th time is obtained. The original text graph mapping model is updated based on the similarity to obtain the updated text graph mapping model. The similarity obtained by processing with the updated text graph mapping model is greater than the similarity threshold.

[0137] Step S205: Obtain the full test business scenario constraint vector corresponding to the description text of the business scenario to be tested, input the full test business scenario constraint vector into the updated text graph mapping model for processing, obtain the test request message graph corresponding to the description text of the business scenario to be tested, decode the test request message graph to obtain the test message, and use the test message to test the business scenario to be tested.

[0138] It is obvious to those skilled in the art that the modules or steps of the present invention described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. They can be implemented using computer-executable program code, and thus can be stored in a storage device for execution by a computing device. In some cases, the steps shown or described can be performed in a different order than those described herein, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0139] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0140] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0141] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0142] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0143] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0144] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0145] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0146] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0147] As can be seen from the above description, the embodiments of this application achieve the following technical effects:

[0148] 1) The method for automating cascading tests in this application converts the business scenario description text obtained within a historical time period into a full business scenario constraint vector; compresses the request message graph corresponding to the business scenario description text into a feature vector, wherein the request message graph uses the interface encoding node as the information aggregation center, and the data item node and value node serve as a graph form describing the interface function; inputs the feature vector and the full business scenario constraint vector arranged chronologically into a temporal self-attention module for learning, thereby obtaining an updated business scenario constraint vector that updates over time; and obtains the time-to-time constraint vector based on the feature vector from the first time to the Tth time and the updated business scenario constraint vector obtained at the Tth time. The process involves predicting feature vectors and obtaining the similarity between the predicted feature vectors at time T+1 and the feature vectors at time T+1. Based on this similarity, the original text graph mapping model is updated to obtain an updated text graph mapping model, where the similarity obtained using the updated model is greater than a similarity threshold. Next, the process acquires the full-scale test scenario constraint vectors corresponding to the description text of the business scenario to be tested. These constraint vectors are then input into the updated text graph mapping model for processing, resulting in a test request message graph corresponding to the description text of the business scenario to be tested. The test request message graph is then decoded to obtain the test message, which is used to test the business scenario to be tested. By constructing a knowledge graph-based matching model and using NLP semantic recognition to obtain target unit cases, cascade unit cases are matched, reducing the workload of assembling functional entity points. This model can also automatically reuse or modify highly similar messages to generate test messages, thereby shortening the software development cycle and improving work efficiency.

[0149] 2) The device for cascading test automation in this application includes a conversion unit that converts business scenario description text acquired within a historical time period into a full business scenario constraint vector; a compression unit that compresses the request message graph corresponding to the business scenario description text into a feature vector, wherein the request message graph uses the interface encoding node as the information aggregation center, and the data item node and value node serve as a graph form describing the interface function; a learning unit that inputs the feature vectors arranged chronologically and the full business scenario constraint vector into a temporal self-attention module for learning, thereby obtaining an updated business scenario constraint vector that updates over time; and a first processing unit that obtains the first business scenario constraint vector based on the feature vectors from the first time to the Tth time and the updated business scenario constraint vector obtained at the Tth time. The predicted feature vector at time T+1 is obtained, and the similarity between the predicted feature vector at time T+1 and the feature vector at time T+1 is acquired. Based on the similarity, the original text graph mapping model is updated to obtain the updated text graph mapping model. The similarity obtained by processing the updated text graph mapping model is greater than the similarity threshold. The second processing unit obtains the test full business scenario constraint vector corresponding to the description text of the business scenario to be tested. The test full business scenario constraint vector is input into the updated text graph mapping model for processing to obtain the test request message graph corresponding to the description text of the business scenario to be tested. The test request message graph is decoded to obtain the test message, and the test message is used to test the business scenario to be tested. By constructing a knowledge graph-based matching model and obtaining target unit cases through semantic recognition of NLP, the workload of assembling functional point entities is reduced. This model can also automatically reuse or modify highly similar messages to realize the generation of test messages, thereby shortening the software development cycle and improving work efficiency.

[0150] The above are merely preferred embodiments of this application and are not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for automating cascading test execution, characterized in that, include: Convert the business scenario description text obtained within the historical time period into a full business scenario constraint vector; The request message graph corresponding to the business scenario description text is compressed into a feature vector. The request message graph uses the interface encoding node as the information aggregation center, and the data item node and value node are a graph form that describes the interface function. The feature vectors arranged in chronological order and the full business scenario constraint vectors are input into the temporal self-attention module for learning, so as to obtain the updated business scenario constraint vectors that are updated over time. Based on the feature vectors from the first time to the Tth time and the updated business scenario constraint vector obtained at the Tth time, the predicted feature vector at the T+1th time is obtained, and the similarity between the predicted feature vector at the T+1th time and the feature vector at the T+1th time is obtained. The original text graph mapping model is updated based on the similarity to obtain the updated text graph mapping model, wherein the similarity obtained by processing with the updated text graph mapping model is greater than the similarity threshold. Obtain the full business scenario constraint vector corresponding to the description text of the business scenario to be tested, input the full business scenario constraint vector into the updated text graph mapping model for processing, obtain the test request message graph corresponding to the description text of the business scenario to be tested, decode the test request message graph to obtain the test message, and use the test message to test the business scenario to be tested; Converting business scenario description text acquired within a historical time period into a full business scenario constraint vector includes: acquiring a pre-trained language model BERT based on knowledge representation learning (KRL); and using the BERT to convert the business scenario description text acquired within the historical time period into the full business scenario constraint vector. The process of compressing the request message graph corresponding to the business scenario description text into a feature vector includes: obtaining an EvolveGCN graph convolutional network model based on a temporal knowledge graph (TKG); and using the EvolveGCN to compress the request message graph corresponding to the business scenario description text into the feature vector.

