Mobile equipment automatic testing method based on variable optimization

Abstract

The invention provides an automatic testing method for mobile equipment based on variable optimization. The automatic testing method comprises the steps of obtaining a variable set and generating a verification numerical value, optimizing a verification group in combination with historical test data and a machine learning algorithm, dynamically generating content items and recording verification result information. The method comprises the steps of analyzing verification group input, loading a template specification structure, calling a recommendation algorithm to generate content items, and dynamically optimizing generation logic based on user response and external data. And for verification result information, marking content items which do not pass verification and recording reasons, and forming an optimized feedback closed loop by refreshing a template standard structure to adjust a recommendation algorithm weight, optimize a page layout rule and improve a dynamic adjustment strategy. According to the method, the test coverage rate, the efficiency and the adaptability of content generation logic are improved, and the method is suitable for efficient automatic testing in a complex scene.

Description

Technical Field

[0001] The present invention relates to the technical field of mobile device testing, and in particular to an automatic testing method for mobile devices based on variable optimization. Background Art

[0002] With the widespread popularity of mobile devices and their increasing intelligence, mobile content generation and real-time interaction with users are becoming increasingly complex and diverse. In order to meet the needs of users for personalized experience, mobile devices need to be able to quickly generate diverse content based on different scenarios, user needs, and external data. However, to ensure the accuracy and efficiency of these content generation logics, comprehensive testing of mobile content generation units has become an inevitable requirement.

[0003] In the prior art, traditional mobile device testing methods mainly rely on manual testing or automated testing based on simple scripts. Although these methods can complete basic testing tasks, they have significant limitations and shortcomings. The first is the problem of insufficient test coverage. Manual testing and simple scripted testing cannot fully cover all input variable combinations, which makes many extreme scenarios or boundary conditions not fully verified, resulting in abnormal system operation or user experience problems. Secondly, the test efficiency is low. Traditional testing methods usually require a lot of manual intervention and operation, which consumes valuable time and human resources, and is particularly difficult to meet the needs of frequent updates and iterations of modern mobile applications.

[0004] In addition, existing methods also have obvious defects in adaptability and flexibility. Current tests usually rely on static preset templates or rules, and lack the ability to dynamically adjust to real-time data changes and user behavior responses. This defect causes the test results to be unable to accurately reflect the actual performance in actual usage scenarios, thereby weakening the reliability of the test. At the same time, traditional testing methods fail to fully utilize the complex correlations between multiple variables that affect content generation. Typically, these variables are treated independently, which ignores the potential interactions between variables, causing some potential problems to be missed in the test.

[0005] Another significant problem is that the feedback mechanism of existing testing methods is relatively weak. Test results often fail to provide sufficient support for subsequent optimization and improvement, which makes the optimization process of content generation logic slow and inefficient. The lack of an effective feedback mechanism means that even if problems are found, it is difficult to systematically improve the test logic and content generation rules.

[0006] In summary, the existing technologies have many problems and deficiencies in the testing methods for real-time content generation units on mobile terminals, including low test coverage, low efficiency, poor dynamic adaptability, insufficient use of correlation between variables, and lack of effective feedback mechanism. This situation urgently requires an automated testing method that can take into account efficiency, comprehensiveness, and dynamic adaptability to comprehensively improve test coverage and accuracy, and effectively promote the improvement and optimization of content generation logic. Summary of the invention

[0007] In order to solve the above problems in the prior art, the present invention proposes a mobile device automated testing method based on variable optimization, comprising the following steps:

[0008] Step 1: obtaining an initial range of each variable of a plurality of variables in a variable set, wherein the variable set defines a mobile terminal content feature generated by a real-time content generation unit; for each variable in the variable set, generating a plurality of verification values ​​according to its initial range;

[0009] Step 2: Based on historical test data and machine learning algorithms, multiple verification groups are automatically optimized and generated, each verification group contains different verification value combinations of multiple variables in the variable set; multiple optimized verification groups are provided to the mobile terminal real-time content generation unit;

[0010] Step 3: In the real-time content generation unit executed by the mobile device, for each optimized verification group, according to the verification value combination of the verification group as the input of the template specification structure, the mobile content item is dynamically generated;

[0011] Step 4: in the test module executed by the mobile device, for each mobile terminal content item output by the real-time content generation unit, associating it with the verification value of the verification group that generated the content item to obtain relevant verification result information; outputting the verification result information including the mobile terminal content item and its associated verification value;

[0012] Step 5: receiving evaluation inputs of the content items according to the relevant verification result information of each mobile terminal content item output by the real-time content generation unit; determining that the evaluation input of at least one content item indicates that it does not meet the mobile terminal verification specification;

[0013] Step 6: Based on the evaluation input indicating the mobile verification failure, refresh the template specification structure to optimize the content generation logic.

