Software user experience evaluation method and system based on comprehensive data analysis

Abstract

The invention discloses a software user experience evaluation method and system based on comprehensive data analysis, and relates to the technical field of man-machine interaction. The method comprises the steps that a behavior perception engine is embedded in a software operation interface, and the number of use times, the number of jump operation times, the number of flow standard steps and the total number of access users of a user in a task path are collected; and further performing unified normalization processing on the background server, constructing a normalized behavior data set, and calculating a path deviation joint index PDI. On the basis of a path deviation joint index PDI, two dimensions of a path selection information entropy and a path jump rate mean value are considered in a joint manner, so that whether concentrated deviation or process jump behaviors exist in user path use can be effectively identified, and then whether problems of forced guidance, process hiding shortcut and the like exist in system path design can be judged. Compared with a traditional evaluation mode which only depends on clicking a thermodynamic diagram or a jump rate, the method has the advantage that higher-dimension path behavior modeling and deviation recognition capability are realized.

Description

Technical Field

[0001] The present invention relates to the field of human-computer interaction technology, and in particular to a method and system for evaluating software user experience based on comprehensive data analysis. Background Art

[0002] The present invention belongs to the technical field of human-computer interaction and intelligent software evaluation, and in particular relates to a method for quantitatively evaluating software user experience that integrates user behavior collection, path selection modeling, and multi-dimensional risk analysis. With the increasing complexity of software system functions, especially in multi-process, multi-path operation scenarios, such as approval flow systems, online questionnaire systems, e-commerce ordering systems, etc., when users complete a task goal, there are often multiple feasible paths and different interaction methods. Accurate and dynamic data modeling of the selection and use behavior of these interaction paths is the key to achieving software interaction optimization and intelligent intervention in user experience.

[0003] However, currently, existing software user experience evaluation methods are mostly limited to post-event surveys, such as questionnaires, ratings, coarse-grained metrics like click-through rate and page dwell time, or heatmap visualizations. These methods lack the ability to model path-level deviations and identify logical structures for interaction anomalies. Especially in software with multiple completion paths, users are often strongly guided by certain default recommended paths or system-prompted processes, resulting in extremely concentrated path usage while other paths are rarely used. This leads to unbalanced data feedback and makes hidden issues difficult to detect. Furthermore, user behavior such as skipping steps within a path, reworking input, or making errors is difficult to accurately categorize and respond to, forcing the system to make delayed or crude adjustments later, impacting interaction effectiveness.

[0004] This imbalance in path design inherently leads to a perceived lack of freedom for users. While users appear to have multiple options, they are actually being coerced or misled by the system design. This ultimately leads to overuse of a single primary path, while other paths, lacking usage data, are difficult to effectively optimize or identify errors. Furthermore, because the system cannot promptly detect user behavior such as frequent step skipping, repeated rework, or accidental touches, this can lead to negative user experience metrics such as increased cognitive load, increased task interruption rates, and abnormally shortened page dwell time. This not only reduces user satisfaction but also impacts subsequent system recommendations, model optimization, and version iteration decisions during data accumulation, trapping the entire interactive system in a vicious cycle of structural experience bias. Summary of the Invention

[0005] In view of the deficiencies in the prior art, the present invention provides a method and system for evaluating software user experience based on comprehensive data analysis, which solves the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: comprising the following steps: S1, by embedding a behavior perception engine in the software operation interface, collecting user experience behavior data in real time and transmitting the user experience behavior data to the software background server; S2. Preprocess the user experience behavior data in the software backend server to obtain a normalized behavior data set, build a behavior database, and store the normalized behavior data set; S3. Extract the normalized behavior data set, calculate the output path deviation joint index (PDI), and perform preliminary comparative evaluation on the preset deviation interval threshold and the path deviation joint index (PDI); S4. Trigger the risk identification mechanism based on the preliminary comparative assessment results, extract the path risk data, and calculate and output the behavioral risk indicator PRLV based on the path risk data; S5. Based on the path deviation joint index PDI and the behavioral risk index PRLV, a comprehensive calculation is performed to analyze the path experience risk total index RRDV, and the path risk interval threshold is set to conduct a secondary comparative evaluation with the path experience risk total index RRDV, and intervention measures are generated.

[0007] Preferably, said S1 includes S11 and S12; S11. By embedding a behavior perception engine in the software's operating interface, the user automatically starts the behavior perception engine after logging into the software and authorizing behavior perception, and collects user experience behavior data in real time. By setting up an API collection application program interface, all user experience behavior data collected by the behavior perception engine are uniformly collected in a log collector; The behavior perception engine includes a path usage count engine, a non-linear jump operation engine, a path standard step number acquirer, and a server engine; The user experience behavior data includes the number of times the i-th path is used U i , the number of nonlinear jump operations J of the i-th path i , the total number of normal steps T of the i-th path i and the total number of users N; The total number of users N is collected in real time by the server engine at the back end of the software; S12. Set up a wireless communication interface of the software backend server and a 5G wireless communication network, wirelessly connect the wireless communication interface to the API collection application program interface, access the log collector of the software backend in real time, and extract real-time user experience behavior data.

