Software test scheme determination method and device and computer equipment

Abstract

The invention relates to a software test scheme determination method and device and computer equipment. Comprising the following steps: in response to a test scheme generation instruction of a target tester on to-be-tested software, analyzing software test behavior data and emotion feedback data of the target tester through a preset portrait construction model, and generating psychological portrait data of the target tester; inputting the software test behavior data, the emotion feedback data and the psychological portrait data into a preset test agent, and generating a software test strategy of the target tester; and generating a test scheme corresponding to the target tester according to the software test strategy and the psychological portrait data. By means of the software testing method and device, targeted software testing schemes can be formulated for different software testing personnel.

Description

Technical Field

[0001] This application relates to the field of software test optimization technology, and in particular to a method, apparatus and computer equipment for determining a software test scheme. Background Technology

[0002] Software testing refers to the process of testing software for its functionality, performance, security, and other characteristics. In related technologies, companies typically have many software testers, and they often develop common software testing protocols for different software. When testing a particular piece of software, the relevant testers will use this common testing protocol and follow the established procedures. However, different software testers often have different areas of expertise, operating habits, and focuses. This means that using a common testing protocol may not maximize the testing efficiency of the testers, nor can it guarantee the accuracy of the test results.

[0003] Therefore, how to develop targeted software testing plans for different software testers has become an urgent technical problem to be solved. Summary of the Invention

[0004] Therefore, it is necessary to provide a method, apparatus, and computer equipment for determining a software testing scheme to address the aforementioned technical problems.

[0005] Firstly, this application provides a method for determining a software test plan, applied to a software test plan generation system, comprising:

[0006] In response to the test plan generation instructions from the target testers for the software under test, the system analyzes the software testing behavior data and emotional feedback data of the target testers through a pre-set profile building model to generate psychological profile data of the target testers.

[0007] Software testing behavior data, emotional feedback data, and psychological profile data are input into a pre-set testing agent to generate software testing strategies for the target testers.

[0008] Based on software testing strategies and psychological profiling data, test plans are generated for the target testers.

[0009] In one embodiment, the method further includes:

[0010] Based on the software testing strategy and psychological profile data, a corresponding test plan is generated for the target testers. User feedback data is obtained by the target testers testing the software under test according to the test plan.

[0011] Based on user feedback data, test plans, psychological profile data, and software testing behavior data, the profile building model and test agent are updated.

[0012] In one embodiment, updating the profile building model and the test agent includes:

[0013] The test knowledge graph is updated based on user feedback data, test plans, psychological profile data, and software testing behavior data. The test knowledge graph is built based on the target testers' historical test plans and historical user feedback data of the test plans they executed, the target testers' historical psychological profile data, and historical software testing behavior data.

[0014] The profile building model and test agent are updated based on the updated test knowledge graph.

[0015] In one embodiment, updating the profile building model and the test agent based on the updated test knowledge graph includes:

[0016] The updated test knowledge graph is input into a pre-defined graph neural network, and updated tester data is extracted based on the graph neural network. The updated tester data includes user feedback data, psychological profile data, software testing behavior data, and data after the test plan is updated.

[0017] The profile building model and the test agent are updated based on the updated data from the testers.

[0018] In one embodiment, a test plan corresponding to the target tester is generated based on the software testing strategy and psychological profile data, including:

[0019] Obtain user feedback data for the software under test;

[0020] Based on user feedback data, as well as software testing strategies and psychological profile data, corresponding testing plans are generated for the target testers.

[0021] In one embodiment, before generating a corresponding test plan for the target tester based on the software testing strategy and psychological profile data, the method further includes:

[0022] Acquire multimodal interaction data of the target testers; multimodal interaction data includes visual data, voice data, and behavioral data of the target testers obtained during the execution of historical test plans;

[0023] Multimodal fusion is performed on multimodal interaction data to obtain the sentiment recognition results and intent recognition results of the target testers regarding the software testing strategy;

[0024] The software testing strategy was adjusted based on the emotion recognition and intent recognition results.

[0025] In one embodiment, the process of building the test agent includes:

[0026] Acquire sample psychological data, sample behavioral data, and sample emotional data from various sample groups;

[0027] The sample psychological data, sample behavioral data, and sample emotional data of each sample person are input into the initial test agent to train the initial test agent, and the value of the loss function corresponding to the initial test agent is calculated based on the training results of the initial test agent.

[0028] Once the value of the loss function determines that the test agent has reached the convergence condition, the trained test agent is obtained.

[0029] In one embodiment, the initial test agent includes a deep learning model and a simulated test environment established based on the test requirements of the software under test; training the initial test agent and calculating the value of the loss function corresponding to the initial test agent based on the training results of the initial test agent includes:

[0030] The sample psychological data, sample behavioral data, and sample emotional data of each sample person are input into the deep learning model, and the deep learning model is trained to obtain the sample testing strategy.

