Scheduling strategy automatic verification device for saas travel platform

Abstract

The invention discloses a saas travel platform-oriented scheduling strategy automatic verification device, and relates to the technical field of travel platform scheduling management. Comprising a scheduling strategy parameter analysis and matching module, a regression test execution and verification module, a test case optimization and selection module, a dynamic test data adjustment module and a prediction and strategy generation module, and automatically matching with an existing strategy library to generate a test case set. Through automatic generation and matching of the test cases, the efficiency and the automation level of the regression test of the SaaS travel platform are remarkably improved, manual intervention is reduced, and the test coverage rate is increased. Along with the richness of the strategy library, unmatched strategies are reduced, the test cost is reduced, and the flexibility and competitiveness of the platform for coping with multi-tenant and strategy change are improved at the same time.

Description

Technical Field

[0001] The present invention relates to the technical field of travel platform scheduling management, and in particular to a scheduling strategy automatic verification device for a SaaS travel platform. Background Art

[0002] With the rapid development of the shared travel and online car-hailing industries, the application of the SaaS (Software as a Service) model in travel platforms has become more and more extensive. SaaS travel platforms provide services to multiple tenants, and each tenant can customize scheduling strategies according to their own needs. As the core capability of the platform, scheduling strategies determine how to efficiently allocate vehicle and driver resources to meet user needs, directly affecting the tenant's operational efficiency and user experience. Whenever new tenants join or existing tenants adjust their scheduling strategies, it becomes crucial to verify the effectiveness and accuracy of these strategies. However, since scheduling strategies involve complex parameters (such as distance, pick-up time, traffic conditions, etc.), manual verification is prone to errors and inefficient, making it difficult to cope with the large-scale and frequently iterative needs of travel platforms.

[0003] The existing technology has the following deficiencies:

[0004] Currently, traditional scheduling policy verification methods primarily rely on manual testing or regression testing. Testers need to manually prepare test data based on each tenant's specific configuration and verify the effectiveness of the policy item by item. While automated regression testing has improved testing efficiency to a certain extent, test case coverage remains low. Furthermore, each time a new tenant is added or the tenant configuration changes, test cases must be redesigned and rerun, which is time-consuming and cumbersome. More importantly, existing automated testing methods struggle to flexibly adapt to the configuration differences between different tenants, resulting in incomplete and inaccurate test coverage and unable to meet the dynamic needs of large-scale SaaS platforms. Consequently, traditional verification methods suffer from low efficiency, insufficient coverage, and reliance on manual intervention.

[0005] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not form the prior art that is already known to a person of ordinary skill in the art. Summary of the Invention

[0006] The purpose of the present invention is to provide a scheduling strategy automatic verification device for SaaS travel platforms, which significantly improves the efficiency and automation level of SaaS travel platform scheduling strategy regression testing by automatically generating and matching test cases. When a new tenant joins or the policy changes, the system can automatically match and trigger regression testing, reducing manual intervention and improving test coverage. As the policy library becomes richer, the number of unmatched policies gradually decreases, further reducing testing costs, and enhancing the platform's flexibility in dealing with multi-tenants and policy changes, improving the platform's adaptability and scalability, making it more competitive in a dynamic market environment, so as to solve the problems in the above-mentioned background technology.

[0007] To achieve the above objectives, the present invention provides the following technical solution: an automatic verification device for scheduling strategies for SaaS travel platforms, comprising a scheduling strategy parameter analysis and matching module, a regression test execution and verification module, a test case optimization and selection module, a dynamic test data adjustment module, and a prediction and strategy generation module:

[0008] The scheduling policy parameter analysis and matching module automatically analyzes the scheduling policy parameters based on the tenant's scheduling policy configuration information, automatically matches them with the existing policy library, and generates a set of test cases;

[0009] The regression test execution and verification module automatically executes the regression test of the scheduling strategy through an algorithm based on the generated test case set, and determines whether the strategy is effective based on the test results;

[0010] The test case optimization and selection module uses a matching algorithm to calculate the execution efficiency of each test case and optimize the selection of test cases to ensure high coverage and low computational complexity;

[0011] Dynamic test data adjustment module: Based on the regression test results, the algorithm module automatically adjusts the test data to cope with changes and additions in tenant scheduling strategies, and optimizes subsequent test case generation and execution strategies;

[0012] The prediction and strategy generation module, based on machine learning algorithms, analyzes historical data of test results, predicts adjustment trends of tenant scheduling strategies, and automatically generates new test strategies to ensure high test coverage and accuracy in subsequent versions.

[0013] Preferably, the scheduling policy parameters are automatically analyzed based on the tenant's scheduling policy configuration information, and automatically matched with the existing policy library to generate a test case set. The specific steps are as follows:

[0014] First, receive the tenant's scheduling policy configuration information from the SaaS travel platform;

[0015] All relevant scheduling parameters will be extracted from the tenant's scheduling policy;

[0016] Once the core parameters of the scheduling policy are extracted and standardized, the parameters are matched with the existing scheduling policy library;

[0017] Based on the matching results with the strategy library, a test case set is automatically generated.

[0018] Preferably, based on the generated test case set, the specific steps of automatically executing the regression test of the scheduling strategy through the algorithm and judging whether the strategy is effective based on the test results are as follows:

[0019] After generating the test case set, prepare the regression testing environment;

[0020] After the test environment is ready, start automatically executing regression tests;

[0021] During the regression test execution process, the execution status of each test case will be monitored in real time and dynamic feedback will be provided;

[0022] After the regression test is completed and the results are analyzed, the effectiveness of the scheduling strategy is judged based on the preset evaluation criteria.