[0008] Preferably, said S2 includes S21 and S22; S21. Receive user experience behavior data in real time in the software backend server, and pre-process the user experience behavior data in the software backend server to obtain a normalized behavior data set; The preprocessing includes timestamp verification and data normalization; The timestamp verification is performed by unifying the user experience behavior data into a Unix timestamp; The data normalization is performed by using a proportional normalization method on the user experience behavior data after timestamp verification, calculating the difference between the original value and the lower limit value of each parameter in the user experience behavior data, then calculating the difference between the upper limit value and the lower limit value, and performing a ratio calculation based on the two differences, thereby normalizing all parameters in the user experience behavior data to the interval [0, 1], thereby eliminating the dimensional influence of all parameters; S22. Configure an original behavior event table and a processed data table in the behavior database, write the normalized behavior data set obtained after preprocessing into the behavior database, store it in the processed data table, and write the user experience behavior data into the original behavior event table for data storage.

[0009] Preferably, said S3 includes S31 and S32; S31. Extracting a normalized behavior data set from the behavior database, calculating and outputting a path deviation joint index (PDI), and measuring the concentration and path deviation of the user's path selection behavior; The path deviation combined index PDI is calculated and outputted by the following algorithm formula: ; Where n represents the total number of paths and log represents the logarithmic function.

[0010] Preferably, in step S32, the historical path deviation combined index (PDI) is calculated and averaged, and the 85th and 20th percentiles are extracted as deviation interval thresholds. The deviation interval thresholds include an upper deviation threshold F1 and a lower deviation threshold F2. The deviation interval thresholds are preliminarily compared and evaluated with the real-time acquired path deviation combined index (PDI) to determine the user's path deviation. The specific evaluation content is as follows; When the path deviation index PDI is greater than the upper threshold F1, it indicates that the access path deviation is abnormal, and the risk identification mechanism is triggered. When the deviation lower threshold F2 is less than the path deviation joint index PDI and less than the deviation upper threshold F1, it indicates that the current user access path has deviated. In this case, the acquisition frame rate of the behavior perception engine is increased by 50%. When the path deviation index PDI ≤ the deviation lower limit threshold F2, it means that the current software user access experience is stable and the path distribution is balanced. At this time, no intervention is performed.

[0011] Preferably, said S4 includes S41 and S42; S41. After the initial comparative assessment triggers the risk identification mechanism, the software system operation log API is integrated with the software backend server to collect and obtain path risk data from the software system operation log in real time. The path risk data is then normalized to eliminate the dimensionality effect of the path risk data. The path risk data includes the total number of interruption behaviors Zd on the i-th path i , the number of rework behaviors on path i Fg i , the number of wrong operations on the i-th path W i and the average operation time D of the user in path i i .

[0012] Preferably, S42, performing ratio calculation based on the extracted path risk data, outputting a behavior risk index PRLV, measuring the risk status of the path when the current user deviates abnormally when accessing the software path; The behavior risk indicator PRLV is calculated and output by the following algorithm formula: ; Where, PRLV i The risk indicator of the i-th path.

[0013] Preferably, said S5 includes S51 and S52; S51. Based on the behavioral risk indicator PRLV of all paths and the path deviation joint index PDI, a comprehensive calculation is performed to output the path experience risk total index RRDV, which measures the behavioral risk of the software in the event of path deviation. The total path experience risk index RRDV is calculated and output by the following algorithm formula: .

[0014] Preferably, S52, based on the calculated total path experience risk index RRDV in all historical periods, the mean and standard deviation are calculated to set the path risk interval threshold, wherein the path risk interval threshold includes an upper risk threshold R1 and a lower risk threshold R2. The upper risk threshold R1 is set based on the mean plus 75 percent of the standard deviation; The lower risk threshold R2 is set based on the mean plus 35 percent of the standard deviation; The path risk interval threshold is compared with the real-time path experience risk index (RRDV) to evaluate the risk status of the user interaction process, classify the risk level, and implement corresponding intervention measures. The specific evaluation content is as follows: When the total path experience risk index RRDV is greater than the upper risk threshold R1, the entire path of the current software is classified as a level 1 risk, and the first intervention measure is executed. The first intervention measure disables the current software path and prompts the software path to be reconstructed. When the lower risk threshold R2 is less than the total path experience risk index RRDV and less than the upper risk threshold R1, the overall path risk of the current software is classified as a secondary risk and the second intervention measure is implemented; The second intervention measure automatically identifies mis-touch areas by collecting click heat maps of the software operation interface and software system operation logs, adjusts the spatial layout, and prioritizes moving mis-touch controls forward and presenting them in steps. When the total path experience risk index RRDV ≤ the lower risk threshold R2, the overall path of the current software is divided into three levels of risk and no intervention measures are required.

[0015] A software user experience evaluation system based on comprehensive data analysis, including an experience behavior collection module, a collection data processing module, a path deviation analysis module, a behavior risk analysis module and a comprehensive experience risk evaluation module; The experience behavior collection module collects user experience behavior data in real time by embedding a behavior perception engine in the software operation interface, and transmits the user experience behavior data to the software background server; The data collection processing module pre-processes the user experience behavior data in the software background server to obtain a normalized behavior data set, and builds a behavior database to store the normalized behavior data set; The path deviation analysis module extracts the normalized behavior data set, calculates and outputs the path deviation joint index PDI, and performs preliminary comparative evaluation on the preset deviation interval threshold and the path deviation joint index PDI; The behavior risk analysis module triggers a risk identification mechanism based on preliminary comparative assessment results, extracts path risk data, and calculates and outputs a behavior risk indicator PRLV based on the path risk data; The comprehensive experience risk assessment module performs comprehensive calculations based on the path deviation joint index PDI and the behavioral risk index PRLV, analyzes the path experience risk total index RRDV, sets the path risk interval threshold and conducts a secondary comparative assessment with the path experience risk total index RRDV, and generates intervention measures.