2. The method according to claim 1, characterized in that, Before converting the business scenario description text obtained within the historical time period into a full business scenario constraint vector, the method further includes: Obtain historical transaction logs within the specified historical time period; Keyword extraction is performed on the historical transaction logs to obtain the business scenario description text.

3. The method according to claim 1, characterized in that, The predicted feature vector for time T+1 is obtained based on the feature vectors from time 1 to time T and the updated business scenario constraint vector obtained at time T, including: The concat function is used to concatenate the feature vectors from the first time step to the Tth time step with the updated business scenario constraint vector obtained at the Tth time step. After processing by a fully connected layer (FC) and a softmax layer, the variable dimension of the hidden layer is transformed to the dimension of the prediction target space to obtain the prediction feature vector at the T+1th time step.

4. The method according to claim 1, characterized in that, The original text graph mapping model is updated based on the similarity to obtain the updated text graph mapping model, including: Based on the similarity, at least the weight values, bias values, and convolution kernel sizes in the original text graph mapping model are updated to obtain the updated text graph mapping model.

5. The method according to claim 1, characterized in that, The request message graph corresponding to the business scenario description text is compressed into a feature vector, including: The request message graph will be clustered around message nodes; The request message graph is classified according to the distance to the message node, and the information of adjacent nodes and connecting edges is integrated and compressed into the feature vector. The formula used is: ； ； ； ； Determine the feature vector, wherein, σ represents the aggregated feature vector of the message graph; σ represents the activation function, which adopts the form of Leaky ReLU to increase the non-linearity of the model. The weighting coefficient for the aggregation of data item name nodes adjacent to the interface node INF is less than 1; This represents a weight matrix that aggregates information from the data item name nodes. The feature vector representing the name node of the adjacent data item; This indicates that the weighting coefficient for the aggregation of adjacent data item value nodes is less than 1; A weight matrix representing the aggregation information of adjacent data item value nodes; The feature vector representing the node whose value is adjacent to the data item; This represents the information that a data item name node passes to the center; This represents the information passed from any data item name node to the center; This represents the information that a data item's value node transmits to the center; This represents the set of data item name nodes adjacent to the interface node; This represents the set of data item value nodes adjacent to the interface node; This represents the information transmitted from any adjacent node u to the central node v; This represents the adjacency matrix between nodes in the graph. The values ​​of row i and column j of the adjacency matrix are 0 or 1, indicating whether there is a connection between nodes i and j. Represents the attention weight matrix; Let P(distance(u, v)) represent the feature vector of any u node, and let E represent the set of all types of nodes in a certain message graph. P(distance(u, v)) is used as a penalty term to distinguish between interface vertices and data item nodes, and between data item vertices and value vertices.

6. A device for automating cascaded testing, characterized in that, include: The transformation unit is used to convert the business scenario description text obtained within a historical time period into a full business scenario constraint vector. The compression unit is used to compress the request message graph corresponding to the business scenario description text into a feature vector, wherein the request message graph takes the interface encoding node as the information aggregation center, and the data item node and value node as a graph form describing the interface function. The learning unit is used to input the feature vectors arranged in chronological order and the full business scenario constraint vectors into the temporal self-attention module for learning, so as to obtain the updated business scenario constraint vectors that are updated over time. The first processing unit is configured to obtain the predicted feature vector at time T+1 based on the feature vector from the first time to time T and the updated business scenario constraint vector obtained at time T, and to obtain the similarity between the predicted feature vector at time T+1 and the feature vector at time T+1, and to update the original text graph mapping model based on the similarity to obtain the updated text graph mapping model, wherein the similarity obtained by processing with the updated text graph mapping model is greater than the similarity threshold; The second processing unit is used to obtain the test full business scenario constraint vector corresponding to the description text of the business scenario to be tested, input the test full business scenario constraint vector into the updated text graph mapping model for processing, obtain the test request message graph corresponding to the description text of the business scenario to be tested, decode the test request message graph to obtain the test message, and use the test message to test the business scenario to be tested. The conversion unit includes a first acquisition module and a conversion module. The first acquisition module is used to acquire a pre-trained language model BERT based on knowledge representation learning (KRL). The conversion module is used to use the BERT to convert the business scenario description text acquired within the historical time period into the full business scenario constraint vector. The compression unit includes a second acquisition module and a compression module. The second acquisition module is used to acquire an EvolveGCN graph convolutional network model based on temporal knowledge graph (TKG). The compression module is used to compress the request message graph corresponding to the business scenario description text into the feature vector using the EvolveGCN.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored program, wherein, when the program is executed, it controls the device on which the computer-readable storage medium is located to perform the method for automating the cascading test as described in any one of claims 1 to 5.

8. An electronic device, characterized in that, include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including a method for performing automated execution of cascading tests as described in any one of claims 1 to 5.

Images