[0023] Preferably, a matching algorithm is used to calculate the execution efficiency of each test case and optimize the selection of test cases to ensure high coverage and low computational complexity. The specific steps are as follows:

[0024] During the regression test execution process, the execution efficiency of each generated test case is evaluated;

[0025] After evaluating the execution efficiency of each test case, the matching algorithm is used to screen and optimize the test cases;

[0026] After optimization and screening, the test case selection strategy is dynamically adjusted to ensure that the computational complexity of the test remains within a normal range while ensuring high coverage;

[0027] Finally, feedback adjustments are made based on the execution results of the regression test and the implementation effect of the optimization strategy.

[0028] Preferably, after the regression test is completed, the execution results of each test case are evaluated, and the "effectiveness score" and deviation of the tenant scheduling strategy are calculated. The regression test results will score each strategy. Based on the test results, the algorithm module calculates the "deviation" of the strategy and determines whether the test data needs to be adjusted based on the deviation. The deviation calculation formula is as follows:

[0029]

[0030] , where D iis the deviation of the i-th test case, N is the total number of parameters in the test case, T ij is the test value of the jth parameter of the i-th test case, R ij is the expected result value of the jth parameter of the i-th test case, |T ij -R ij | is the error value of the jth parameter of the i-th test case;

[0031] Based on the deviation results, the algorithm module automatically adjusts the parameter range of the test data to adapt to changes and additions to tenant scheduling policies. The parameter range adjustment formula is as follows:

[0032] T ij ′=T ij +α·D i ·sign(T ij -R ij )

[0033] , where α is the adjustment coefficient, which controls the adjustment range, sign(T ij -R ij ) is the sign function used to determine the adjustment direction, T ij ′ is the adjusted test data value;

[0034] After completing the test data adjustment, regenerate the optimized test case set based on the new parameter range. The optimized test case generation formula is as follows:

[0035]

[0036] , where C i is the comprehensive score of the optimized i-th test case, W ij is the weight of the jth parameter of the i-th test case;

[0037] Finally, based on the newly generated optimized test cases, adjust the subsequent test execution strategy to ensure that the new test cases can be executed efficiently with reasonable computing resources. Based on the optimized test case set and the execution efficiency of each case, calculate resource allocation and determine the execution order using the following formula:

[0038]

[0039] , where A i is the computing resource allocation ratio of the i-th test case, R max is the maximum computing resource available to the system, and M is the total number of test cases.

[0040] Preferably, based on machine learning algorithms, historical data of test results is analyzed to predict the adjustment trend of tenant scheduling strategies, and new test strategies are automatically generated to ensure high test coverage and accuracy in subsequent versions. The specific steps are as follows:

[0041] First, we collect and analyze the data of historical regression test results. By analyzing these historical data, we extract the key feature parameters and calculate the weighted value of each feature based on the historical data set. The calculation expression is as follows:

[0042]

[0043] , where W p is feature F p The weight value in the data set, F p is the pth feature in the historical data, D hist is the historical test dataset, n is the total number of features;

[0044] Based on historical data analysis and feature extraction, a machine learning algorithm is used to train a trend prediction model. The model uses the features extracted from historical data and the weighted value W. i To construct a training set, we can predict the changes of future strategies. The calculation expression of the adjustment prediction value of the future scheduling strategy is as follows:

[0045] P adj =α·X train +β·W p

[0046] , where P adj is the adjusted forecast value of the future scheduling strategy, X train is the input feature set used for training, α is the regression coefficient in the regression model, which indicates the influence of the input feature on the prediction result, and β is the feature weighting coefficient;

[0047] Adjusted prediction value P based on future scheduling strategy adj , a new test strategy will be generated. According to the prediction results, a new test case set will be generated, and it will be ensured that the case effectively tests the adjusted scheduling strategy. The new test case set generation formula is as follows:

[0048]

[0049] , where T new is the newly generated test strategy, m is the total number of generated test cases, f(P adj ,F p ) is based on the predicted trend P adj and feature F p Compute the generator function for each test case;

[0050] Finally, regression testing is performed based on the generated new test strategy and test case set. If the test results show that the generated test strategy does not meet expectations in terms of high coverage and accuracy, the prediction model will be further optimized based on the results of the regression test, and the generated test strategy will be adjusted. The optimized measurement expression is as follows:

[0051]

[0052] , where T final is the final optimized test strategy, E test is the evaluation value of the regression test result, E max It is the preset optimal coverage and accuracy target value.

[0053] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0054] The present invention greatly improves the efficiency of the SaaS travel platform's scheduling strategy regression test through an automated test case generation and matching mechanism. When a new tenant joins or an existing tenant's scheduling strategy changes, the system can automatically match the existing scheduling strategy test set and quickly trigger the corresponding regression test. The system automatically identifies and applies matching test cases based on the tenant's scheduling strategy, avoiding the process of manual intervention and ensuring that the platform can verify the effectiveness of the new strategy in a timely and comprehensive manner after each policy change. With the continuous enrichment of the scheduling strategy library, the proportion of unmatched strategies has gradually decreased, the degree of test automation has continued to improve, and the test efficiency has been significantly enhanced. The system can cover a wider range of test scenarios, reduce the workload of repeated manual testing, greatly improve the testing efficiency of the travel platform, and shorten the cycle from strategy adjustment to online verification.

[0055] The present invention not only reduces the cost of manual intervention through automated testing, but also significantly improves the flexibility of the platform in dealing with different tenants and policy changes. With the continuous improvement of the scheduling policy library, the system can quickly adapt to new tenant configurations or policy changes, and intelligently match appropriate test cases for verification, avoiding the tedious work of manually designing new test cases for each change. This automatic matching mechanism reduces the need for manual testing and data preparation, thereby reducing the overall cost of testing. At the same time, when faced with a complex multi-tenant environment, the system can flexibly generate relevant test cases based on the personalized needs of each tenant and efficiently verify their policies. This flexibility ensures that the platform can quickly respond to market changes and tenant needs, enhances the adaptability and scalability of the SaaS platform, and makes it more competitive in the dynamic travel market. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction to the drawings required for use in the embodiments will be given below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0057] Figure 1 This is a module schematic diagram of an automatic verification device for scheduling strategies for a SaaS travel platform according to the present invention. DETAILED DESCRIPTION

[0058] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that the description of this disclosure will be thorough and complete and will fully convey the concepts of the example embodiments to those skilled in the art.