[0016] The present invention provides a method and system for evaluating software user experience based on comprehensive data analysis. It has the following beneficial effects: (1) This method embeds a behavior perception engine in the software operation interface to collect the number of times users use the task path, the number of jump operations, the number of standard steps in the process, and the total number of visiting users. It then performs unified normalization processing on the background server to construct a normalized behavior data set and calculate the path deviation joint index (PDI). Based on the path deviation joint index (PDI), which jointly considers the two dimensions of path selection information entropy and path jump rate mean, it can effectively identify whether there is concentrated deviation or process jump behavior in user path usage, and then determine whether there are problems such as forced guidance and hidden shortcuts in the process in the system path design. Compared with the traditional evaluation method that only relies on click heat maps or bounce rates, this method achieves a higher-dimensional path behavior modeling and deviation identification capability.

[0017] (2) After the path deviation index PDI exceeds the preset threshold, this method further activates the risk identification mechanism, collects risk factors such as the number of interruption behaviors, the number of rework behaviors, the number of erroneous operations, and the average operation time in the path, and realizes the quantitative assessment of the risk level of each path by constructing the behavioral risk index PRLV. The behavioral risk index PRLV formula comprehensively considers the behavior density and the operation time compensation effect. While realizing the expression of the behavioral risk density per unit time, it has the adaptive detection capability of high-frequency rework and erroneous touch paths. The total path experience risk index RRDV calculated by further combining the behavioral risk index PRLV with the path deviation index PDI can not only form a unified risk stratification for all paths in the software, but also provide hierarchical decision support for subsequent interaction structure adjustment and control-level intervention, solving the problems of fuzzy risk identification and uncontrollable response strategies in existing technologies.

[0018] (3) This method sets a path risk interval threshold and dynamically compares the path experience risk index (RRDV) with historical behavior data, thereby achieving a three-level classification of risk status during user interaction. Among them, paths above the upper risk threshold R1 will trigger the first intervention strategy, including path disabling and structural reconstruction suggestions; paths between the path risk interval thresholds will trigger the second intervention strategy, such as spatial layout adjustment based on heat map, control rearrangement and operation guidance optimization; paths below the lower risk threshold R2 do not require intervention and maintain the current structure. The above intervention measures are all generated based on real-time risk assessment and have adaptive, closed-loop, and multi-level intervention characteristics. They can effectively improve the user experience performance of software systems in complex multi-path scenarios, reduce false touch rates, interruption rates, and rework rates, and significantly enhance system version optimization capabilities and user satisfaction. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 This is a schematic diagram of the steps of a software user experience evaluation method based on comprehensive data analysis of the present invention; Figure 2This is a flow chart of a software user experience evaluation system based on comprehensive data analysis according to the present invention; Figure 3 Schematic diagram of the PRLV risk value distribution of risk indicators for each path. DETAILED DESCRIPTION

[0020] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0021] Example 1: Please refer to Figure 1 and Figure 3 The present invention provides a method for evaluating software user experience based on comprehensive data analysis. To achieve the above objectives, the present invention is implemented through the following technical solutions: comprising the following steps: S1, by embedding a behavior perception engine in the software operation interface, collecting user experience behavior data in real time and transmitting the user experience behavior data to the software background server; S2. Preprocess the user experience behavior data in the software backend server to obtain a normalized behavior data set, build a behavior database, and store the normalized behavior data set; S3. Extract the normalized behavior data set, calculate the output path deviation joint index (PDI), and perform preliminary comparative evaluation on the preset deviation interval threshold and the path deviation joint index (PDI); S4. Trigger the risk identification mechanism based on the preliminary comparative assessment results, extract the path risk data, and calculate and output the behavioral risk indicator PRLV based on the path risk data; S5. Based on the path deviation joint index PDI and the behavioral risk index PRLV, a comprehensive calculation is performed to analyze the path experience risk total index RRDV, and the path risk interval threshold is set to conduct a secondary comparative evaluation with the path experience risk total index RRDV, and intervention measures are generated.

[0022] In this embodiment, the method embeds a multi-type behavior perception engine in the software operation interface, collects key behavior parameters including the number of path uses, nonlinear jumps, and task completion status in real time, and uploads the data to the server side through the 5G communication interface to achieve real-time capture of user interactive behavior. The original data is then time-synchronized and normalized to build a behavior database in a unified format to provide a high-quality data basis for subsequent analysis. Furthermore, by constructing a path deviation joint index PDI, the concentration degree and jump rate changes of the path selection distribution are accurately portrayed, and it is identified whether there are guidance deviations or path shortcut problems in the software process design. When the path deviation joint index PDI index exceeds the deviation threshold, the risk identification mechanism is triggered, path risk data is collected, and a behavior risk index PRLV is calculated to quantify the risk level of each path. Finally, the path deviation joint index PDI and the risk index PRLV are comprehensively analyzed to output the path experience risk total index RRDV, and according to its value falling in different risk intervals, a hierarchical intervention strategy is automatically generated, including path reconstruction suggestions, control forward movement, process disassembly, visual feedback optimization, etc. By implementing this method, it is possible to dynamically identify, respond in real time, and adaptively optimize potential user experience risks in multi-path processes without interrupting user task execution. Compared to traditional user experience evaluation methods that rely on static heat maps or questionnaire scoring, this invention can detect process design flaws earlier and more accurately, improving the software's ability to automatically identify and handle abnormal path behavior. This significantly enhances the software's path elasticity design capabilities, user behavior fault tolerance, and overall experience stability, ultimately achieving the technical effects of increasing task completion rates, reducing rework and misoperation, and improving user satisfaction and system reliability.