[0014] The variable set includes user behavior data variables, external environment parameter variables and built-in parameter variables of content generation logic.

[0015] The generation rule of the verification value includes a uniform distribution generation method, a random distribution generation method, an extreme value generation method or a distribution function generation method.

[0016] The process of optimizing the validation group based on the machine learning algorithm in step 2 includes:

[0017] Extract the eigenvector of the validation group variables;

[0018] Training classification models based on historical test data;

[0019] The importance of each variable on the quality of content generation is calculated based on the classification model, and the verification group is screened.

[0020] The step 3 further comprises

[0021] Parse the validation group input parameters and map them to the corresponding parameters of the content generation logic;

[0022] Load the template specification structure,

[0023] Call the recommendation algorithm to generate recommended content, dynamically adjust the resource loading strategy according to network-related parameters, and generate content items in combination with the page layout module;

[0024] Dynamically optimize content generation logic based on real-time user responses or external data changes;

[0025] The data associated with the content item and the validation set is recorded and stored for subsequent testing and evaluation.

[0026] The template specification structure includes a page layout module, a recommendation algorithm module and a dynamic rule adjustment module.

[0027] The verification result information of step 4 includes the display quality index of the content item, user interaction data and content generation time.

[0028] The step 5 further includes comparing the verification result information item by item based on the preset evaluation specification to determine whether the content item meets the quality and performance requirements;

[0029] If the content item fails the verification, the failed content item is marked, and the content item identifier, the failed indicator and the reason are recorded;

[0030] Generates recommended adjustments for failed content items;

[0031] Store tag information in logs or databases for subsequent analysis and optimization

[0032] Refresh the template specification structure according to the analysis results, including adjusting the recommendation algorithm weights, optimizing page layout rules, and improving dynamic adjustment strategies.

[0033] Reintroduce the refreshed template specification structure into the test process, generate new test data and record the verification results to form an optimization feedback loop.

[0034] Beneficial effects:

[0035] The present invention provides a mobile device automation testing method based on variable optimization, which significantly improves the test coverage and efficiency by obtaining a variable set and generating a verification value, combining historical test data and a machine learning algorithm to optimize the verification group. In a real-time content generation unit, content items are dynamically generated and optimized in combination with user responses or external data, thereby improving the adaptability of the content generation logic. By recording the associated data of the verification group and the content item and generating verification result information, the comprehensiveness and traceability of the test results are achieved. In addition, based on the verification result information, the content items that have not passed the verification are marked, and the recommendation algorithm, page layout and dynamic adjustment strategy are optimized by refreshing the template specification structure, thereby effectively improving the content generation quality and system performance. The method of the present invention forms an optimization feedback closed loop, which can continuously improve the test process and content generation logic, is suitable for efficient automated testing in complex multi-variable scenarios, and has significant practical value and promotion significance. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of the present application, but do not constitute an improper limitation of the present invention. In the drawings:

[0037] Figure 1 A flow chart of the method of the present invention is shown. DETAILED DESCRIPTION

[0038] The present invention will be described in detail below in conjunction with the accompanying drawings and specific embodiments, wherein the illustrative embodiments and descriptions are only used to explain the present invention but are not intended to limit the present invention.

[0039] The following describes in detail an implementation of an automated testing method for a mobile device of the present invention in conjunction with the accompanying drawings. Figure 1 As shown, the specific contents of the present invention are as follows:

[0040] Step 1: Get the variable range and generate verification values

[0041] In the technical implementation scheme of this invention, a variable set is first defined, which includes multiple key variables that affect the content characteristics of the mobile content generation unit. The content of the variable set is determined according to the actual needs of the application scenario. For example, in a typical mobile application scenario, the variable set includes the following types of variables:

[0042] 1. User behavior data variables: including user click frequency, page dwell time, sliding speed, etc. These variables directly reflect the user's interaction behavior with the mobile terminal. For example, the initial range of the user click frequency variable can be set to 0 to 100 times per minute to cover low-frequency to high-frequency click scenarios.

