Method and system for analyzing use data of mobile terminal application program

Abstract

The invention provides a mobile terminal application program use data analysis method and system, and relates to the technical field of data analysis, and the method comprises the steps: obtaining user operation behavior information, calculating a behavior weight coefficient and an event weight coefficient, obtaining an event storage node, a sub-event behavior path and a key event behavior path, carrying out the path aggregation, and obtaining an aggregation path; calculating a key event exception coefficient of a key event node of the application program to obtain an exceptional key event node, and further calculating a sub-event exception coefficient to obtain an exceptional sub-event node; and performing node information adjustment on the abnormal sub-event nodes, obtaining the number of the abnormal sub-event nodes and the number of the abnormal key event nodes after adjustment, judging whether adjustment is continued or not, and obtaining an adjustment result. And more accurate and efficient data support is provided for optimization and operation decision making of the mobile application program.

Description

Technical Field

[0001] The present invention proposes a method and system for analyzing data used by a mobile application, and relates to the technical field of data analysis, and in particular to the technical field of analyzing data used by a mobile application. Background Art

[0002] With the rapid development of mobile Internet technology, mobile applications (APPs) have become an indispensable part of people's daily life and work. These applications have greatly improved user efficiency and satisfaction by providing rich functions and convenient operation experience. However, with the increasing complexity and functions of applications, how to effectively monitor and analyze user usage behavior to optimize application performance, improve user experience and prevent potential problems has become an important challenge faced by developers and operators.

[0003] Existing mobile application data analysis methods mainly rely on log collection and statistical analysis. By collecting user operation logs, users' behavior patterns and usage habits are statistically analyzed to provide data support for product optimization and operational decisions. However, this method has many shortcomings when dealing with large-scale user data and complex behavior paths. First, traditional log analysis methods can often only provide relatively coarse-grained statistical information, and it is difficult to conduct detailed analysis at the level of specific events and sub-events. Secondly, due to the diversity and complexity of user behavior data, traditional analysis methods are prone to false positives or omissions when processing these data, resulting in insufficient accuracy of the analysis results. Finally, traditional analysis methods also have bottlenecks in data storage and anomaly analysis, making it difficult to cope with the real-time processing needs of large-scale user data. Summary of the invention

[0004] The present invention provides a method and system for analyzing mobile application usage data to solve the above problems:

[0005] The present invention provides a method and system for analyzing mobile application usage data, the method comprising:

[0006] S1. Obtain user operation behavior information, classify the behavior information and calculate the behavior weight coefficient, and then calculate the event weight coefficient corresponding to the preset time period to obtain key event nodes and sub-event nodes;

[0007] S2. Distribute and store event node information to obtain event storage nodes, obtain sub-event behavior paths and key event behavior paths according to the event storage nodes, aggregate the paths, and obtain aggregated paths;

[0008] S3, calculating the key event abnormality coefficient of the key event node of the application program to obtain the abnormal key event node, and then calculating the sub-event abnormality coefficient to obtain the abnormal sub-event node;

[0009] S4. Adjust the node information of the abnormal sub-event nodes, obtain the number of abnormal sub-event nodes and the number of abnormal key event nodes after adjustment, determine whether to continue adjustment, and obtain the adjustment result.

[0010] Furthermore, the S1 includes:

[0011] Acquire user operation behavior information, classify the user operation behavior information according to preset behavior categories, and obtain behavior classification information;

[0012] Perform edge computing on each behavior classification information, and calculate the behavior weight coefficient of each behavior classification information through edge computing method;

[0013] Obtaining the behavior weight coefficient of each behavior classification information in the same preset time period to calculate the event weight coefficient corresponding to the preset time period;

[0014] According to the event weight coefficient, the user operation behavior information in the preset time period is annotated with events of corresponding weights to obtain key event nodes, and each behavior classification of the key event node is annotated with sub-events to obtain sub-event nodes.

[0015] Further, the S2 includes:

[0016] Obtain data of multiple sub-event nodes and their key event nodes and upload them to the cloud;

[0017] Distribute and store each sub-event node of the key event node through the cloud to obtain the event storage node;

[0018] Connect the sub-event nodes of each behavior classification of the event storage node to obtain the sub-event behavior path of the corresponding behavior classification;

[0019] Connect the key event nodes of each event storage node to obtain the key event behavior path;

[0020] Aggregate the sub-event behavior path with the key event behavior path to obtain the aggregate path.

[0021] Further, the S3 includes:

[0022] Obtain application performance data on the aggregation path, and calculate the key event anomaly coefficient of the key event node of the application program based on the application performance data combined with the event weight coefficient;

[0023] Anomaly marking each key event node according to the key event anomaly coefficient to obtain an abnormal key event node;

[0024] Obtaining sub-event node information of the abnormal key event node, and calculating the sub-event abnormality coefficient of each sub-event node of the abnormal key event node according to the application performance information combined with the sub-event information;

[0025] Each sub-event node is annotated as abnormal according to the sub-event abnormal coefficient to obtain an abnormal sub-event node.