[0023] Example 2: Please refer to Figure 1 ,Specifically: S1 includes S11 and S12; S11. By embedding a behavior perception engine in the software's operating interface, the user automatically starts the behavior perception engine after logging into the software and authorizing behavior perception, and collects user experience behavior data in real time. By setting up an API collection application program interface, all user experience behavior data collected by the behavior perception engine are uniformly collected in a log collector; The behavior perception engine includes a path usage count engine, a nonlinear jump operation engine, a path standard step count acquirer, and a server engine; User experience behavior data includes the number of times the i-th path is used U i , the number of nonlinear jump operations J of the i-th path i , the total number of normal steps T of the i-th path i and the total number of users N; The number of times the i-th path is used U iBy embedding the path usage count engine in the task completion button and submission event in the software's operation interface, we can collect information about which path the user has selected and whether the path task has been successfully completed. The number of nonlinear jump operations J on the i-th path i By embedding a non-linear jump operation engine on each step control in the software's operation interface, the click sequence queue is recorded in real time; The total number of normal steps T for path i i Configure the standard path process through the software backend, use the backend path standard step number obtainer to load the user access path process, extract the normal step process, and register the defined standard process node when each path enters; The total number of users N is collected in real time through the server engine at the back end of the software; S12. Set up a wireless communication interface of the software backend server and a 5G wireless communication network, wirelessly connect the wireless communication interface to the API collection application program interface, access the log collector of the software backend in real time, and extract real-time user experience behavior data.

[0024] In this embodiment, the method embeds multiple types of behavior perception engines within the software interface, building a front-end and back-end linkage mechanism capable of real-time collection, transmission, and management of user experience behavior data. Specifically, the path usage count engine records the frequency of task completion for each path, the nonlinear jump operation engine tracks step jumps and path detours during user interaction, the path standard step count acquirer provides a system-preset sequence of process nodes, and the server-side engine dynamically counts the total number of currently accessing users, forming the basic parameter set required for path behavior modeling. Data collected by all engines is centrally transmitted to a log collector via a configured API collection interface. Furthermore, relying on the 5G wireless communication network, high-speed, low-latency data upload is achieved through the software backend's wireless communication interface, ensuring the integrity and timeliness of collected behavior. Through the above implementation methods, not only is high-precision, low-interference, full-link collection of user path selection and behavior characteristics achieved, but the real-time and uniformity of data collection is also ensured, providing a complete, continuous, and structured data foundation for subsequent behavior modeling and path deviation analysis. Compared with existing technologies, this method can effectively circumvent problems such as traditional log lag, incomplete sampling, and behavior omissions, significantly improving the accuracy and availability of user behavior data. Ultimately, this approach provides underlying support for building a highly reliable, multi-path dynamically optimized software process evaluation mechanism, significantly enhancing the system's automated perception and real-time response capabilities in user experience feedback, and providing a key foundation for data-driven UI structure improvement and process redesign.

[0025] Example 3: Please refer to Figure 1, specifically: S2 includes S21 and S22; S21. Receive user experience behavior data in real time in the software backend server, and pre-process the user experience behavior data in the software backend server to obtain a normalized behavior data set; Preprocessing includes timestamp verification and data normalization; Timestamp verification is achieved by unifying user experience behavior data into Unix timestamps; Data normalization uses a proportional normalization method on the user experience behavior data after timestamp verification. The original value of each parameter in the user experience behavior data is calculated as the difference between the lower limit value, the upper limit value and the lower limit value, and the ratio of the two differences is calculated. All parameters in the user experience behavior data are normalized to the [0, 1] interval to eliminate the dimension effects of all parameters. S22. Configure an original behavior event table and a processed data table in the behavior database, write the normalized behavior data set obtained after preprocessing into the behavior database, store it in the processed data table, and write the user experience behavior data into the original behavior event table for data storage.

[0026] In this embodiment, the method realizes the standardized preprocessing and structured storage management of user experience behavior data by establishing a two-layer data processing mechanism in the software background server. Specifically, by unifying the timestamp format, all user experience behavior data are standardized into Unix timestamps to ensure that the data collection results under different terminals and different operating system environments have consistent time series characteristics; then the proportional normalization method is used to perform dimensionless processing on various behavior parameters, such as the number of path usages and jump frequencies, and their numerical ranges are uniformly mapped to the [0,1] interval, effectively eliminating the dimensional differences between the parameters and improving the accuracy and comparability of subsequent model calculations. By configuring the original behavior event table and the processed data table respectively in the behavior database, a hierarchical storage structure is established, so that the system can quickly extract and analyze the normalized data set while retaining the traceability of the original data, taking into account both data integrity and analysis efficiency. The introduction of this implementation method significantly improves the processing capabilities of large-scale, multi-dimensional user behavior data, ensures the high quality and uniformity of the algorithm input data, and provides a solid foundation for the subsequent accurate calculation of the path deviation index PDI and the risk level indicator PRLV. Compared with existing data processing methods, this solution effectively reduces data redundancy and information bias through normalization calibration and structured warehousing, improves the model's sensitivity to abnormal behavior recognition, and ultimately enhances the system's data analysis accuracy and processing stability in software user experience evaluation, promoting a seamless connection between user behavior perception and policy intervention.