[0043] 2. External environment parameter variables: including network conditions (such as latency, bandwidth, etc.), device type (such as mobile phone model, operating system version, etc.), and external hardware limitations (such as memory usage, battery status, etc.). For example, the initial range of network latency can be set to 10 milliseconds to 200 milliseconds to cover content generation in low-latency and high-latency environments.

[0044] 3. Built-in parameter variables of content generation logic: including template structure attributes (such as content arrangement order, page layout parameters, etc.) and parameters in the recommendation algorithm (such as weight distribution, priority setting, etc.). These variables are used to test the responsiveness of content generation units under different template configurations.

[0045] For each variable in the above variable set, first determine its initial range. The initial range of the variable can be obtained through empirical rules, industry standards, or actual data statistics. For example, the range of some variables is uniformly distributed (such as user click frequency distributed between 0 and 100 times per minute), while other variables follow normal distribution or exponential distribution (such as network delay is concentrated in a certain range, but occasionally extreme values ​​appear).

[0046] Next, generate multiple verification values ​​based on the initial range of each variable. The verification value generation rules can be selected according to the actual test requirements to ensure test coverage. The following are several methods for generating verification values:

[0047] Uniform distribution generation method: Divide the initial range of the variable into several segments evenly, and take one or more values ​​in each segment. For example, for network delay (10 milliseconds to 200 milliseconds), it can be divided into sub-ranges with intervals of 10 milliseconds, and each sub-range takes a verification value, such as 10, 20, 30 milliseconds, up to 200 milliseconds.

[0048] Random distribution generation method: Generate a certain number of random values ​​based on the initial range. Constraints can be added during random generation to control deviations. For example, for the user's click frequency (0 to 100 times per minute), you can randomly generate several integer values, such as 15, 35, 70, etc., to ensure that each generated value is within a reasonable range.

[0049] Extreme value generation method: select boundary values ​​or values ​​close to the boundary within the initial range for verification. For example, for the page dwell time (0 to 300 seconds), select values ​​close to 0 seconds (extremely short dwell) and close to 300 seconds (extremely long dwell) for testing to cover the boundary cases.

[0050] Distribution function generation method (if applicable): If the initial range of some variables conforms to a known distribution model (such as normal distribution, exponential distribution, etc.), the distribution function can be used to generate verification values. For example, for the user's sliding speed, assuming that it presents a normal distribution (mean is 50 pixels / second, standard deviation is 10 pixels / second), the normal distribution function can be used to generate verification values ​​to reflect the actual user behavior characteristics.

[0051] The verification value generation result can be expressed as a multi-dimensional array, where each dimension corresponds to a variable. For example, for the three variables of user click frequency, network delay, and page layout, the generated verification value array is: [

[0053] [10,20,30],#User click frequency

[0054] [50,100,150],#Network delay

[0055] [1,2,3]#Page layout ]

[0057] The generated validation values ​​will serve as the basic input data set for the test, providing support for subsequent optimization and validation steps. In the data generation process, automated processing can be achieved through programs, such as using programming languages ​​(such as Python or Matlab) to write value generation scripts to ensure efficiency and accuracy.

[0058] Through the above process, we can ensure that the variable range is reasonably defined and the verification value generation is comprehensive, which can provide efficient input support for the testing of the mobile content generation unit.

[0059] Step 2: Optimize the validation set

[0060] Based on the verification value set generated in step 1, this implementation scheme optimizes and generates multiple verification groups by combining historical test data and machine learning algorithms. Each verification group is composed of different verification value combinations of multiple variables in the variable set, with the purpose of reducing redundant data to improve test efficiency while ensuring test coverage.

[0061] The specific implementation steps are as follows:

[0062] 1. Verify the initial combination of values

[0063] A preliminary validation set is generated by permuting and combining the multi-dimensional validation values ​​generated in step 1. For example, suppose there are two variables:

[0064] User click frequency (value range: 30, 50, 70 times per minute)

[0065] Network delay (value range: 50, 100, 150 milliseconds)

[0066] Through the Cartesian product, 9 preliminary validation groups can be generated:

[0067] [30,50],[30,100],[30,150],

[0068] [50,50],[50,100],[50,150],

[0069] [70,50],[70,100],[70,150]

[0070] 2. Historical test data analysis

[0071] The representativeness and redundancy of the above preliminary verification group are analyzed by using historical test data. For example, by counting previous test records, it can be determined whether certain variable combinations appear frequently or have a critical impact. For example:

[0072] If historical data shows that "network latency = 150 milliseconds" has less impact on content generation, you can lower the priority of this value in the verification group;

[0073] If it is found that certain problems are more easily exposed when "user click frequency = 70 times per minute", such combination will be retained first.