[0031] The sample test strategy is input into the simulated test environment and executed to obtain the strategy execution result;

[0032] The reward value corresponding to the strategy execution result is calculated based on the preset evaluation indicators;

[0033] The value of the loss function is calculated based on the reward value and the reinforcement algorithm.

[0034] Secondly, this application also provides a software testing scheme determination apparatus, comprising:

[0035] The acquisition module is used to respond to the test plan generation instructions of the target testers for the software to be tested. It analyzes the software testing behavior data and emotional feedback data of the target testers through a preset profile building model to generate psychological profile data of the target testers.

[0036] The calculation module is used to input software testing behavior data, emotional feedback data, and psychological profile data into a preset testing agent to generate software testing strategies for the target testers.

[0037] The generation module is used to generate test plans corresponding to target testers based on software testing strategies and psychological profile data.

[0038] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0039] In response to the test plan generation instructions from the target testers for the software under test, the system analyzes the software testing behavior data and emotional feedback data of the target testers through a pre-set profile building model to generate psychological profile data of the target testers.

[0040] Software testing behavior data, emotional feedback data, and psychological profile data are input into a pre-set testing agent to generate software testing strategies for the target testers.

[0041] Based on software testing strategies and psychological profiling data, test plans are generated for the target testers.

[0042] The aforementioned method, apparatus, and computer equipment for determining a software testing plan first respond to a test plan generation instruction from a target tester for the software under test. They analyze the target tester's software testing behavior data and emotional feedback data using a pre-defined profile building model to generate a psychological profile of the target tester. Then, the software testing behavior data, emotional feedback data, and psychological profile data are input into a pre-defined testing agent to generate a software testing strategy for the target tester. Finally, a test plan corresponding to the target tester is generated based on the software testing strategy and the psychological profile data. This application allows for the development of tailored software testing plans for different software testers, thereby maximizing the strengths of software testers and improving the efficiency and accuracy of software testing. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 This is a diagram illustrating the application environment of a software testing scheme determination method in one embodiment.

[0045] Figure 2 This is a flowchart illustrating a method for determining a software testing scheme in one embodiment;

[0046] Figure 3 This is a flowchart illustrating the model update method in one embodiment;

[0047] Figure 4This is a flowchart illustrating a test plan generation method in one embodiment;

[0048] Figure 5 This is a flowchart illustrating a software strategy adjustment method in one embodiment;

[0049] Figure 6 This is a flowchart illustrating a method for constructing a test agent in one embodiment;

[0050] Figure 7 This is a schematic diagram of the process of training an initial test agent in one embodiment;

[0051] Figure 8 This is a flowchart illustrating the method for determining a software testing scheme in a preferred embodiment;

[0052] Figure 9 This is a structural block diagram of a software test scheme determination device in one embodiment. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0054] The technical background of this application will be explained below.

[0055] In related technologies, when testing software, a generalized testing strategy is usually adopted. That is, a general software testing plan is formulated for different software, and different software testers also perform software testing operations on the software under test based on this general software testing plan.

[0056] However, different testers have different areas of expertise, skills, and operating habits. Using a universal software testing solution for all testers will not maximize their testing efficiency, limit their ability to utilize their skills, and fail to guarantee the accuracy of the software testing results.

[0057] In summary, the above description illustrates the main problems with the relevant technologies: the software testing methods in these technologies use generalized testing strategies, making it difficult to accurately match the personalized needs of different testers.

[0058] Based on this, embodiments of this application provide a method for determining a software testing plan: This method is applied to a software testing plan generation system. Responding to a test plan generation instruction from a target tester for the software under test, it analyzes the target tester's software testing behavior data and emotional feedback data using a preset profile building model to generate psychological profile data for the target tester. The software testing behavior data, emotional feedback data, and psychological profile data are input into a preset testing agent for simulated learning to generate a software testing strategy for the target tester. A test plan corresponding to the target tester is generated based on the software testing strategy and psychological profile data. This application allows for targeted testing plans tailored to the characteristics and psychology of different software testers, enabling each tester to fully leverage their strengths, reducing unnecessary testing steps, and improving testing efficiency and the accuracy of test results.

[0059] The software testing scheme determination method provided in this application embodiment can be applied to, for example, Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated on server 104, or it can be located on the cloud or other network servers, and the software test plan generation system can be integrated on server 104. First, in response to the test plan generation instruction from the target tester for the software under test, the system analyzes the target tester's software testing behavior data and emotional feedback data through a preset profile building model to generate psychological profile data of the target tester. The software testing behavior data, emotional feedback data, and psychological profile data are then input into a preset test agent for simulation learning to generate the target tester's software testing strategy. Finally, a test plan corresponding to the target tester is generated based on the software testing strategy and psychological profile data. Terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, etc. Server 104 can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services.