[0027] Example 4: Please refer to Figure 1,Specifically: S3 includes S31 and S32; S31. Extracting a normalized behavior data set from the behavior database, calculating and outputting a path deviation joint index (PDI), and measuring the concentration and path deviation of the user's path selection behavior; The path deviation index PDI is calculated and output by the following algorithm formula: ; Where n represents the total number of paths, and log represents the logarithmic function; The path selection entropy is the normalized form of information entropy, which measures the degree of distribution balance when users choose paths. A high entropy close to 1 indicates that the paths are used evenly and the distribution is balanced; a low entropy close to 0 indicates that most users are concentrated on a very small number of paths, which is severely skewed. represents the average jump rate, The jump rate of each path reflects whether the user performs the task according to the system design process. A higher value indicates more interaction jumps in the path, which may be due to users getting lost, complex interfaces, unclear guidance, etc. The average value represents the user process compliance at the system level. The product of the two composite indicators reflects whether the user's path selection behavior is concentrated and whether the user's behavior deviates within the path. In essence, it is a quantitative expression of the deviation of the path structure design. The Path Deviation Index (PDI) is used to assess whether users are focused on selecting multiple paths in a task process and whether they frequently deviate from the system's preset standard process. In essence, it is used to determine whether the process structure is reasonable and the guidance is clear. Path selection entropy is a standard normalized information entropy that measures selection concentration. The unit used is "bit", but because it is a normalized ratio, the result is a unitless number in the interval [0,1].

[0028] The jump rate mean is a nonlinear jump rate with dimensionless ratio, which measures the deviation of behavior.

[0029] The product of the two is a dimensionless number, which avoids the problem of superposition of different dimensions.

[0030] S32. Calculate the average of historical path deviation combined indexes (PDIs). The 85th and 20th percentiles are set as deviation thresholds. The deviation thresholds include an upper deviation threshold F1 and a lower deviation threshold F2. A preliminary comparison is performed between the deviation thresholds and the real-time PDIs to determine the user's path deviation. The specific evaluation details are as follows. When the path deviation index PDI is greater than the upper threshold F1, it indicates that the access path deviation is abnormal, the path selection is overly concentrated, and the operation jumps significantly. At this time, the risk identification mechanism is triggered; When the deviation lower threshold F2 is less than the path deviation joint index PDI and less than the deviation upper threshold F1, it indicates that the current user access path has deviated. In this case, the acquisition frame rate of the behavior perception engine is increased by 50%. When the path deviation index PDI ≤ the deviation lower limit threshold F2, it means that the current software user access experience is stable and the path distribution is balanced. At this time, no intervention is performed.

[0031] In this embodiment, the method establishes a structural deviation analysis method for user path selection behavior by constructing a calculation mechanism for the path deviation index (PDI). Specifically, by extracting normalized data from a behavioral database and constructing a joint calculation model using information entropy and jump rate, this method not only measures the degree of balance in user path distribution but also incorporates the deviation of interactive behavior within a path. This allows the system to not only identify the central tendency of path selection but also quantify whether the internal process of a path is frequently skipped, thereby achieving a comprehensive assessment of the rationality of path design and the consistency of user interaction. Furthermore, a historical behavior statistical model is introduced to dynamically generate deviation threshold intervals, where the deviation interval thresholds include an upper deviation threshold F1 and a lower deviation threshold F2. The real-time path deviation index (PDI) value is compared with this interval to determine the stability, deviation trend, or abnormal concentration of the current user path behavior, ultimately driving the acquisition density adjustment or risk identification mechanism. The introduction of this implementation significantly improves the ability to identify path deviation patterns and effectively addresses the technical blind spot of traditional user behavior analysis methods that only consider click counts without considering the path structure and behavioral characteristics. By converting path behavior into a quantifiable structural deviation index and combining it with a dynamic threshold judgment strategy, this method provides systematic early warning of potential process bottlenecks, hidden shortcuts, or forced design. Compared to existing static indicator systems, this method offers greater real-time performance, sensitivity, and depth of structural analysis. It provides a proactive foundation for subsequent risk level identification and intervention strategy generation, and overall improves the quality of user process design and the transparency of interactive behaviors within the software system.

[0032] Example 5: Please refer to Figure 1 , specifically: S4 includes S41 and S42; S41. After the initial comparative assessment triggers the risk identification mechanism, the software system operation log API is integrated with the software backend server to collect and obtain path risk data from the software system operation log in real time. The path risk data is then normalized to eliminate the dimensionality effect of the path risk data. Path risk data includes the total number of interruption behaviors on path i, Zd i , the number of rework behaviors on path i Fg i , the number of wrong operations on the i-th path W i and the average operation time D of the user in path i i ; The total number of interruption behaviors on path i Zd i By identifying whether the user completes the target page, if there is no end event and the session ends, it is interrupted; Number of rework behaviors on path i Fg i When the user clicks the previous step or modifies an entered field, an update event is triggered, which records the differences before and after the modification. Rework can be refined into repeated modifications, submission failure rollback, etc. Number of wrong operations on path i W i Collect DOM element attributes of click events from software system operation logs to determine whether they are accidental touches. This can include clicking non-interactive elements, repeatedly clicking the submit button, clicking Cancel immediately after clicking, etc. You can also collect information based on the triggering frequency of error message pop-up windows. The average operation time of users on path i is D i When each user accesses path i, the software system operation log records the timestamps of the start node and the completion node. The difference between the two is the time consumed by the user, and the average of path i can be taken.