[0074] 3. Machine Learning Algorithm Screening

[0075] In the process of optimizing the validation group, a machine learning algorithm is used to model and screen the effectiveness of the validation group. Specifically, the following steps can be taken:

[0076] Feature extraction: Treat the variable values ​​of each validation group as a feature vector. For example, for the validation group [30,50], the feature vector is [30,50].

[0077] Classification model training: Label the test results (such as success or failure) corresponding to each feature vector based on historical data. Classification models such as decision trees, random forests, or support vector machines can be used for training.

[0078] Importance assessment: The model calculates the importance of each variable to the classification result. For example, the user click frequency has a greater impact on the quality of content generation, while the network delay has a smaller impact. Based on the importance assessment results, the variable combination weights in the validation group can be adjusted.

[0079] Validation group clustering: Use a clustering algorithm (such as K-means or hierarchical clustering) to cluster the preliminary validation groups and remove validation groups with high repetitiveness or small effect differences. For example, if the test results of validation groups [30,50] and [30,100] are similar, you can selectively keep one.

[0080] 4. Optimize verification group generation

[0081] Based on the above analysis and screening, an optimized validation group is generated. For example, after machine learning model screening and clustering optimization, the following validation groups are finally retained:

[0082] [30,50],[50,100],[70,150]

[0083] These verification groups can more effectively cover key test scenarios and significantly reduce unnecessary redundancy.

[0084] 5. Verify group sorting and priority setting

[0085] For the optimized validation groups, sort them according to test importance or variable priority. For example, if "user click frequency = 70 times per minute" is a high-priority scenario, [70,150] is used as the primary test group. In addition, test weights can be assigned to each validation group to ensure that key scenario tests are completed first when resources are limited.

[0086] 6. Verify group output

[0087] The validation group generated by the optimization is output as structured data for use in subsequent test steps. For example, the validation group can be stored in JSON format so that it can be directly input into the mobile real-time content generation unit: [

[0089] {"User click frequency":30,"Network delay":50},

[0090] {"User click frequency":50,"Network delay":100},

[0091] {"User click frequency":70,"Network delay":150} ]

[0093] Through the above steps, the generated verification group can not only cover a wide range of test scenarios, but also reduce redundancy through the support of historical data and machine learning algorithms, and improve test efficiency and resource utilization. This optimization method is applicable to scenarios with various numbers and distributions of variables, and can significantly improve the effect of automated testing.

[0094] Step 3: Generate content items

[0095] In this embodiment, in the real-time content generation unit of the mobile device, the mobile terminal content items are dynamically generated according to the optimized verification group. The specific process is as follows:

[0096] 1. Verify group input parsing

[0097] The content generation unit receives input parameters from the optimization verification group. The verification group includes a combination of multiple variables and their verification values, such as "user click frequency = 30 times per minute" and "network delay = 50 milliseconds". The system first parses the input data and maps the values ​​of each variable to the corresponding parameters of the content generation logic. For example:

[0098] Map "user click frequency" to page interaction priority parameters;

[0099] Map "network latency" to resource loading speed adjustment parameters.

[0100] 2. Template specification structure loading

[0101] The content generation unit loads the established template specification structure to guide the content generation process. The template specification structure may include the following modules:

[0102] Page layout module: defines the arrangement rules of content items, such as displaying news items in a waterfall flow or displaying video recommendations in a grid.

[0103] Recommendation algorithm module: Call the corresponding recommendation algorithm based on the verification group input parameters. For example, when the user clicks frequently, the recommendation algorithm will give priority to returning short-term and high-frequency content.

[0104] Dynamic rule adjustment module: allows content generation rules to be adjusted according to real-time changes, such as dynamically reducing the priority of large-capacity resources in combination with network delay parameters.

[0105] 3. Content Generation Process

[0106] After parsing the validation group and loading the template, the content generation unit generates the content item according to the following steps:

[0107] Calling the recommendation algorithm: Calling the recommendation algorithm based on the user behavior-related variables (such as click frequency and dwell time) in the verification group. For example, if the click frequency is 30 times per minute, the system selects algorithm model A to push popular short video content.