[0026] Further, the S4 includes:

[0027] Obtaining node information of each abnormal sub-event node, adjusting the node information, and obtaining an adjusted sub-event abnormality coefficient;

[0028] Obtaining the number of abnormal sub-event nodes according to the adjusted sub-event abnormal coefficient;

[0029] Determine whether to calculate the corresponding key event abnormality coefficient according to the number of abnormal sub-event nodes;

[0030] Then the number of abnormal key event nodes after adjustment is obtained.

[0031] Furthermore, the system comprises:

[0032] The weight node acquisition module is used to obtain user operation behavior information, classify the behavior information and calculate the behavior weight coefficient, and then calculate the event weight coefficient corresponding to the preset time period, and obtain key event nodes and sub-event nodes;

[0033] The event path acquisition module is used to perform distributed storage of event node information, obtain event storage nodes, obtain sub-event behavior paths and key event behavior paths according to the event storage nodes, aggregate the paths, and obtain aggregated paths;

[0034] The abnormal calculation and analysis module is used to calculate the key event abnormality coefficient of the key event node of the application program, obtain the abnormal key event node, and then calculate the sub-event abnormality coefficient to obtain the abnormal sub-event node;

[0035] The abnormal adjustment judgment module is used to adjust the node information of the abnormal sub-event node, obtain the number of abnormal sub-event nodes and the number of abnormal key event nodes after adjustment, determine whether to continue adjustment, and obtain the adjustment result.

[0036] Furthermore, the weight node acquisition module includes:

[0037] A behavior classification module is used to obtain user operation behavior information, classify the user operation behavior information according to preset behavior types, and obtain behavior classification information;

[0038] A classification weight calculation module is used to perform edge calculation on each behavior classification information and calculate the behavior weight coefficient of each behavior classification information by edge calculation method;

[0039] An event weight calculation module is used to obtain the behavior weight coefficient of each behavior classification information in the same preset time period and calculate the event weight coefficient corresponding to the preset time period;

[0040] The event node acquisition module is used to mark the user operation behavior information in a preset time period with events of corresponding weights according to the event weight coefficient to obtain key event nodes, and to mark each behavior classification of the key event nodes with sub-events to obtain sub-event nodes.

[0041] Furthermore, the event path acquisition module includes:

[0042] A distributed storage module is used to obtain data of multiple sub-event nodes and their key event nodes and upload them to the cloud;

[0043] Distribute and store each sub-event node of the key event node through the cloud to obtain the event storage node;

[0044] A path generation module is used to connect the sub-event nodes of each behavior classification of the event storage node to obtain the sub-event behavior path of the corresponding behavior classification;

[0045] Connect the key event nodes of each event storage node to obtain the key event behavior path;

[0046] The path aggregation module is used to aggregate the sub-event behavior path with the key event behavior path to obtain an aggregated path.

[0047] Furthermore, the abnormal calculation analysis module includes:

[0048] A key anomaly calculation module is used to obtain application performance data on the aggregation path, and calculate the key event anomaly coefficient of the key event node of the application program based on the application performance data combined with the event weight coefficient;

[0049] A key node anomaly acquisition module is used to annotate each key event node according to the key event anomaly coefficient to obtain an abnormal key event node;

[0050] A sub-anomaly calculation module, used to obtain sub-event node information of the abnormal key event node, and calculate the sub-event anomaly coefficient of each sub-event node of the abnormal key event node according to the application performance information combined with the sub-event information;

[0051] The sub-node anomaly acquisition module is used to annotate each sub-event node according to the sub-event anomaly coefficient to obtain an abnormal sub-event node.

[0052] Furthermore, the abnormal adjustment judgment module includes:

[0053] An adjustment sub-node acquisition module is used to obtain node information of each abnormal sub-event node, adjust the node information, and obtain an adjusted sub-event abnormality coefficient;

[0054] Obtaining the number of abnormal sub-event nodes according to the adjusted sub-event abnormal coefficient;

[0055] The key node acquisition module is adjusted to determine whether to calculate the corresponding key event abnormality coefficient according to the number of abnormal sub-event nodes;

[0056] Then the number of abnormal key event nodes after adjustment is obtained.