[0059] The present invention provides Figure 1 The automatic verification device for scheduling strategies for SaaS travel platforms shown in the figure includes a scheduling strategy parameter analysis and matching module, a regression test execution and verification module, a test case optimization and selection module, a dynamic test data adjustment module, and a prediction and strategy generation module:

[0060] The scheduling policy parameter analysis and matching module automatically analyzes the scheduling policy parameters based on the tenant's scheduling policy configuration information, automatically matches them with the existing policy library, and generates a set of test cases;

[0061] Based on the tenant's scheduling policy configuration information, the scheduling policy parameters are automatically analyzed and automatically matched with the existing policy library. The specific steps for generating a test case set are as follows:

[0062] First, receive the tenant's scheduling policy configuration information from the SaaS travel platform;

[0063] This information may include, but is not limited to, passenger pickup time, distance, vehicle type, traffic conditions, route selection, and other parameters. Tenants may customize these parameters based on their operating environment, geographic location, or specific business needs. Upon receiving the configuration information, the system automatically structures it and extracts key information for subsequent analysis and matching. The core of this step is the standardization and classification of information, ensuring that each parameter is accurately mapped to the corresponding position in the existing policy library for subsequent matching operations.

[0064] All relevant scheduling parameters will be extracted from the tenant's scheduling policy;

[0065] These parameters may be numerical, such as time, distance, and cost, or categorical, such as driver level and passenger demand type. During the extraction process, the system will classify and standardize these parameters based on preset rules or algorithms. For example, it will unify time limits in different regions into a standard format, or convert categorical data such as "express routes" and "regular routes" into actionable standard values. Through such parameter analysis, the system can fully understand the configuration of tenant scheduling strategies, thereby identifying which key factors affect scheduling decisions, which parameters are adjustable, and which are the core content that the platform needs to verify.

[0066] Once the core parameters of the scheduling policy are extracted and standardized, the parameters are matched with the existing scheduling policy library;

[0067] The policy library is a database that stores various scheduling policy templates and historical verification data. It contains scheduling policy examples for different tenants and different scenarios. Using algorithms, the system compares and matches newly received tenant policies with similar policies in the library to find the most similar policy template. This process can be accomplished using pattern recognition, similarity calculation, or machine learning algorithms to ensure efficient and accurate matching. For example, if a tenant configures a new pick-up time limit, the system will compare it with previous policies for similar scenarios and automatically generate a corresponding verification case.

[0068] Automatically generate a test case set based on the matching results with the strategy library;

[0069] These test cases represent the possible execution of tenant scheduling strategies in different scenarios, covering the possibility of various scheduling decisions. For example, if a tenant's scheduling strategy involves specific rules for "peak hours", the system will generate a test case to simulate the scheduling needs during peak hours, and compare it with the existing strategy to evaluate its rationality. The generated test cases will be stored in the system, awaiting automatic execution in the regression testing phase. In this step, the system must not only consider routine situations, but also predict and generate special situations, such as sudden traffic changes, sudden surges in orders, etc. These generated test cases will provide comprehensive data support for subsequent scheduling strategy verification, ensuring the comprehensiveness and accuracy of the verification.

[0070] The regression test execution and verification module automatically executes the regression test of the scheduling strategy through an algorithm based on the generated test case set, and determines whether the strategy is effective based on the test results;

[0071] Based on the generated test case set, the algorithm is used to automatically execute the regression test of the scheduling strategy, and the specific steps to determine whether the strategy is effective based on the test results are as follows:

[0072] After generating the test case set, prepare the regression testing environment;

[0073] This step involves initializing the system state, loading test data, setting relevant test parameters, and loading the test case collection generated in the previous step. Because regression testing of scheduling policies involves multiple parameters (such as time, location, and pickup requirements), test case collections often contain different scenarios and data combinations. Therefore, before execution, the system automatically allocates computing resources based on the complexity of the test cases to ensure efficient testing. The regression testing environment simulates the actual platform's operating state, ensuring that every parameter during the test is verified under conditions close to those in the real world, thus avoiding biased test results.

[0074] After the test environment is ready, start automatically executing regression tests;

[0075] During the execution process, the actual execution of the scheduling strategy will be simulated for each test case, including parameters such as pick-up time, driver allocation, and order priority. Each test case will compare the scheduling strategy configured by the tenant with the policy template in the existing policy library to evaluate whether the strategy has achieved the expected results. For example, the system will check whether the order allocation is completed within the specified time, whether the scheduling priority is adjusted as required during peak traffic periods, etc. Through this process, the system will automatically record the execution results of each test case and automatically determine whether the scheduling strategy is effective based on preset standards (such as whether the pick-up time meets the requirements, whether the resource allocation is reasonable, etc.).

[0076] During the regression test execution process, the execution status of each test case will be monitored in real time and dynamic feedback will be provided;

[0077] Whenever a test case is executed, the system automatically analyzes the execution results based on pre-set test standards (such as time thresholds, resource allocation accuracy, etc.). If a strategy fails to achieve the expected goal, the system will mark it as a failure and record the reasons for the failure in detail (for example, the pick-up time exceeds the maximum tolerance value, resource allocation is uneven, etc.). At the same time, based on the execution results of multiple test cases, the system will analyze whether the performance of the strategy is consistent in different scenarios and whether there is potential for optimization. If certain strategies perform poorly in specific scenarios, the system will automatically generate improvement suggestions by analyzing the test data for tenants to adjust and optimize.