[0033] S42. Calculate a ratio based on the extracted path risk data and output a behavior risk indicator PRLV to measure the risk status of the path when the current user deviates from the software path during access. The behavioral risk indicator PRLV is calculated and output using the following algorithm formula; ; Where, PRLV i Risk indicator of path i; The numerator is represented by the sum of the user's negative interaction behaviors on path i, that is, the weighted set of risk factors. The higher the value, the more serious the potential risk of the path. The denominator introduces an operation time factor for two purposes: Compensation effect: If an operation takes a long time, it indicates that the user is willing to invest time, which may partially buffer the negative effects of high-risk behavior; Time pressure penalty mechanism: If an operation takes an unusually short time, it may be used carelessly and there is a high risk of misuse; The risk indicator PRLV is a behavioral risk density per unit time, which can also be understood as the average number of abnormal interactive behaviors generated per second. It measures the frequency and density of "interactive risk behaviors" occurring during user use of a specific path, such as interruptions, rework, and incorrect operations. Furthermore, combined with the operation time, the risk density per unit time is: The formula of the risk indicator PRLV is in proportional form. At the same time, the parameters involved are all parameters after data normalization processing. The output result of the risk indicator PRLV is dimensionless.

[0034] In this embodiment, the method further enhances the ability to conduct refined analysis after path deviation behavior by establishing a risk identification and level quantification mechanism based on behavioral logs. Through the software system operation log API and the backend integration connection, core risk factor data such as path interruption, rework, misoperation and average time consumption are extracted in real time, and after unified normalization processing, a structured path risk data set is formed. This process realizes the systematic capture and standardized expression of users' atypical operation behaviors, providing an accurate behavioral quantification basis for subsequent risk calculations. In constructing the behavioral risk indicator PRLV based on the above data, the ratio calculation method of "the total amount of negative behavior divided by the average time consumption of the user" is adopted, and the time density dimension is integrated. While identifying the frequency of abnormal user operations, the "time compensation effect" is introduced, taking into account the dual dimensions of behavioral intensity and operation rhythm. Through this implementation method, the system can automatically enter the risk identification state after the user path deviation is detected, and quickly complete the quantitative assessment of the path risk level, ensuring that the problem path can be accurately identified and treated differently. Compared with the traditional method of judging risks based solely on the number of operations, this method significantly improves the recognition granularity and context perception capabilities. The comprehensive risk indicator PRLV output not only reflects the complexity of the problem path itself but also reveals potential cognitive or interaction barriers behind user behavior, providing a reliable basis for subsequent intervention decisions. Overall, this method effectively promotes the progressive transformation from "detecting deviations" to "determining risks," making software systems more targeted and intelligently responsive in user experience management.

[0035] Example 6: Please refer to Figure 1 and Figure 3 ,Specifically: S5 includes S51 and S52; S51. Based on the behavioral risk indicator PRLV of all paths and the path deviation joint index PDI, a comprehensive calculation is performed to output the path experience risk total index RRDV, which measures the behavioral risk of the software in the event of path deviation. The total path experience risk index RRDV is calculated and output by the following algorithm formula; ; Where, Represents the average value of behavioral risk indicators across all paths, integrating the two key dimensions through multiplication; The total risk assessment result of the path experience risk index RRDV is evaluated by combining the behavioral risk index PRLV of all paths with the path deviation joint index PDI to evaluate the comprehensive experience issues of the entire task process; At the same time, the output results of the behavioral risk index PRLV and the path deviation joint index PDI are both dimensionless, and the output result of the path experience risk total index RRDV is also dimensionless.

[0036] S52: Based on the calculated total path experience risk index RRDV in all historical periods, the mean and standard deviation are calculated to set the path risk interval threshold. The path risk interval threshold includes the risk upper limit threshold R1 and the risk lower limit threshold R2. The upper risk threshold R1 is set based on the mean plus 75 percent of the standard deviation; The lower risk threshold R2 is set based on the mean plus 35 percent of the standard deviation; The path risk interval threshold is compared with the real-time path experience risk index (RRDV) to evaluate the risk status of the user interaction process, classify the risk level, and implement corresponding intervention measures. The specific evaluation content is as follows: When the total path experience risk index RRDV is greater than the upper risk threshold R1, the overall path of the current software is classified as a level one risk, that is, there is a significant deviation in the user experience of the software path design and a large number of risky behaviors are generated during the user experience. The first intervention measure is executed, which disables the current software path and prompts the software path to be reconstructed. When the lower risk threshold R2 is less than the total path experience risk index RRDV ≤ the upper risk threshold R1, the overall path risk of the current software is classified as a secondary risk, that is, the user behavior fluctuates significantly but the path structure is not out of control, and the second intervention measure is implemented; The second intervention measure collects click heat maps of the software operation interface and software system operation logs to automatically identify the mis-touch areas, that is, where the mis-touch frequency is greater than the mis-touch threshold. The spatial layout is then adjusted to prioritize the forward movement and step-by-step presentation of mis-operated controls. When the total path experience risk index RRDV ≤ the lower risk threshold R2, the overall path of the current software is divided into three levels of risk. That is, there is a certain deviation in the current overall path experience of the software, but the operation is risk-free and no intervention measures are required.