[0108] Dynamically adjust content logic: adjust resource loading strategies based on network-related parameters (such as latency, bandwidth). For example, when the network latency is 50 milliseconds, the system allows loading of high-definition images; if the latency is close to 200 milliseconds, it automatically switches to low-resolution resources.

[0109] Generate content layout: Combined with the page layout module, the recommended content is arranged according to the established template structure. For example, a news content card containing a title, thumbnail, and description information is generated.

[0110] 4. Real-time response optimization

[0111] The content generation unit can further optimize the generation logic by combining real-time user responses or external data changes. For example:

[0112] If the system detects that the user's click rate for the current recommended content is low (eg, less than 5%), the weight parameters of the recommendation algorithm are dynamically adjusted to increase the proportion of content related to the user's interests.

[0113] If external data (such as server status) changes, the system can dynamically adjust resource loading strategies. For example, in high-load situations, the frequency of generating high-resolution content is reduced.

[0114] 5. Content Item Generation Example

[0115] For a certain verification group ("user click frequency = 30 times per minute", "network delay = 50 milliseconds"), the content generation unit outputs the following content items:

[0116] Video recommendation list: Push a recommendation list containing 5 short videos, each of which is no longer than 1 minute.

[0117] News content page: Generates a page containing 10 news items, each with a thumbnail and a brief description of no more than 50 words.

[0118] 6. Generate records and store them

[0119] Each generated content item will record its association with the verification group and store it in the log. For example:

[0120] The recorded associated data will be used in subsequent testing and evaluation steps.

[0121] Through the above implementation scheme, the real-time content generation unit can efficiently generate content items based on the optimized verification group, and at the same time has the ability to dynamically adapt to the external environment and user response, thereby improving the pertinence of content generation and user experience.

[0122] Step 4: Associate the verification results

[0123] In this embodiment, the testing module of the mobile device is responsible for associating each content item output by the content generation unit with the corresponding verification group, and outputting verification result information to provide data support for subsequent evaluation.

[0124] The specific implementation process is as follows:

[0125] 1. Records of association between content items and verification groups

[0126] After the content generation unit generates content items, the test module records the source of each content item, including feature information of the content item (such as type, generation time, display style, etc.) and corresponding verification group information (such as variables in the verification group and their value combinations).

[0127] For example, for verification group A ("user click frequency = 30 times per minute" and "network delay = 50 milliseconds"), if the content generation unit outputs a list of recommended videos, the test module will record the following data:

[0128] 2. Storage format of linked data

[0129] The test module stores the associated data in a structured form, which can be a table, log file or database record, for subsequent access and analysis.

[0130] Tabular format: Data is stored in a two-dimensional table format, with each row recording a content item and its associated verification group. For example:

[0131] |Verification Group|Content Item|Generation Time|

[0132] |Click frequency=30, delay=50 |Recommended video list (5 videos) |500ms |

[0133] |Click frequency=70,delay=100|News page (10 news) |800ms |

[0134] Log file format: Data is stored as text log files, and the formatted strings are easy to read and parse:

[0135] [2024-11-26 10:00:00] Verification group: [Click frequency = 30 times / minute, delay = 50 milliseconds] -> Content item: Recommended video list (5 videos), generation time 500ms

[0136] Database record form: It uses relational database storage, and the fields include verification group ID, variable value, content item description, generation index, etc., which is convenient for large-scale query and statistical analysis.

[0137] 3. Generation of verification result information

[0138] The test module further generates verification result information describing the performance indicators and presentation quality of the content item. For example, the verification result information may include the following:

[0139] Display quality indicators: record the display effect of content items on the interface, such as resolution, image loading completeness, recommended content relevance score, etc.

[0140] User interaction data: records the response of content items in actual interactions, such as user click rate, page dwell time, number of sliding operations, etc.

[0141] Generation time: records the time consumed in the content generation process to reflect the performance of the generation unit.

[0142] For some indicators (such as display quality), scoring rules can be introduced. For example:

[0143] Recommended content relevance score: Based on preset user interest tags, the score ranges from 1 to 5;

[0144] Image loading completeness: Calculates the percentage of loaded resources to total resources.

[0145] 4. Verify the output of the results

[0146] The verification result information is output by the system and can be used for real-time monitoring or subsequent analysis. The output forms include:

[0147] Real-time monitoring interface: The verification results are displayed in the monitoring interface in a dynamically updated form, which is convenient for testers to observe and adjust.

[0148] Batch export files: Generate CSV or JSON format files containing the test results of all validation groups for offline analysis.