[0078] After the regression test is completed and the results are analyzed, the effectiveness of the scheduling strategy is judged based on the preset evaluation criteria;

[0079] If the policy performs well in most test scenarios and meets the tenant's operational needs, the system will determine that the policy is valid and generate a corresponding test report and execution log, which will be fed back to the tenant. Conversely, if the policy fails to meet the expected goals during regression testing, the system will mark it as invalid and provide a detailed failure report. This report will include the specific circumstances of the test failure, such as delayed pick-up time and unbalanced resource allocation, and include possible improvement suggestions. Ultimately, the tenant can further optimize the scheduling policy based on the information provided by the system to ensure that it will pass the next regression test.

[0080] The test case optimization and selection module uses a matching algorithm to calculate the execution efficiency of each test case and optimize the selection of test cases to ensure high coverage and low computational complexity;

[0081] The specific steps for using a matching algorithm to calculate the execution efficiency of each test case and optimize the selection of test cases to ensure high coverage and low computational complexity are as follows:

[0082] During the regression test execution process, the execution efficiency of each generated test case is evaluated;

[0083] An evaluation model is constructed based on multiple dimensions, including the operational complexity, execution time, and required computing resources involved in each test case. This model considers the resource consumption of each test case during actual execution, including metrics such as the system's CPU load, memory usage, and I / O operation frequency. By analyzing these factors, the system can assign an execution efficiency score to each test case, facilitating appropriate choices during subsequent optimization. The core purpose of this step is to assign an "efficiency" label to each test case, ensuring that the testing process prioritizes resources for executing those test cases that can efficiently evaluate the effectiveness of the strategy.

[0084] After evaluating the execution efficiency of each test case, the matching algorithm is used to screen and optimize the test cases;

[0085] The algorithm's goal is to prioritize test cases that demonstrate high execution efficiency and cover a wider range of test scenarios. To achieve this, the system calculates the "coverage" of each test case—that is, the key scheduling policy parameter combinations covered by the test case. The system compares the coverage of different test cases and selects those that cover more important scenarios, while avoiding repeated testing of the same scenario. The matching algorithm leverages the similarities between test cases, prioritizing those that effectively verify policies in multiple scenarios to ensure high coverage. This algorithm effectively reduces redundant testing and optimizes the combination of test cases, saving computing resources and time.

[0086] After optimization and screening, the test case selection strategy is dynamically adjusted to ensure that the computational complexity of the test remains within a normal range while ensuring high coverage;

[0087] To achieve this goal, the system adjusts the order of test case execution based on the real-time performance of regression testing. Specifically, the system prioritizes test cases known to be highly efficient and with broad coverage, while scheduling inefficient or resource-intensive test cases for later in the testing process. This strategy effectively focuses computing resources on efficient test cases while reducing the frequency of inefficient ones, ensuring a smooth and efficient testing process. This dynamically adjusted strategy is also continuously optimized based on historical test data, ensuring that the system can adaptively select the appropriate test case combination to avoid situations where computational complexity is excessive.

[0088] Finally, feedback adjustments are made based on the execution results of the regression test and the implementation effect of the optimization strategy;

[0089] By analyzing the execution results and efficiency of each test case during the testing process, the system can continuously adjust its optimization strategy. For example, if a test case has low execution efficiency but covers a very important test scenario, the system may decide to increase computing resources appropriately to ensure the execution of that test case. Furthermore, based on historical feedback, the system gradually adjusts the parameter weights in the algorithm, allowing the optimization strategy to more accurately reflect actual execution conditions, further improving test efficiency and accuracy. As the test case library grows and testing requirements change, the system can dynamically adapt to new situations and optimize test case selection, ensuring that coverage is maintained while maintaining low computational complexity.

[0090] Dynamic test data adjustment module: Based on the regression test results, the algorithm module automatically adjusts the test data to cope with changes and additions in tenant scheduling strategies, and optimizes subsequent test case generation and execution strategies;

[0091] Based on the regression test results, the algorithm module automatically adjusts the test data to cope with changes and additions to tenant scheduling policies. The specific steps for optimizing subsequent test case generation and execution strategies are as follows:

[0092] After the regression test is completed, the execution results of each test case are evaluated, and the "effectiveness score" and deviation of the tenant scheduling policy are calculated. The regression test results will score each policy to measure whether its performance in different scenarios meets the expected goals. Based on the test results, the algorithm module calculates the "deviation" of the policy and determines whether the test data needs to be adjusted based on the deviation. The deviation calculation formula is as follows:

[0093]

[0094] , where D i is the deviation of the i-th test case, N is the total number of parameters in the test case, T ij is the test value of the jth parameter of the i-th test case, R ij is the expected result value of the jth parameter of the i-th test case, |T ij -R ij | is the error value of the jth parameter of the i-th test case;

[0095] Deviation D i This represents the average deviation between the execution result of test case i and the expected target, reflecting the "performance deviation" of the scheduling policy. A high deviation indicates that the policy does not meet expectations in certain scenarios, and configuration adjustments or the addition of new test data may be necessary. By calculating the deviation for all test cases, we can comprehensively assess the need for policy adjustments.

[0096] Based on the deviation results, the algorithm module automatically adjusts the parameter range of the test data to adapt to changes and additions to tenant scheduling policies. Specifically, the upper and lower limits of the original test data are adjusted according to the deviation value in the test results to ensure that subsequent test cases can cover new test scenarios or policy adjustments. The parameter range adjustment formula is as follows:

[0097] T ij ′=T ij +α·D i ·sign(T ij -R ij )

[0098] , where α is the adjustment coefficient, which controls the adjustment range, sign(T ij -R ij ) is the sign function used to determine the adjustment direction, T ij ′ is the adjusted test data value;

[0099] The core of the adjustment formula is through the deviation D i The adjustment coefficient α controls the adjustment amplitude of each parameter in the test data, and determines the direction of adjustment (i.e., adjusting to a larger or smaller range) based on the actual error direction. The adjustment coefficient α controls the adjustment amplitude of each parameter, ensuring that the adjustment of the test data is neither too large, resulting in over-computation, nor too small, ignoring potential adjustment needs. This ensures that the generated test data is more adaptable to policy changes or new scenarios.