[0060] In one exemplary embodiment, such as Figure 2 As shown, a method for determining a software testing scheme is provided, which is applied to... Figure 1 Taking server 104 as an example, this system integrates a software test plan generation system, including:

[0061] Step S210: In response to the test plan generation instruction of the target tester for the software to be tested, the software testing behavior data and emotional feedback data of the target tester are analyzed through a preset profile construction model to generate psychological profile data of the target tester.

[0062] In this embodiment, the target tester is any software test engineer; that is, this application aims to generate a corresponding test plan for the target tester. The aforementioned test plan generation instruction is a trigger signal issued by the target tester, used to instruct the software test plan generation system to generate a test plan suitable for the target tester.

[0063] First, the software testing behavior data and emotional feedback data of the target tester are continuously acquired and recorded. This data can be historical data recorded when the target tester previously used the software test plan generation system. For example, if a test plan needs to be generated for the target tester for the nth time, the software testing behavior data and emotional feedback data would be the behavior data and emotional feedback data recorded when the target tester performed software testing before the nth time. Understandably, if this is the target tester's first time using the software test plan generation system, the parameters in the system can use the initialization parameters.

[0064] The aforementioned software testing behavior data refers to data recorded during the target tester's historical software testing work (i.e., the software testing operations prior to the nth time mentioned above), reflecting the target tester's operating habits and preferences. Examples include the execution order of test cases, the depth and type of software defects discovered, commonly used testing tools, dwell time and testing time on specific functions / modules, etc. The aforementioned emotional feedback data refers to data recorded during the target tester's historical software testing work, reflecting the target tester's emotional state. Examples include facial expressions (which can be captured by a computer webcam), the force and frequency of keyboard typing, mouse movement trajectory, etc. In practical applications, the target tester's emotional characteristics can also be obtained through a pre-provided test questionnaire.

[0065] Having obtained the software testing behavior data and emotional feedback data of the target tester through the above methods, the software testing behavior data and emotional feedback data can be input into a preset profile building model to construct psychological profile data of the target tester. This profile building model can be a machine learning model, such as a classifier or clustering model, or a deep learning model, such as an attention-based deep learning model, like BERT (Bidirectional Encoder Representations from Transformers) or GPT (Generative Pre-trained Transformer). The output psychological profile data is structured data used to quantify the psychological characteristics, emotional characteristics, and operational habits of the target tester. For example, the psychological profile data of the target tester could be: conservative, detail-oriented, highly patient, etc.

[0066] Step S220: Input the software testing behavior data, emotional feedback data, and psychological profile data into the preset testing agent to generate the software testing strategy for the target tester.

[0067] In this embodiment, the aforementioned testing agent is composed of a Markov decision process and a deep learning model. Specifically, the software testing strategy recommendation problem is modeled as a Markov decision process. The agent continuously learns through trial and error in a simulated environment to optimize the testing strategy. The deep learning model is used to process complex test data, improving the efficiency and accuracy of strategy learning. The deep learning model includes, but is not limited to, models such as convolutional neural networks and recurrent neural networks. In a preferred embodiment, a Transformer-based encoder-decoder model can be used. The encoder processes historical test data (i.e., the aforementioned software testing behavior data and emotional feedback data) and psychological profile data, while the decoder generates a testing strategy corresponding to the target tester. In summary, in practical applications, the target tester's software testing behavior data, emotional feedback data, and psychological profile data can be input into the testing agent. First, the testing agent fuses the various types of data into a state vector, and then inputs this state vector into the deep learning network within the testing agent to generate the corresponding testing strategy. It is understood that during the testing phase, multiple simulations are unnecessary; only one forward computation is required to obtain the aforementioned software testing strategy.

[0068] Among them, the software testing strategy is the general strategy of the testing plan that is adapted to the target tester. For example, the target tester is more concerned with the software's problem-solving, security and usability, as well as whether the software's documentation and instructions are clear and accurate, so as to provide guidance for the generation of specific test plans.

[0069] Step S230: Generate a test plan corresponding to the target tester based on the software testing strategy and psychological profile data.

[0070] In this embodiment, the aforementioned test plan refers to a specific, directly executable test plan that is ultimately generated and adapted to the target tester. This includes, but is not limited to, test methods such as stress testing, security testing, and usability testing to evaluate software performance and user experience; and test tools such as those capable of automatically detecting software performance, security vulnerabilities, and user experience issues, including stress testing tools (e.g., JMeter), security scanning tools (e.g., Open Web Application Security Project Zed Attack Proxy, OWASP ZAP), and usability testing tools (e.g., Silverback). It is also understood that the psychological preferences of testers are fully considered during the test plan generation process to improve their execution efficiency and comfort. For example, for detail-oriented target testers, the test plan may include more granular test step checkpoints, etc. In practical applications, corresponding test plans can be generated through pre-trained deep learning networks, or the final test plan can be generated based on preset test strategies and psychological profiles, and multiple mapping relationships between these test plans.