[0149] Database interface: The verification results are obtained through the database query interface to provide support for the evaluation module.

[0150] 5. Sample Records

[0151] Taking a verification group and content item as an example, the verification result is recorded as follows:

[0152] Verification group: [user click frequency = 30 times / minute, network delay = 50 milliseconds]

[0153] Content item: Recommended video list (contains 5 videos)

[0154] Display quality indicators: resolution 1080p, relevance score 4.5, loading completeness 100%

[0155] User interaction data: click-through rate 25%, page dwell time 120 seconds

[0156] Generation time: 500 milliseconds

[0157] The record shows the specific input parameters of the validation group and the test performance of its generated content items, providing sufficient data support for subsequent evaluation.

[0158] Through the above implementation, the test module can effectively associate the output of the content generation unit with the verification group, and provide support for evaluation and optimization through detailed verification result information, thereby ensuring the test coverage and accuracy of the content generation process.

[0159] Step 5: Evaluate content items

[0160] In this implementation, based on the verification result information output by the test module, the system evaluates the content item to determine whether it meets the expected mobile terminal verification specifications. The specific evaluation process is as follows:

[0161] 1. Evaluate the reception of input

[0162] The test module provides the verification result information to the evaluation module, and the evaluation module receives all key data about the content item, including but not limited to:

[0163] Display quality indicators: such as the relevance score of recommended content, page loading completeness, etc.

[0164] User interaction data: such as click rate, dwell time, number of sliding operations, etc.

[0165] System performance indicators: such as content generation time, resource consumption (such as memory usage, CPU usage, etc.).

[0166] The evaluation module provides basic information for the performance and compliance of content items by parsing and summarizing these verification result data.

[0167] 2. Evaluation criteria setting

[0168] Evaluation specifications are the criteria for judging the quality and performance of content items, which are usually preset by the system and stored in configuration files or databases. The following are examples of common evaluation specifications:

[0169] The relevance score of the recommended content should be greater than or equal to 4.0 (out of 5.0);

[0170] The page loading completeness should be no less than 95%;

[0171] Content generation time should be less than 1000 milliseconds;

[0172] The click-through rate should be no less than 15%.

[0173] During the evaluation process, the evaluation module compares the verification result data with these evaluation specifications one by one.

[0174] 3. Evaluation process of content items

[0175] The evaluation module checks the verification result data item by item and determines whether the content item has passed the verification according to the set evaluation specifications. The specific process is as follows:

[0176] Single evaluation: For each verification result indicator, compare it with the corresponding evaluation specification. For example:

[0177] If the relevance score of the recommended content is 3.8, which is lower than the set 4.0, this item will be judged as unqualified;

[0178] If the page loading completeness is 98%, which is higher than the set 95%, this item is judged as qualified.

[0179] Comprehensive evaluation: Summarize all individual evaluation results and determine the final evaluation result according to system rules. For example, if any key indicator fails, the overall evaluation will be considered as failed.

[0180] 4. Mark content items that have not passed verification

[0181] If the evaluation module determines that a content item has failed verification, it is flagged and the following information is recorded:

[0182] Content item identification: such as content type, generation time, associated verification group, etc.;

[0183] Indicators and reasons for failure: such as "recommendation content relevance score is too low" or "generation time exceeds the specified range";

[0184] Recommended adjustment measures: such as optimizing the recommendation algorithm weight or adjusting resource loading priority.

[0185] The tag information will be stored in the system log or database for subsequent analysis and optimization.

[0186] 5. Evaluation Examples

[0187] For a content item generated for a verification group (user click frequency = 30 times / minute, network delay = 50 milliseconds), assume that the verification result is:

[0188] Recommended content relevance score: 3.5;

[0189] Page loading completeness: 97%;

[0190] Generation time: 1200 milliseconds.

[0191] The system evaluation results are as follows:

[0192] The relevance score of the recommended content is unqualified (less than 4.0);

[0193] The page loading completeness is qualified (above 95%);

[0194] The generation time is unqualified (greater than 1000 milliseconds).

[0195] After comprehensive judgment, the content item failed the verification and generated the marking information:

[0196] Content item identifier: recommended video list, verification group: [click frequency = 30 times / minute, delay = 50 milliseconds]

[0197] Reason for failure: relevance score is too low (3.5<4.0); generation time exceeds the limit (1200ms>1000ms)

[0198] Suggested measures: Optimize the recommendation algorithm to improve relevance; optimize the loading process to shorten the generation time.