[0100] After the test data is adjusted, an optimized test case set is regenerated based on the new parameter range. The goal of this process is to ensure that the new test cases cover more possible scheduling scenarios and address issues exposed in previous tests. The formula for generating optimized test cases is as follows:

[0101]

[0102] , where C i is the comprehensive score of the optimized i-th test case, W ij is the weight of the jth parameter of the i-th test case, indicating the importance of the parameter to the test case;

[0103] Optimized test case C i It is the comprehensive value obtained by multiplying the adjusted test data with the corresponding weight. Weight W ij This reflects the importance of each parameter in the test case, helping the system prioritize certain parameters for testing based on the actual needs of the policy. This approach allows the system to generate optimized test cases, ensuring higher coverage and better verification of policy effectiveness.

[0104] Finally, based on the newly generated optimized test cases, adjust the subsequent test execution strategy to ensure that the new test cases can be executed efficiently with reasonable computing resources. Based on the optimized test case set and the execution efficiency of each case, calculate resource allocation and determine the execution order using the following formula:

[0105]

[0106] , where A i is the computing resource allocation ratio of the i-th test case, R max is the maximum computing resource available to the system, and M is the total number of test cases.

[0107] By calculating the comprehensive score C for each test case i , dynamically allocates computing resources based on the importance and execution efficiency of test cases. Test cases with higher scores receive more resources, ensuring that high-priority test scenarios are validated first. This optimizes computing resource usage, improves overall testing efficiency, and ensures maximum test coverage.

[0108] The prediction and strategy generation module uses machine learning algorithms to analyze historical test results, predict the adjustment trend of tenant scheduling strategies, and automatically generate new test strategies to ensure high test coverage and accuracy in subsequent versions.

[0109] The specific steps for analyzing historical test results based on machine learning algorithms, predicting the adjustment trends of tenant scheduling policies, and automatically generating new test policies to ensure high test coverage and accuracy in subsequent versions are as follows:

[0110] First, we collect and analyze the data of historical regression test results. This data contains the test results of all past scheduling strategies, including the execution status of each test case, test pass / fail status, execution time, resource consumption, and other information. By analyzing this historical data, we extract key feature parameters. These feature data provide the basis for subsequent trend prediction. Based on the historical data set, we calculate the weighted value of each feature. The calculation expression is as follows:

[0111]

[0112] , where W p is feature F p The weight value in the data set, F p is the pth feature in the historical data, D hist is a historical test dataset, which contains the execution results of historical test cases, strategy change records and their impact on test coverage, and n is the total number of features;

[0113] The weight value of each feature is used to determine its importance in the subsequent prediction model.

[0114] Based on historical data analysis and feature extraction, a trend prediction model is trained using machine learning algorithms (such as regression models). The goal of this model is to predict the adjustment trend of future tenant scheduling strategies, including which strategies may be adjusted, the magnitude of the adjustment, and the impact of these adjustments on system testing. The model uses features extracted from historical data and weighted values W i To construct a training set, we can predict the changes of future strategies. The calculation expression of the adjustment prediction value of the future scheduling strategy is as follows:

[0115] P adj =α·X train +β·W p

[0116] , where P adj is the adjustment forecast value of the future scheduling strategy (such as the frequency and magnitude of the adjustment), X train is the input feature set used for training (including historical features and weighted values), α is the regression coefficient in the regression model, which indicates the degree of influence of the input features on the prediction results, and β is the feature weighting coefficient, which indicates the influence of the feature W in the historical data. p the extent of the impact on the forecast adjustment trend;

[0117] P adjIt is used to judge the changing trend of future scheduling strategies and provide a basis for generating new test strategies.

[0118] Adjusted prediction value P based on future scheduling strategy adj , a new test strategy will be generated. The new test strategy will be optimized based on the predicted possible adjustments to ensure that the new strategy can cover possible changes in future versions. Based on the prediction results, a new test case set will be generated to ensure that the case effectively tests the adjusted scheduling strategy. The formula for generating the new test case set is as follows:

[0119]

[0120] , where T new is the newly generated test strategy, m is the total number of generated test cases, f(P adj ,F p ) is based on the predicted trend P adj and feature F p Compute the generator function for each test case;

[0121] Finally, regression testing is performed based on the generated new test strategy and test case set. If the test results show that the generated test strategy does not meet expectations in terms of high coverage and accuracy, the prediction model will be further optimized based on the regression test results, and the generated test strategy will be adjusted. Through this feedback mechanism, the test strategy is dynamically optimized to ensure that high coverage and accuracy can be maintained in subsequent versions. The optimized measurement expression is as follows:

[0122]

[0123] , where T final is the final optimized test strategy, E test is the evaluation value of the regression test results, measuring the coverage and accuracy of the new strategy, E max It is the preset optimal coverage and accuracy target value.

[0124] Through this feedback mechanism, new strategies are continuously adjusted to ensure their effectiveness in practical applications. Specific implementation 1:

[0126] In the SaaS travel platform, tenants will customize scheduling strategies based on their own operational needs and business scenarios. The scheduling strategy includes multiple key parameters, such as pick-up time, vehicle selection, driver allocation method, passenger needs, etc. The system first automatically obtains the tenant's scheduling strategy configuration information through the API interface or the tenant's configuration interface. This information is generally transmitted to the test system in a structured manner (such as JSON, XML or database tables, etc.). In order to ensure the smooth execution of subsequent strategy matching and test execution, the system will format and standardize the received configuration data, remove redundant information, and convert it into a standard format that can be used for subsequent analysis and matching.