[0071] Through the embodiments of this application, psychological profile data corresponding to the target tester is constructed based on the acquired behavioral data and emotional feedback data. This allows for the efficient and accurate determination of the target tester's psychological characteristics, personality traits, etc., and then the corresponding software testing strategy is calculated based on the software testing behavioral data, emotional feedback data, and psychological profile data. This method effectively combines the target tester's personality, psychology, and operating habits. Finally, a specific test plan is generated based on the software testing strategy and psychological profile data. This step effectively integrates the target tester's testing habits, psychological profile, and other data to generate a test plan highly adapted to the target tester, facilitating the target tester to leverage their strengths. Furthermore, since the testing focus, specific testing methods, and testing tools used in the test plan all conform to the target tester's personality traits and psychological preferences, the target tester can complete the testing task more efficiently, ensuring the accuracy, efficiency, and depth of the testing task.

[0072] In one exemplary embodiment, such as Figure 3 As shown, the method also includes:

[0073] Step S310: After generating a corresponding test plan for the target tester based on the software testing strategy and psychological profile data, obtain user feedback data obtained by the target tester testing the software under test according to the test plan.

[0074] In this embodiment of the application, after obtaining the above-mentioned test plan, user feedback data of the target testers on the test plan can be obtained accordingly. The user feedback data includes, but is not limited to, subjective feedback data, such as the target testers' evaluation of the test plan, and objective feedback data, such as the test results after performing test operations based on the test plan, such as the test comprehensiveness, the number of defects detected based on the test plan, the severity level, the test efficiency of the test plan, etc.

[0075] Step S320: Update the profile building model and the testing agent based on user feedback data, test plan, psychological profile data and software testing behavior data.

[0076] After obtaining user feedback data, test plans, psychological profile data, and software testing behavior data, the profile building model and the testing agent are updated respectively. That is, the profile building model and the testing agent are fine-tuned respectively. The update strategy in this embodiment includes, but is not limited to, periodic batch updates or real-time incremental learning updates, etc.

[0077] Through the embodiments of this application, the profile building model and the testing agent can continuously learn, constantly absorb feedback data such as new user data, optimize model performance, and enhance the model's self-evolution capability.

[0078] In one exemplary embodiment, updating the profile building model and the test agent includes:

[0079] The test knowledge graph is updated based on user feedback data, test plans, psychological profile data, and software testing behavior data. The test knowledge graph is built based on the target testers' historical test plans and historical user feedback data of the test plans they executed, the target testers' historical psychological profile data, and historical software testing behavior data.

[0080] The profile building model and test agent are updated based on the updated test knowledge graph.

[0081] This application provides a specific method for updating the profile construction model and the testing agent. The method includes first constructing a testing knowledge graph, which can be built from all acquired testing data. Specifically, it is constructed from historical test plans corresponding to each target tester and their historical user feedback data, historical psychological profile data of the target testers, and historical software testing behavior data. Specifically, the heterogeneous data can be used as nodes, such as tester A, defect B, testing tool C, and emotional feature D, and the relationships between them can be used as edges, such as "detected by...testing tool...", "...defect has...features", etc., thereby obtaining a unified and scalable testing knowledge graph for the testing domain.

[0082] Furthermore, the newly acquired user feedback data, test plans, psychological profile data, and software testing behavior data are integrated into the test knowledge graph, and the knowledge graph is updated. Based on the data contained in the updated test knowledge graph, the profile building model and the test agent are updated.

[0083] Through the embodiments of this application, the profile building model and the test agent can be updated respectively based on the high-quality, high-information-density data contained in the test knowledge graph, transforming the originally independent data into an interconnected knowledge network. It also allows for more comprehensive utilization of the full amount of data when updating the model and agent, ensuring the efficiency of the model and agent update.

[0084] In an exemplary embodiment, updating the profile building model and the test agent based on the updated test knowledge graph includes:

[0085] The updated test knowledge graph is input into a pre-defined graph neural network, and updated tester data is extracted based on the graph neural network. The updated tester data includes user feedback data, psychological profile data, software testing behavior data, and data after the test plan is updated.

[0086] The profile building model and the test agent are updated based on the updated data from the testers.

[0087] In this embodiment, the updated test knowledge graph is input into a preset graph neural network (such as a GNN). Based on this GNN, a series of high-quality tester update data can be inferred, such as "testers with high conscientiousness and moderate neuroticism are most efficient at discovering security defects when using test strategy S to test financial software." Thus, the profile building model and the test agent are updated based on the tester update data provided by the GNN. In practical applications, the model and agent can be updated based solely on the tester update data provided by the GNN, or based on data concatenated from the tester update data and the original data.

[0088] In practical applications, the above-mentioned methods for updating models and agents include, but are not limited to, incremental learning, transfer learning, and so on.