[0037] In this embodiment, this method calculates a composite path experience risk index (RRDV) by multiplying the previously derived path deviation combined index (PDI) and the behavioral risk indicator (PRLV) for each path. This index accurately reflects the combined risks of "structural imbalance" and "interaction anomalies" in the user experience, comprehensively characterizing the controllability and stability of the current software path design in actual use. This index breaks away from a single-dimensional judgment mechanism and significantly enhances the overall visualization of experience risk. Furthermore, based on the fluctuations of the path experience risk index (RRDV) over historical periods, a dynamic path risk interval threshold is constructed using a "mean and multi-level standard deviation" approach, forming a three-level responsive intervention mechanism. For level 1 high-risk paths, structural disabling and reconstruction prompts are automatically executed; for level 2 risk, click heatmap analysis and control space re-arrangement optimization are intelligently triggered; and for level 3 low-risk states, no intervention is performed, avoiding false positives and redundant adjustments. This dynamic risk-based response mechanism significantly improves the accuracy, adaptability, and practicality of intervention strategies. This step not only quantifies and integrates path deviations and behavioral risks, but also proactively triggers actionable intervention recommendations based on threshold models, achieving a closed-loop process from "risk identification" to "experience optimization." The ultimate result is improved user experience, reduced false positives, and a more flexible and rational path structure, effectively promoting continuous improvement of the software interaction experience and supporting scientific decision-making for design iterations.

[0038] Example 7: Please refer to Figure 1 and Figure 2 , a software user experience evaluation system based on comprehensive data analysis, including an experience behavior collection module, a collection data processing module, a path deviation analysis module, a behavior risk analysis module and a comprehensive experience risk evaluation module; The experience behavior collection module embeds a behavior perception engine in the software operation interface to collect user experience behavior data in real time and transmit the user experience behavior data to the software background server; The data collection and processing module pre-processes the user experience behavior data in the software background server to obtain a normalized behavior data set, build a behavior database, and store the normalized behavior data set; The path deviation analysis module extracts the normalized behavior data set, calculates and outputs the path deviation joint index (PDI), and performs preliminary comparative evaluation with the preset deviation interval threshold and the path deviation joint index (PDI). The behavioral risk analysis module triggers the risk identification mechanism based on the preliminary comparative assessment results, extracts the path risk data, and calculates and outputs the behavioral risk indicator PRLV based on the path risk data; The comprehensive experience risk assessment module performs comprehensive calculations based on the path deviation joint index PDI and the behavioral risk index PRLV, analyzes the path experience risk total index RRDV, sets the path risk interval threshold and conducts a secondary comparative assessment with the path experience risk total index RRDV, and generates intervention measures.

[0039] While the embodiments of the present invention have been shown and described, it will be apparent to those skilled in the art that various changes, modifications, substitutions, and alterations can be made to the embodiments without departing from the principles and spirit of the invention.

Claims

1. A software user experience evaluation method based on comprehensive data analysis, characterized by: The following steps are involved: S1, by embedding a behavior perception engine in the software operation interface, collecting user experience behavior data in real time and transmitting the user experience behavior data to the software background server; S2. Preprocess the user experience behavior data in the software backend server to obtain a normalized behavior data set, build a behavior database, and store the normalized behavior data set; S3. Extract the normalized behavior data set, calculate the output path deviation joint index (PDI), and perform preliminary comparative evaluation on the preset deviation interval threshold and the path deviation joint index (PDI); S4. Trigger the risk identification mechanism based on the preliminary comparative assessment results, extract the path risk data, and calculate and output the behavioral risk indicator PRLV based on the path risk data; S5. Based on the path deviation joint index PDI and the behavioral risk index PRLV, a comprehensive calculation is performed to analyze the path experience risk total index RRDV, and the path risk interval threshold is set to conduct a secondary comparative evaluation with the path experience risk total index RRDV, and intervention measures are generated.

2. The software user experience evaluation method based on comprehensive data analysis according to claim 1, characterized in that: Said S1 includes S11 and S12; S11. By embedding a behavior perception engine in the software's operating interface, the user automatically starts the behavior perception engine after logging into the software and authorizing behavior perception, and collects user experience behavior data in real time. By setting up an API collection application program interface, all user experience behavior data collected by the behavior perception engine are uniformly collected in a log collector; The behavior perception engine includes a path usage count engine, a non-linear jump operation engine, a path standard step number acquirer, and a server engine; The user experience behavior data includes the number of times the i-th path is used U i , the number of nonlinear jump operations J of the i-th path i , the total number of normal steps T of the i-th path i and the total number of users N; The total number of users N is collected in real time by the server engine at the back end of the software; S12. Set up a wireless communication interface of the software backend server and a 5G wireless communication network, wirelessly connect the wireless communication interface to the API collection application program interface, access the log collector of the software backend in real time, and extract real-time user experience behavior data.

3. The software user experience evaluation method based on comprehensive data analysis according to claim 2, characterized in that: Said S2 includes S21 and S22; S21. Receive user experience behavior data in real time in the software backend server, and pre-process the user experience behavior data in the software backend server to obtain a normalized behavior data set; The preprocessing includes timestamp verification and data normalization; The timestamp verification is performed by unifying the user experience behavior data into a Unix timestamp; The data normalization is performed by using a proportional normalization method on the user experience behavior data after timestamp verification, calculating the difference between the original value and the lower limit value of each parameter in the user experience behavior data, then calculating the difference between the upper limit value and the lower limit value, and performing a ratio calculation based on the two differences, thereby normalizing all parameters in the user experience behavior data to the interval [0, 1], thereby eliminating the dimensional influence of all parameters; S22. Configure an original behavior event table and a processed data table in the behavior database, write the normalized behavior data set obtained after preprocessing into the behavior database, store it in the processed data table, and write the user experience behavior data into the original behavior event table for data storage.