[0127] Next, the system analyzes the tenant's scheduling strategy using a specific parsing algorithm, automatically extracting key parameters. These parameters can include timeliness requirements (such as the maximum tolerance for pickup times), resource allocation rules (such as driver and vehicle scheduling priorities), and business scenarios (such as scheduling differences during peak and off-peak periods). Through parameterization, the system transforms every detail of the scheduling strategy into an operational, standardized format, providing accurate input for subsequent matching algorithms.

[0128] The system categorizes and hierarchizes each dispatch strategy. For example, for dispatch strategies involving multiple cities, the system automatically identifies differences in dispatch strategies across different regions based on city-specific traffic rules and travel needs. For another example, when a strategy includes time periods (such as peak hours and nighttime), the system subdivides the dispatch strategy based on these time dimensions to ensure that the dispatch rules for each time period are verified. This automated parsing process significantly reduces manual intervention, reduces human error, and improves data accuracy and consistency.

[0129] After parsing the tenant's scheduling policy, the system enters the policy matching phase. At this point, the system compares the tenant's scheduling policy with the existing policy library. The policy library stores the platform's historical scheduling policies and their associated test results for different tenants. By establishing a similarity assessment model, the system automatically matches the most similar policy template based on the templates in the existing policy library. To improve matching accuracy, the system comprehensively evaluates policy similarity based on the matching degree of multiple parameters (such as time period, resource requirements, and traffic conditions).

[0130] The matching process isn't limited to a single, exact match. Instead, intelligent algorithms perform fuzzy matching across policies, identifying those with potential similarities. For example, the system can determine whether different policies can share test cases based on factors such as traffic patterns in the tenant's area, similarities in travel needs, and policy complexity. This process, which combines pattern recognition and data mining techniques, enables automated and efficient matching, reducing the need for manual intervention and significantly improving the efficiency of policy testing.

[0131] Based on policy matching, the system automatically generates a set of test cases. Each test case represents a specific execution scenario for a particular scheduling policy, covering the policy's key parameters and possible business scenarios. For example, for a policy with a time limit for pickup, the system generates multiple test cases covering different time periods and pickup conditions (such as road conditions and passenger location). The generated test cases not only simulate common scenarios but also cover boundary conditions and abnormal situations (such as traffic congestion and sudden orders) to ensure the robustness of the scheduling policy.

[0132] These test cases are stored in a test case library. The system uses intelligent search and filtering mechanisms to quickly locate the test cases that need to be executed. During regression testing, the system automatically extracts relevant test cases from the test case library and automatically adjusts the case combination based on changes in strategy to ensure comprehensiveness and accuracy of testing.

[0133] Through automated test case generation and matching, the system significantly reduces the need for manual intervention and improves testing efficiency and accuracy. Automatically parsing, matching, and generating test cases for tenant scheduling policies effectively covers all possible scenarios and parameter combinations. This approach not only reduces errors prone to traditional manual testing but also minimizes resource waste during testing, providing an efficient, intelligent, and cost-effective scheduling policy verification solution for SaaS platforms.

[0134] Specific implementation method 2: After the system generates test cases, a key step in regression testing is how to efficiently execute these test cases. During this process, the system first analyzes the execution efficiency of each test case. Execution efficiency is a key criterion for evaluating test case effectiveness, and involves factors such as execution time, resource consumption, and computational complexity. The system automatically evaluates the efficiency of each test case based on historical execution data and resource consumption models. For example, some test cases may require simulating complex traffic conditions or resource allocation during rush hour, which are more expensive to execute. However, simpler test scenarios execute faster and consume fewer resources.

[0135] To this end, the system uses an algorithm to weight each test case, assigning each case an execution efficiency score based on its execution time, resource consumption, and test scenarios covered. This score helps the system prioritize test cases with high execution efficiency and coverage of more critical scenarios during subsequent test execution. This approach effectively avoids the over-execution of inefficient tests and ensures the overall efficiency of regression testing.

[0136] Based on each test case's execution efficiency score, the system prioritizes efficient test cases according to a specific strategy. These efficient test cases typically have high resource utilization, can quickly verify multiple key scenarios, and have low computational complexity. The system automatically selects test cases that maximize coverage and test efficiency, thereby reducing the execution of irrelevant or redundant test cases.

[0137] Furthermore, the system dynamically adjusts the execution order during testing based on real-time feedback. For example, if a test case exhibits high resource consumption early in the test, the system will use intelligent scheduling to delay its execution, ensuring that more efficient test cases are prioritized. The system can also dynamically optimize the execution order and combination of test cases based on the real-time performance of regression tests, ensuring both efficiency and comprehensiveness.

[0138] During regression testing, the system monitors the execution of each test case in real time and dynamically adjusts computing resource allocation based on execution time, results, and resource consumption. If certain test cases are inefficient, the system schedules computing resources to ensure that their execution does not impact the smooth flow of the entire regression testing process. For example, inefficient test cases may be assigned to more powerful computing resources or distributed computing platforms, while efficient test cases can be executed with lower resources, thus optimizing overall resource usage.

[0139] Dynamic adjustments aren't limited to test case selection; they also involve optimizing the test environment. For example, during test execution, the system can dynamically allocate more computing resources to resource-intensive test cases, ensuring that test execution isn't delayed due to resource bottlenecks. Furthermore, the system can dynamically optimize test case selection based on the testing requirements and policy changes of different tenants, ensuring that each regression test covers new changes quickly and effectively.