[0089] Through the embodiments of this application, by introducing a test knowledge graph, the deep relationships and features between the profile building model and the test agent's learning data can be realized, thereby improving the updating effect and intelligence level of the profile building model and the test agent, realizing the effective organization and utilization of knowledge, and enhancing the system's self-evolution capability.

[0090] In one exemplary embodiment, such as Figure 4 As shown, based on the software testing strategy and psychological profile data, a corresponding testing plan for the target tester is generated, including:

[0091] Step S410: Obtain usage feedback data for the software under test.

[0092] In this embodiment of the application, the software to be tested is the software that the target tester needs to test. The usage feedback data is the feedback data provided by the end user. The end user is the person who uses the software. The usage feedback data is the feedback data generated by the end user during actual use that can reflect the quality, performance, ease of use and other issues of the software. In practical applications, it can be obtained through data acquisition components embedded in the software, software usage logs and questionnaires.

[0093] Step S420: Based on user feedback data, software testing strategies, and psychological profile data, generate a corresponding testing plan for the target testers.

[0094] In this embodiment, a test plan is generated based on the aforementioned user feedback data, software testing strategies, and psychological profile data. This means that the opinions of users of the software under test, the psychological profile of the target testers, and the testing strategies are combined to generate a corresponding test plan. In actual operation, the test plan can be generated through a fully trained neural network.

[0095] Through the embodiments of this application, the characteristics and psychology of testers, as well as the opinions of software users, are comprehensively considered to ensure that the final generated test plan not only matches the strengths of testers, but also improves the practicality of the test plan, and enhances the coverage and defect detection rate of the test plan.

[0096] In one exemplary embodiment, such as Figure 5 As shown, before generating the corresponding test plan for the target tester based on the software testing strategy and psychological profile data, the method also includes:

[0097] Step S510: Obtain multimodal interaction data of the target tester; the multimodal interaction data includes visual data, voice data and behavioral data of the target tester obtained during the execution of historical test plans.

[0098] In this embodiment, the aforementioned multimodal interaction data includes, but is not limited to, the visual data of the target tester (such as detecting facial expressions such as frowning), voice data (receiving voice commands from the tester, collecting the voice signals of the tester, or detecting the tester uttering interjections), and text information (which can be implemented through interactive pop-ups), etc., to achieve efficient and natural interaction with the target tester.

[0099] Step S520: Perform multimodal fusion on the multimodal interaction data to obtain the sentiment recognition results and intent recognition results of the target testers regarding the software testing strategy.

[0100] In this embodiment, a multimodal Transformer model can be used to achieve the fusion processing of visual, speech, and text information, improving the accuracy and completeness of information understanding. In one embodiment, the fused multimodal information can be input into a deep learning model to obtain the emotional state of the test subject. In another embodiment, the emotional state of the test subject can be identified and classified in real time according to a preset emotion dictionary. In summary, the emotion recognition result of the test subject can be obtained.

[0101] Similarly, the intent recognition result of the target tester can also be obtained based on the fused multimodal information. The multimodal information can be directly analyzed. For example, the analysis of the voice signal can reveal that the tester is dissatisfied with a certain part of the testing strategy. Or, if the tester's interface is detected to be stuck on a certain module for a long time without producing any effective action, it may indicate that the tester has a negative opinion of the module (such as unclear operation guidance or unfamiliarity with the recommended testing tools), and so on.

[0102] Step S530: Adjust the software testing strategy based on the emotion recognition results and the intent recognition results.

[0103] The software testing strategy can be adjusted based on the analyzed sentiment recognition and intent recognition results, thereby facilitating the generation of corresponding test plans based on a more accurate testing strategy. Understandably, in one embodiment, when generating an initial testing strategy, the initial testing strategy can be shown to the testers, and sentiment recognition and intent recognition results can be extracted based on the testers' multimodal interaction data, thereby adjusting the initial software testing strategy based on these results. In another embodiment, when generating the initial testing strategy, it can be that the initial testing strategy is not directly shown, but rather the multimodal interaction data saved by the testers during the execution of historical test tasks can be automatically retrieved, and the testing strategy can be adjusted directly based on this multimodal interaction data.

[0104] This application introduces multimodal interaction data, and based on the analysis of the multimodal interaction data, a testing strategy that is more suitable for the target tester can be obtained.

[0105] In one exemplary embodiment, such as Figure 6 As shown, the process of building the test agent includes:

[0106] Step S610: Obtain sample psychological data, sample behavioral data, and sample emotional data of various sample personnel.

[0107] In this embodiment, sample psychological data, sample behavioral data, and sample emotional data of preset sample personnel are obtained. The sample personnel can be a number of testers participating in the model training phase. It is understood that different testers represent different psychological characteristics, test behaviors, and emotions during the test. Using data from multiple sample personnel can ensure that the trained agent has broad applicability.

[0108] Step S620: Input the sample psychological data, sample behavioral data, and sample emotional data of each sample person into the initial test agent, train the initial test agent, and calculate the value of the loss function corresponding to the initial test agent based on the training results of the initial test agent.