[0199] Through the above-mentioned evaluation process, this implementation plan can accurately determine the compliance of content items, providing a reliable basis for subsequent optimization and adjustment.

[0200] Step 6: Refresh the template specification structure

[0201] In this implementation, for content items marked as verification failure in the evaluation results, the system refreshes and adjusts the template specification structure based on its associated verification group and generation logic. The specific process is as follows:

[0202] 1. Analysis of tagged content items

[0203] The system first obtains the content items marked as verification failures in the evaluation results and their related information. The analysis of content items includes the following aspects:

[0204] Associated verification group: Extract the source verification group information of the failed verification content item, including each variable and its specific value. For example, "user click frequency = 30 times / minute, network delay = 50 milliseconds".

[0205] Failure reasons: Based on the tag information provided by the evaluation module, identify the key reasons that led to the validation failure, such as insufficient content relevance score, excessive generation time, or insufficient page load completeness.

[0206] Scope of impact: Analyze the scope of the failure cause to determine whether it is limited to a specific verification group or has a broader impact.

[0207] 2. Update of template specification structure

[0208] Based on the analysis results, the template specification structure is refreshed. The template specification structure is the core framework of the content generation logic, including modules such as recommendation algorithms, page layout rules, and dynamic adjustment strategies. The refresh process may include the following:

[0209] Recommendation algorithm weight adjustment

[0210] If it detects that certain combinations of variables result in a low relevance score for the recommended content, the system will adjust the weights of the recommendation algorithm. For example:

[0211] When the user click frequency is low, the recommendation algorithm needs to lower the priority of high-profile content and increase the weight of long-tail content.

[0212] When network latency is high, the recommendation algorithm needs to prioritize content that takes up less resources.

[0213] Example of the adjusted recommendation algorithm weight formula (assuming that the recommendation score is determined by content popularity and user interest):

[0214] S=w1·H+w2·I

[0215] Among them, S represents the content relevance score, H represents the content popularity, I represents the user interest, and w1 and w2 are weight factors. If the evaluation result shows that the current content popularity weight is too high, the system will reduce the value of w1 and increase w2.

[0216] Page layout rule optimization

[0217] If certain layout rules result in poor display of content items (such as insufficient loading completeness or low user interaction rate), the system will adjust the page layout rules. For example:

[0218] Reduce the amount of content displayed on each screen to reduce the pressure of loading resources;

[0219] Adjust the order of content modules and place content that users are most interested in on the first screen.

[0220] The adjusted layout rules can be implemented through parameterized configuration, for example:

[0221] Layout parameters:

[0222] {"Number of modules":3,"Priority module":"Recommended videos","Second priority module":"News content"}

[0223] Dynamic adjustment strategy improvements

[0224] Optimize the dynamic adjustment strategy for real-time generation logic. For example:

[0225] If the generation time exceeds the system requirements, add an asynchronous processing mechanism in the generation logic to load the core content in advance;

[0226] If the loading is not complete enough, the resources in the visible area will be loaded dynamically first.

[0227] 3. Refresh the template feedback loop

[0228] The refreshed template specification structure will be reintroduced into the testing process to form an optimization feedback loop. The specific steps include:

[0229] Generate new test data: Generate new content items using the optimized template specification structure and record their verification groups and test results.

[0230] Verify the improvement: Compare the verification results of the content items before and after the update to determine whether the template update is effective. For example, if the relevance score is improved from 3.5 to 4.2 and the loading time is reduced from 1200 milliseconds to 800 milliseconds after optimization, the update effect is considered significant.

[0231] 4. Sample refresh scenario

[0232] Assume that the evaluation result of a content item is as follows:

[0233] Verification group: [user click frequency = 30 times / minute, network delay = 50 milliseconds]

[0234] Reason for failure: The recommendation content relevance score is low (3.5<4.0); the generation time is too long (1200ms>1000ms).

[0235] Based on the above information, the system performs the following refresh operations:

[0236] Adjust the weight of the recommendation algorithm: reduce the weight of content popularity and increase the weight of user interest;

[0237] Optimize page layout rules: reduce the amount of content per screen from 5 to 3;

[0238] Improve dynamic adjustment strategy: add asynchronous loading mechanism and prioritize loading core content.