[0140] By dynamically optimizing test case selection and execution strategies, the system maximizes test coverage while maintaining low computational complexity. Compared to traditional testing methods, this optimization approach significantly reduces redundant testing, improves test efficiency, and conserves computing resources. Through intelligent dynamic adjustments, the system optimizes test case execution order and resource allocation based on real-time feedback from regression testing, ensuring that regression testing can efficiently and accurately verify the effectiveness of scheduling strategies in multi-tenant and multi-scenario scenarios.

[0141] Specific Implementation 3: To enhance the foresight and intelligence of regression testing, this implementation incorporates machine learning algorithms to analyze and predict based on historical test data. The system collects and stores the results of each regression test, including the performance of each test case, adjustments to tenant scheduling policies, and execution status under different test scenarios. By analyzing this data, the system can identify the impact of policy adjustments on test results and which policy changes have a significant impact on platform operations.

[0142] system

[0143] Machine learning models (such as decision trees, support vector machines, and neural networks) are used to train historical data and build predictive models. This model can predict potential future scheduling policy adjustments for tenants based on patterns in the historical data. For example, based on historical data from certain policy adjustments, the system can determine which adjustments might lead to delays in pickup times or uneven resource allocation, and prepare corresponding test cases in advance.

[0144] Based on trend predictions from historical data, the system can pre-generate new test strategies and test cases, ensuring that scheduling policy adjustments in future versions can be promptly verified. For example, if historical data indicates that a tenant frequently adjusts their pickup strategy within a specific time period, the system will predict that the tenant is likely to make similar adjustments in the future and pre-generate test cases related to pickup times. This way, the system can proactively respond to changes in tenant scheduling policies and ensure that test cases always cover new scenarios that may arise.

[0145] This data-driven test strategy generation not only improves the intelligence of testing but also significantly reduces test preparation time and avoids testing blind spots caused by policy adjustments. By analyzing patterns in historical data, the system can intelligently generate the most appropriate test strategies and test cases based on tenants' business needs and historical performance.

[0146] Over time, the system continuously optimizes its machine learning models based on new test data and policy adjustments. This continuous learning allows the system to dynamically adjust its policies and test case generation process based on the results of each regression test. For example, if a policy adjustment results in a high test failure rate, the system will learn from the failure patterns in historical data and optimize its test case generation strategy to ensure better coverage of future risks associated with these adjustments.

[0147] This continuous optimization process, powered by machine learning, enables the system to adapt to evolving travel market demands, proactively identify potential scheduling policy issues, and ensure the accuracy and coverage of testing. Through a data-driven approach, the system generates customized testing strategies and test cases based on each tenant's operational status and business model, enhancing the foresight and adaptability of testing.

[0148] Machine learning-based historical data analysis and policy optimization not only improves the intelligence of testing but also enables regression testing to be more proactive and dynamically adapt to tenant policy adjustments. Through continuous learning and optimization of machine learning models, the system can predict future policy adjustments based on trends in historical data and prepare corresponding test cases in advance. This intelligent prediction and optimization significantly improves testing efficiency, reduces the need for manual intervention, and enhances the system's adaptability, providing strong support for dynamic scheduling policy verification on SaaS platforms.

[0149] The present invention greatly improves the efficiency of the SaaS travel platform's scheduling strategy regression test through an automated test case generation and matching mechanism. When a new tenant joins or an existing tenant's scheduling strategy changes, the system can automatically match the existing scheduling strategy test set and quickly trigger the corresponding regression test. The system automatically identifies and applies matching test cases based on the tenant's scheduling strategy, avoiding the process of manual intervention and ensuring that the platform can verify the effectiveness of the new strategy in a timely and comprehensive manner after each policy change. With the continuous enrichment of the scheduling strategy library, the proportion of unmatched strategies has gradually decreased, the degree of test automation has continued to improve, and the test efficiency has been significantly enhanced. The system can cover a wider range of test scenarios, reduce the workload of repeated manual testing, greatly improve the testing efficiency of the travel platform, and shorten the cycle from strategy adjustment to online verification.

[0150] The present invention not only reduces the cost of manual intervention through automated testing, but also significantly improves the flexibility of the platform in dealing with different tenants and policy changes. With the continuous improvement of the scheduling policy library, the system can quickly adapt to new tenant configurations or policy changes, and intelligently match appropriate test cases for verification, avoiding the tedious work of manually designing new test cases for each change. This automatic matching mechanism reduces the need for manual testing and data preparation, thereby reducing the overall cost of testing. At the same time, when faced with a complex multi-tenant environment, the system can flexibly generate relevant test cases based on the personalized needs of each tenant and efficiently verify their policies. This flexibility ensures that the platform can quickly respond to market changes and tenant needs, enhances the adaptability and scalability of the SaaS platform, and makes it more competitive in the dynamic travel market.

[0151] The above description is merely illustrative of certain exemplary embodiments of the present invention. It goes without saying that those skilled in the art will be able to modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and description are illustrative in nature and should not be construed as limiting the scope of protection of the claims.

Claims

1. A dispatching strategy automatic verification device for SaaS travel platform, characterized in that: It includes scheduling strategy parameter analysis and matching module, regression test execution and verification module, test case optimization and selection module, dynamic test data adjustment module, and prediction and strategy generation module: The scheduling policy parameter analysis and matching module automatically analyzes the scheduling policy parameters based on the tenant's scheduling policy configuration information, automatically matches them with the existing policy library, and generates a set of test cases; The regression test execution and verification module automatically executes the regression test of the scheduling strategy through an algorithm based on the generated test case set, and determines whether the strategy is effective based on the test results; The test case optimization and selection module uses a matching algorithm to calculate the execution efficiency of each test case and optimize the selection of test cases to ensure high coverage and low computational complexity; Dynamic test data adjustment module: Based on the regression test results, the algorithm module automatically adjusts the test data to cope with changes and additions in tenant scheduling strategies, and optimizes subsequent test case generation and execution strategies; The prediction and strategy generation module, based on machine learning algorithms, analyzes historical data of test results, predicts adjustment trends of tenant scheduling strategies, and automatically generates new test strategies to ensure high test coverage and accuracy in subsequent versions.