[0109] In this embodiment, the core of the initial test agent is a deep learning network (such as a multilayer perceptron). The initial test agent is trained based on the sample psychological data, sample behavioral data, and sample emotional data of each sample person. This process typically employs mainstream reinforcement learning algorithms. The loss function corresponding to the current initial test agent is calculated based on its training results.

[0110] Step S630: If the value of the loss function determines that the test agent has reached the convergence condition, the trained test agent is obtained.

[0111] If the value of the loss function no longer decreases significantly after multiple iterations, but fluctuates within a very small range, it indicates that the test agent has reached the convergence condition, and the trained test agent is obtained.

[0112] In one exemplary embodiment, such as Figure 7 As shown, the initial test agent includes a deep learning model and a simulated test environment established based on the test requirements of the software under test; the initial test agent is trained, and the value of the loss function corresponding to the initial test agent is calculated based on the training results, including:

[0113] Step S710: Input the sample psychological data, sample behavioral data, and sample emotional data of each sample person into the deep learning model, and train the deep learning model to obtain the sample testing strategy.

[0114] The initial test agent includes a deep learning model and a simulated test environment corresponding to the software under test. The deep learning model may include, but is not limited to, architectures such as multilayer perceptron, convolutional neural network, or Transformer. The simulated environment is a containerized test environment, which can be built based on Docker and Kubernetes, supporting rapid deployment and configuration of different versions of software to simulate real test scenarios.

[0115] The sample psychological data, sample behavioral data, and sample emotional data of each sample person are input into the deep learning model to obtain the corresponding sample testing strategy. That is, the testing strategy is the outline of the specific testing plan generated by the deep learning model based on a certain sample person.

[0116] Step S720: Input the sample test strategy into the simulation test environment for execution to obtain the strategy execution result.

[0117] In this embodiment of the application, the sample testing strategy is input into a simulated testing environment, and the sample testing strategy is executed in the simulated testing environment corresponding to the software under test, and the strategy execution results are generated (including but not limited to code coverage, defect discovery, execution efficiency, etc.).

[0118] Step S730: Calculate the reward value corresponding to the strategy execution result based on the preset evaluation indicators.

[0119] In this embodiment, a reward value corresponding to the testing strategy is calculated based on preset evaluation indicators and the execution results of the aforementioned strategy. These evaluation indicators can be set by relevant technical personnel according to actual needs, such as multi-dimensional indicators like user satisfaction, test coverage, and defect detection rate. The reward value reflects a comprehensive score of the strategy execution results, quantifying the quality of the strategy execution outcome.

[0120] Step S740: Calculate the value of the loss function based on the reward value and the reinforcement algorithm.

[0121] In this embodiment, the sample test strategies, policy execution results, and reward values ​​obtained above can be stored. When training the deep learning model, a batch of sample test strategies, policy execution results, and reward value data can be obtained from the stored data. This batch of data is used as the training set, and the corresponding loss function value is calculated using the reward value in this batch of data according to the preset reinforcement algorithm (e.g., Proximal Policy Optimization, PPO).

[0122] This application also provides a method for determining a software test plan, which is applied to a software test plan generation system, such as... Figure 8 The diagram shown is a flowchart illustrating a method for determining a software testing scheme in one embodiment.

[0123] Step S810: Load the system parameters of the software test plan generation system, and initialize the model and configuration.

[0124] Step S820: Collect software testing behavior data and emotional feedback data of the target testers. In practical applications, after collecting the software testing behavior data and emotional feedback data of the target testers, the data can be cleaned, normalized, and feature extracted to construct a high-quality dataset.

[0125] Step S830: Generate psychological profile data corresponding to the target tester based on software testing behavior data and emotional feedback data. In practical applications, the software testing behavior data and emotional feedback data can be input into a well-trained profile building model to generate the corresponding psychological profile data. It is understood that when building the profile building model, methods such as transfer learning and adaptive learning rates can be considered to achieve efficient model building and parameter optimization. The preferred profile building model is a deep learning model based on an attention mechanism (such as BERT, GPT, etc.). It should be noted that in subsequent calculations, behavioral pattern change detection and anomaly detection of the target tester can be introduced to achieve dynamic updates of the psychological profile, ensuring its accuracy and timeliness.

[0126] Step S840: Generate a corresponding testing strategy based on psychological profile data, software testing behavior data, and emotional feedback data. The psychological profile data, software testing behavior data, and emotional feedback data can be input into a preset testing agent to obtain the corresponding software testing strategy.

[0127] Step S850: Adjust the software testing strategy based on the collected multimodal interaction data.

[0128] Step S860: Obtain user feedback data for the software under test, and generate a corresponding test plan based on the user feedback data, the adjusted software testing strategy, and the user profile data. The user feedback data consists of suggestions provided by users of the software under test.