4. The software user experience evaluation method based on comprehensive data analysis according to claim 3 is characterized by: Said S3 includes S31 and S32; S31. Extracting a normalized behavior data set from the behavior database, calculating and outputting a path deviation joint index (PDI), and measuring the concentration and path deviation of user path selection behaviors; The path deviation combined index PDI is calculated and outputted by the following algorithm formula: ; Where n represents the total number of paths and log represents the logarithmic function.

5. The software user experience evaluation method based on comprehensive data analysis according to claim 4 is characterized by: S32. Calculate the average of historical path deviation combined indexes (PDIs). The 85th and 20th percentiles are set as deviation thresholds. The deviation thresholds include an upper deviation threshold F1 and a lower deviation threshold F2. A preliminary comparison is performed between the deviation thresholds and the real-time PDIs to determine the user's path deviation. The specific evaluation details are as follows. When the path deviation index PDI is greater than the upper threshold F1, it indicates that the access path deviation is abnormal, and the risk identification mechanism is triggered. When the deviation lower threshold F2 is less than the path deviation joint index PDI and less than the deviation upper threshold F1, it indicates that the current user access path has deviated. In this case, the acquisition frame rate of the behavior perception engine is increased by 50%. When the path deviation index PDI ≤ the deviation lower limit threshold F2, it means that the current software user access experience is stable and the path distribution is balanced. At this time, no intervention is performed.

6. The software user experience evaluation method based on comprehensive data analysis according to claim 1, characterized in that: Said S4 includes S41 and S42; S41. After the initial comparative assessment triggers the risk identification mechanism, the software system operation log API is integrated with the software backend server to collect and obtain path risk data from the software system operation log in real time. The path risk data is then normalized to eliminate the dimensionality effect of the path risk data. The path risk data includes the total number of interruption behaviors Zd on the i-th path i , the number of rework behaviors on path i Fg i , the number of wrong operations on the i-th path W i and the average operation time D of the user in path i i .

7. The software user experience evaluation method based on comprehensive data analysis according to claim 6, characterized in that: S42. Calculate a ratio based on the extracted path risk data and output a behavior risk indicator PRLV to measure the risk status of the path when the current user deviates from the software path during access. The behavior risk indicator PRLV is calculated and output by the following algorithm formula: ; Where, PRLV i The risk indicator of the i-th path.

8. The software user experience evaluation method based on comprehensive data analysis according to claim 6, characterized in that: Said S5 includes S51 and S52; S51. Based on the behavioral risk indicator PRLV of all paths and the path deviation joint index PDI, a comprehensive calculation is performed to output the path experience risk total index RRDV, which measures the behavioral risk of the software in the event of path deviation. The total path experience risk index RRDV is calculated and output by the following algorithm formula: 。 9. The software user experience evaluation method based on comprehensive data analysis according to claim 8, characterized in that: S52: Based on the calculated total path experience risk index RRDV in all historical periods, the mean and standard deviation are calculated to set the path risk interval threshold, which includes the upper risk threshold R1 and the lower risk threshold R2. The upper risk threshold R1 is set based on the mean plus 75 percent of the standard deviation; The lower risk threshold R2 is set based on the mean plus 35 percent of the standard deviation; The path risk interval threshold is compared with the real-time path experience risk index (RRDV) to evaluate the risk status of the user interaction process, classify the risk level, and implement corresponding intervention measures. The specific evaluation content is as follows: When the total path experience risk index RRDV is greater than the upper risk threshold R1, the entire path of the current software is classified as a level 1 risk, and the first intervention measure is executed. The first intervention measure disables the current software path and prompts the software path to be reconstructed. When the lower risk threshold R2 is less than the total path experience risk index RRDV and less than the upper risk threshold R1, the overall path risk of the current software is classified as a secondary risk and the second intervention measure is implemented; The second intervention measure automatically identifies mis-touch areas by collecting click heat maps of the software operation interface and software system operation logs, adjusts the spatial layout, and prioritizes moving mis-touch controls forward and presenting them in steps. When the total path experience risk index RRDV ≤ the lower risk threshold R2, the overall path of the current software is divided into three levels of risk and no intervention measures are required.

10. A software user experience evaluation system based on comprehensive data analysis, applied to the software user experience evaluation method based on comprehensive data analysis according to any one of claims 1 to 9, characterized in that: It includes experience behavior collection module, collection data processing module, path deviation analysis module, behavior risk analysis module and comprehensive experience risk assessment module; The experience behavior collection module collects user experience behavior data in real time by embedding a behavior perception engine in the software operation interface, and transmits the user experience behavior data to the software background server; The data collection processing module pre-processes the user experience behavior data in the software background server to obtain a normalized behavior data set, and builds a behavior database to store the normalized behavior data set; The path deviation analysis module extracts the normalized behavior data set, calculates and outputs the path deviation joint index PDI, and performs preliminary comparative evaluation on the preset deviation interval threshold and the path deviation joint index PDI; The behavior risk analysis module triggers a risk identification mechanism based on preliminary comparative assessment results, extracts path risk data, and calculates and outputs a behavior risk indicator PRLV based on the path risk data; The comprehensive experience risk assessment module performs comprehensive calculations based on the path deviation joint index PDI and the behavioral risk index PRLV, analyzes the path experience risk total index RRDV, sets the path risk interval threshold and conducts a secondary comparative assessment with the path experience risk total index RRDV, and generates intervention measures.

Images