2. The automatic verification device for scheduling strategy of SaaS travel platform according to claim 1 is characterized in that: Based on the tenant's scheduling policy configuration information, the scheduling policy parameters are automatically analyzed and automatically matched with the existing policy library. The specific steps for generating a test case set are as follows: First, receive the tenant's scheduling policy configuration information from the SaaS travel platform; All relevant scheduling parameters will be extracted from the tenant's scheduling policy; Once the core parameters of the scheduling policy are extracted and standardized, the parameters are matched with the existing scheduling policy library; Based on the matching results with the strategy library, a test case set is automatically generated.

3. The automatic verification device for scheduling strategy of SaaS travel platform according to claim 1 is characterized in that: Based on the generated test case set, the algorithm is used to automatically execute the regression test of the scheduling strategy, and the specific steps to determine whether the strategy is effective based on the test results are as follows: After generating the test case set, prepare the regression testing environment; After the test environment is ready, start automatically executing regression tests; During the regression test execution process, the execution status of each test case will be monitored in real time and dynamic feedback will be provided; After the regression test is completed and the results are analyzed, the effectiveness of the scheduling strategy is judged based on the preset evaluation criteria.

4. The automatic verification device for scheduling strategy of SaaS travel platform according to claim 1 is characterized in that: The specific steps for using a matching algorithm to calculate the execution efficiency of each test case and optimize the selection of test cases to ensure high coverage and low computational complexity are as follows: During the regression test execution process, the execution efficiency of each generated test case is evaluated; After evaluating the execution efficiency of each test case, the matching algorithm is used to screen and optimize the test cases; After optimization and screening, the test case selection strategy is dynamically adjusted to ensure that the computational complexity of the test remains within a normal range while ensuring high coverage; Finally, feedback adjustments are made based on the execution results of the regression test and the implementation effect of the optimization strategy.

5. The automatic verification device for scheduling strategy of SaaS travel platform according to claim 1 is characterized in that: After the regression test is completed, the execution results of each test case are evaluated, and the "effectiveness score" and deviation of the tenant scheduling policy are calculated. The regression test results will score each policy. Based on the test results, the algorithm module calculates the "deviation" of the policy and determines whether the test data needs to be adjusted based on the deviation. The deviation calculation formula is as follows: ， Where D i is the deviation of the i-th test case, N is the total number of parameters in the test case, T ij is the test value of the jth parameter of the i-th test case, R ij is the expected result value of the jth parameter of the i-th test case, |T ij -R ij | is the error value of the jth parameter of the i-th test case; Based on the deviation results, the algorithm module automatically adjusts the parameter range of the test data to adapt to changes and additions to tenant scheduling policies. The parameter range adjustment formula is as follows: T ij ′=T ij +α·D i ·sign(T ij -R ij ) Where α is the adjustment coefficient, which controls the adjustment range, sign(T ij -R ij ) is the sign function used to determine the adjustment direction, T ij ′ is the adjusted test data value; After completing the test data adjustment, regenerate the optimized test case set based on the new parameter range. The optimized test case generation formula is as follows: ， Where C i is the comprehensive score of the optimized i-th test case, W ij is the weight of the jth parameter of the i-th test case; Finally, based on the newly generated optimized test cases, adjust the subsequent test execution strategy to ensure that the new test cases can be executed efficiently with reasonable computing resources. Based on the optimized test case set and the execution efficiency of each case, calculate resource allocation and determine the execution order using the following formula: ， Where A i is the computing resource allocation ratio of the i-th test case, R max is the maximum computing resource available to the system, and M is the total number of test cases.

6. The automatic verification device for scheduling strategy of SaaS travel platform according to claim 1 is characterized in that: The specific steps for analyzing historical test results based on machine learning algorithms, predicting the adjustment trends of tenant scheduling policies, and automatically generating new test policies to ensure high test coverage and accuracy in subsequent versions are as follows: First, we collect and analyze the data of historical regression test results. By analyzing these historical data, we extract the key feature parameters and calculate the weighted value of each feature based on the historical data set. The calculation expression is as follows: ， Where W p is feature F p The weight value in the data set, F p is the pth feature in the historical data, D hist is the historical test dataset, n is the total number of features; Based on historical data analysis and feature extraction, a machine learning algorithm is used to train a trend prediction model. The model uses the features extracted from historical data and the weighted value W. i To construct a training set, we can predict the changes of future strategies. The calculation expression of the adjustment prediction value of the future scheduling strategy is as follows: P adj =α·X train +β·W p , Where, P adj is the adjusted forecast value of the future scheduling strategy, X train is the input feature set used for training, α is the regression coefficient in the regression model, which indicates the influence of the input feature on the prediction result, and β is the feature weighting coefficient; Adjusted prediction value P based on future scheduling strategy adj , a new test strategy will be generated. According to the prediction results, a new test case set will be generated, and it will be ensured that the case effectively tests the adjusted scheduling strategy. The new test case set generation formula is as follows: ， Where, T new is the newly generated test strategy, m is the total number of generated test cases, f(P adj ,F p ) is based on the predicted trend P adj and feature F p Compute the generator function for each test case; Finally, regression testing is performed based on the generated new test strategy and test case set. If the test results show that the generated test strategy does not meet expectations in terms of high coverage and accuracy, the prediction model will be further optimized based on the results of the regression test, and the generated test strategy will be adjusted. The optimized measurement expression is as follows: ， Where, T final is the final optimized test strategy, E test is the evaluation value of the regression test result, E max It is the preset optimal coverage and accuracy target value.

Images