[0129] Step S870: Obtain user feedback from the target testers when they conduct software testing according to the above test plan, and update the profile building model and test agent based on the user feedback.

[0130] In summary, this application enables the use of high-quality testing solutions through deep learning frameworks, reducing unnecessary testing steps and improving testing efficiency. Furthermore, this application generates personalized testing solutions tailored to the target testers, reducing resource waste during tester task execution and lowering testing costs. The high compatibility of the testing solution with testers also increases tester satisfaction.

[0131] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0132] Based on the same inventive concept, this application also provides a software test plan determination apparatus for implementing the software test plan determination method described above. The solution provided by this apparatus is similar to the implementation scheme described in the above method; therefore, the specific limitations in one or more software test plan determination apparatus embodiments provided below can be found in the limitations of the software test plan determination method described above, and will not be repeated here.

[0133] In one exemplary embodiment, such as Figure 9 As shown, a software testing scheme determination device is provided, comprising: an acquisition module 91, a calculation module 92, and a generation module 93, wherein:

[0134] The acquisition module 91 is used to respond to the test plan generation instruction of the target tester for the software to be tested. It analyzes the software testing behavior data and emotional feedback data of the target tester through a preset profile construction model to generate psychological profile data of the target tester.

[0135] The calculation module 92 is used to input software testing behavior data, emotional feedback data, and psychological profile data into a preset testing agent to generate a software testing strategy for the target tester.

[0136] The generation module 93 is used to generate test plans corresponding to target testers based on software testing strategies and psychological profile data.

[0137] The aforementioned software testing scheme determines that each module in the device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of a computer device in hardware form or independent of it, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0138] In one exemplary embodiment, the software test scheme determination device further includes a second feedback module and a second update module;

[0139] The second feedback module is used to obtain user feedback data obtained by the target testers when they test the software under test according to the test plan, based on the software testing strategy and psychological profile data to generate the corresponding test plan for the target testers.

[0140] The second update module is used to update the profile building model and the testing agent based on user feedback data, test plans, psychological profile data, and software testing behavior data.

[0141] In one exemplary embodiment, the software test scheme determination device further includes a third determination module and a third update module;

[0142] The third module is used to update the test knowledge graph based on user feedback data, test plans, psychological profile data, and software testing behavior data. The test knowledge graph is built based on the target testers' historical test plans and historical user feedback data of the execution of historical test plans, the target testers' historical psychological profile data, and historical software testing behavior data.

[0143] The third update module is used to update the profile building model and the test agent based on the updated test knowledge graph.

[0144] In one exemplary embodiment, the software test scheme determination apparatus further includes a fourth calculation module and a fourth generation module;

[0145] The fourth calculation module is used to input the updated test knowledge graph into a preset graph neural network, and extract the updated data of testers based on the graph neural network; the updated data of testers includes user feedback data, psychological profile data, software testing behavior data, and data after the test plan is updated;

[0146] The fourth generation module is used to update the profile building model and the test agent based on the updated data from the testers.

[0147] In one exemplary embodiment, the software test scheme determination apparatus further includes a fifth acquisition module and a fifth generation module;

[0148] The fifth acquisition module is used to acquire user feedback data of the software under test;

[0149] The fifth generation module is used to generate corresponding test plans for target testers based on user feedback data, software testing strategies, and psychological profile data.

[0150] In an exemplary embodiment, the software test scheme determination device further includes a sixth acquisition module, a sixth calculation module, and a sixth generation module;

[0151] The sixth acquisition module is used to acquire multimodal interaction data of the target testers; the multimodal interaction data includes visual data, voice data and behavioral data of the target testers acquired during the execution of historical test plans;

[0152] The sixth calculation module is used to perform multimodal fusion on multimodal interaction data to obtain the emotion recognition results and intent recognition results of the target testers on the testing strategy;

[0153] The sixth generation module is used to adjust the software testing strategy based on the emotion recognition results and intent recognition results.

[0154] In an exemplary embodiment, the software test scheme determination device further includes a seventh acquisition module, a seventh calculation module, and a seventh determination module;

[0155] The seventh acquisition module is used to acquire sample psychological data, sample behavioral data, and sample emotional data of various sample personnel;

[0156] The seventh calculation module is used to input the sample psychological data, sample behavioral data, and sample emotional data of each sample person into the initial test agent, train the initial test agent, and calculate the value of the loss function corresponding to the initial test agent based on the training results of the initial test agent.

[0157] The seventh determination module is used to obtain the trained test agent when the value of the loss function determines that the deep learning model has reached the convergence condition.

[0158] In an exemplary embodiment, the software test scheme determination device further includes an eighth acquisition module, an eighth calculation module, an eighth generation module, and an eighth determination module;

[0159] The eighth acquisition module is used to input the sample psychological data, sample behavioral data, and sample emotional data of each sample person into the deep learning model, and train the deep learning model to obtain the sample testing strategy.