[0239] The refreshed template specification structure is as follows:

[0240] Template specification structure:

[0241] -Recommendation algorithm weight: {"heat weight":0.4,"interest weight":0.6}

[0242] -Page layout rules: {"Number of screen contents":3,"Priority loading area":"Top module"}

[0243] -Dynamically adjust strategy: {"asynchronous loading": true,"priority resources":"core content"}

[0244] Through the above process, this implementation scheme realizes efficient refreshing of the template specification structure, can effectively solve the problem of content items that fail verification, further optimize the content generation logic and improve the system adaptability.

[0245] The above description is only a preferred embodiment of the present invention, so all equivalent changes or modifications made according to the structure, characteristics and principles described in the scope of the patent application of the present invention are included in the scope of the patent application of the present invention.

Claims

1. A mobile device automated testing method based on variable optimization, characterized in that: The automated testing method for the mobile device comprises the following steps: Step 1: obtaining an initial range of each variable of a plurality of variables in a variable set, wherein the variable set defines a mobile terminal content feature generated by a real-time content generation unit; and generating a plurality of verification values ​​for each variable in the variable set according to its initial range; Step 2: Based on historical test data and machine learning algorithms, multiple verification groups are automatically optimized and generated, each verification group contains different verification value combinations of multiple variables in the variable set; multiple optimized verification groups are provided to the mobile terminal real-time content generation unit; Step 3: In the real-time content generation unit executed by the mobile device, for each optimized verification group, according to the verification value combination of the verification group as the input of the template specification structure, the mobile content item is dynamically generated; Step 4: in the test module executed by the mobile device, for each mobile terminal content item output by the real-time content generation unit, associating it with the verification value of the verification group that generates the content item to obtain relevant verification result information; outputting the verification result information including the mobile terminal content item and its associated verification value; Step 5: receiving evaluation inputs for the content items according to the relevant verification result information of each mobile terminal content item output by the real-time content generation unit; and determining that the evaluation input of at least one content item indicates that it does not meet the mobile terminal verification specification; Step 6: Based on the evaluation input indicating the mobile verification failure, refresh the template specification structure to optimize the content generation logic.

2. The mobile device automated testing method based on variable optimization as claimed in claim 1, characterized in that: The variable set includes user behavior data variables, external environment parameter variables and built-in parameter variables of content generation logic.

3. A mobile device automated testing method based on variable optimization as claimed in claim 1 or 2, characterized in that: The generation rule of the verification value includes a uniform distribution generation method, a random distribution generation method, an extreme value generation method or a distribution function generation method.

4. The mobile device automated testing method based on variable optimization as claimed in claim 1, characterized in that: The process of optimizing the validation group based on the machine learning algorithm in step 2 includes: Extract the eigenvector of the validation group variables; Training classification models based on historical test data; The importance of each variable on the quality of content generation is calculated based on the classification model, and the verification group is screened.

5. The mobile device automated testing method based on variable optimization as claimed in claim 1, characterized in that: The step 3 further comprises: Parse the validation group input parameters and map them to the corresponding parameters of the content generation logic; Load the template specification structure, Call the recommendation algorithm to generate recommended content, dynamically adjust the resource loading strategy according to network-related parameters, and generate content items in combination with the page layout module; Dynamically optimize content generation logic based on real-time user responses or external data changes; The data associated with the content item and the validation set is recorded and stored for subsequent testing and evaluation.

6. The mobile device automated testing method based on variable optimization as claimed in claim 5, characterized in that: The template specification structure includes a page layout module, a recommendation algorithm module and a dynamic rule adjustment module.

7. The mobile device automated testing method based on variable optimization as claimed in claim 1, characterized in that: The verification result information of step 4 includes the display quality index of the content item, user interaction data and content generation time.

8. The mobile device automated testing method based on variable optimization as claimed in claim 1, characterized in that: The step 5 further includes comparing the verification result information item by item based on the preset evaluation specification to determine whether the content item meets the quality and performance requirements; If the content item fails the verification, the failed content item is marked, and the content item identifier, the failed indicator and the reason are recorded; Generates recommended adjustments for failed content items; Store the tag information in logs or databases for subsequent analysis and optimization.

9. The mobile device automated testing method based on variable optimization as claimed in claim 1, characterized in that: Refresh the template specification structure according to the analysis results, including adjusting the recommendation algorithm weights, optimizing page layout rules, and improving dynamic adjustment strategies.

10. The mobile device automated testing method based on variable optimization as claimed in claim 9, characterized in that: Reintroduce the refreshed template specification structure into the test process, generate new test data and record the verification results to form an optimization feedback loop.

Images