[0160] The eighth calculation module is used to input the sample test strategy into the simulation test environment for execution and obtain the strategy execution result;

[0161] The eighth generation module is used to calculate the corresponding reward value of the strategy execution result based on the preset evaluation indicators;

[0162] The eighth determination module is used to calculate the value of the loss function based on the reward value and the reinforcement algorithm.

[0163] In one exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement any of the software test scheme determination methods described above.

[0164] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements any of the software test scheme determination methods described above.

[0165] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements any of the software test scheme determination methods described above.

[0166] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0167] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0168] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for determining a software testing plan, characterized in that, The method, applied to a software test plan generation system, includes: In response to the test plan generation instruction of the target tester for the software to be tested, the software testing behavior data and emotional feedback data of the target tester are analyzed through a preset profile building model to generate psychological profile data of the target tester. The software testing behavior data, the emotional feedback data, and the psychological profile data are input into a preset testing agent to generate the software testing strategy for the target tester. Based on the software testing strategy and the psychological profile data, a test plan is generated for the target tester.

2. The method according to claim 1, characterized in that, The method further includes: In the case of generating a corresponding test plan for the target tester based on the software testing strategy and the psychological profile data, user feedback data obtained by the target tester testing the software under test according to the test plan is acquired. Based on the user feedback data, the test plan, the psychological profile data, and the software test behavior data, the profile construction model and the test agent are updated.

3. The method according to claim 2, characterized in that, The updating of the profile construction model and the test agent includes: The test knowledge graph is updated based on the user feedback data, the test plan, the psychological profile data, and the software testing behavior data; the test knowledge graph is built based on the target tester's historical test plans and the historical user feedback data of executing the historical test plans, the target tester's historical psychological profile data, and historical software testing behavior data; The profile building model and the test agent are updated based on the updated test knowledge graph.

4. The method according to claim 3, characterized in that, The step of updating the profile construction model and the test agent based on the updated test knowledge graph includes: The updated test knowledge graph is input into a preset graph neural network, and tester update data is extracted based on the graph neural network; the tester update data includes user feedback data, psychological profile data, software testing behavior data, and data after the test plan is updated; The profile building model and the test agent are updated based on the updated data from the testers.

5. The method according to any one of claims 1 to 4, characterized in that, The step of generating a corresponding test plan for the target tester based on the software testing strategy and the psychological profile data includes: Obtain usage feedback data for the software under test; Based on the user feedback data, the software testing strategy, and the psychological profile data, a corresponding testing plan is generated for the target tester.

6. The method according to any one of claims 1 to 4, characterized in that, Before generating the corresponding test plan for the target tester based on the software testing strategy and the psychological profile data, the method further includes: Acquire the multimodal interaction data of the target tester; the multimodal interaction data includes the visual data, voice data and behavioral data of the target tester obtained during the execution of historical test plans; Multimodal fusion is performed on the multimodal interaction data to obtain the emotion recognition result and intent recognition result of the target tester regarding the software testing strategy; The software testing strategy is adjusted based on the emotion recognition results and the intent recognition results.

7. The method according to any one of claims 1 to 4, characterized in that, The construction process of the test agent includes: Acquire sample psychological data, sample behavioral data, and sample emotional data from various sample groups; The sample psychological data, sample behavioral data, and sample emotional data of each of the sample personnel are input into the initial test agent to train the initial test agent, and the value of the loss function corresponding to the initial test agent is calculated based on the training result of the initial test agent. Once the value of the loss function determines that the test agent has reached the convergence condition, the trained test agent is obtained.

8. The method according to claim 7, characterized in that, The initial test agent includes a deep learning model and a simulated test environment established based on the test requirements of the software under test; training the initial test agent and calculating the value of the loss function corresponding to the initial test agent based on the training results of the initial test agent includes: The sample psychological data, sample behavioral data, and sample emotional data of each of the sample personnel are input into the deep learning model, and the deep learning model is trained to obtain the sample testing strategy. The sample testing strategy is input into the simulated testing environment and executed to obtain the strategy execution result; The reward value corresponding to the strategy execution result is calculated based on the preset evaluation index; The value of the loss function is calculated based on the reward value and the reinforcement algorithm.

9. A software testing scheme determination device, characterized in that, The device includes: The acquisition module is used to respond to the test plan generation instruction of the target tester for the software to be tested, and analyze the software testing behavior data and emotional feedback data of the target tester through a preset profile construction model to generate the psychological profile data of the target tester. The calculation module is used to input the software testing behavior data, the emotional feedback data, and the psychological profile data into a preset testing agent to generate the software testing strategy for the target tester. The generation module is used to generate a test plan corresponding to the target tester based on the software testing strategy and the psychological profile data.

10. A computer device, comprising a memory and a processor, characterized in that, The processor is configured to form a software test scheme generation system as described in any one of claims 1 to 8, and to implement the steps of the method as described in any one of claims 1 to 8.

Images