[0057] Beneficial effects of the present invention: By accurately identifying anomalies in user behavior and making targeted adjustments, the user's satisfaction and efficiency in the process of using the application can be significantly improved. Based on the results of big data analysis, application developers can formulate product strategies more scientifically, such as adjusting functional layout, optimizing recommendation algorithms, etc., to better meet user needs. By real-time monitoring of abnormal events in user behavior and handling them in a timely manner, the risk of application crashes or performance degradation can be effectively reduced, and the stability and reliability of the application can be improved. Distributed storage and efficient data processing mechanisms enable the system to respond quickly to data changes. Based on in-depth analysis of user behavior data, the application can provide a more personalized service experience, such as personalized recommendations, customized content, etc., thereby enhancing user stickiness and loyalty. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 A schematic diagram of a data analysis method used for a mobile application;

[0059] Figure 2 A schematic diagram of the aggregation path. DETAILED DESCRIPTION

[0060] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0061] In one embodiment of the present invention, a method and system for analyzing mobile application usage data is provided, and the method includes:

[0062] S1. Obtain user operation behavior information, classify the behavior information and calculate the behavior weight coefficient, and then calculate the event weight coefficient corresponding to the preset time period to obtain key event nodes and sub-event nodes;

[0063] S2. Distribute and store event node information to obtain event storage nodes, obtain sub-event behavior paths and key event behavior paths according to the event storage nodes, aggregate the paths, and obtain aggregated paths;

[0064] S3, calculating the key event abnormality coefficient of the key event node of the application program to obtain the abnormal key event node, and then calculating the sub-event abnormality coefficient to obtain the abnormal sub-event node;

[0065] S4, adjust the node information of the abnormal sub-event node, obtain the number of abnormal sub-event nodes and the number of abnormal key event nodes after adjustment, determine whether to continue adjustment, and obtain the adjustment result, such as Figure 1 shown.

[0066] The working principle of the above technical solution is as follows: the system first collects various operation behavior information of users in the application, such as clicks, slides, dwell time, etc. These behavior information are classified to identify different types of user activities, such as browsing, searching, purchasing, etc. The weight coefficient of each behavior type is calculated to reflect its importance to the overall user behavior. Based on the behavior weight coefficient, the weight coefficient of each event within the preset time period is further calculated, which helps to identify which application usage events have a greater impact on the overall performance or user experience of the application. Key event nodes and their sub-event nodes are identified, which represent the key paths or important turning points of user behavior in the application. Event node information is distributedly stored to improve the efficiency and scalability of data processing. Based on the stored event nodes, sub-event behavior paths and key event behavior paths are constructed and aggregated to form a complete user behavior path map. The abnormal coefficient of the key event node is calculated, and the abnormal key event node is identified by comparing the difference between the actual behavior and the expected or historical average behavior. The sub-event abnormal coefficient is calculated to further identify the specific abnormal sub-event node. The identified abnormal sub-event node is adjusted for information, which may be refreshing the operation data or refreshing the running data. Monitor the effect after adjustment, evaluate whether the adjustment is effective by comparing the number of abnormal sub-event nodes and key event nodes before and after adjustment, and decide whether to continue the adjustment.

[0067] The technical effect of the above technical solution is: by accurately identifying abnormal points in user behavior and making targeted adjustments, the user's satisfaction and efficiency in the process of using the application can be significantly improved. Based on the results of big data analysis, application developers can formulate product strategies more scientifically, such as adjusting functional layout, optimizing recommendation algorithms, etc., to better meet user needs. By real-time monitoring of abnormal events in user behavior and handling them in a timely manner, the risk of application crashes or performance degradation can be effectively reduced, and the stability and reliability of the application can be improved. Distributed storage and efficient data processing mechanisms enable the system to respond quickly to data changes. Based on in-depth analysis of user behavior data, the application can provide a more personalized service experience, such as personalized recommendations, customized content, etc., thereby enhancing user stickiness and loyalty.

[0068] In one embodiment of the present invention, the S1 includes:

[0069] Acquire user operation behavior information, classify the user operation behavior information according to preset behavior categories, and obtain behavior classification information;

[0070] Perform edge computing on each behavior classification information, and calculate the behavior weight coefficient of each behavior classification information through edge computing method;

[0071] The calculation formula of the behavior weight coefficient is:

[0072]

[0073] Among them, Q xw is the behavior weight coefficient, c is the number of behavior operation types that need to be calculated, Z ix is the average operation data of the i-th operation type, Z yxi is the preset operation data of the i-th operation type, n is the total number of types of behavior classification information, Z ia is the average operation data of the ith operation type of the ath behavior classification information, Z ya The preset operation data of the ith operation type of the ath behavior classification information;

[0074] Obtaining the behavior weight coefficient of each behavior classification information in the same preset time period to calculate the event weight coefficient corresponding to the preset time period;

[0075] The calculation formula of the event weight coefficient is:

[0076]

[0077] Among them, Q sj is the event weight coefficient, s is the number of behavior classification information types within the preset time period, Q xwdis the behavior weight coefficient of the dth behavior classification information type within the preset time period, Q xwa is the behavior weight coefficient of the a-th behavior classification information;

[0078] According to the event weight coefficient, the user operation behavior information in the preset time period is annotated with events of corresponding weights to obtain key event nodes, and each behavior classification of the key event node is annotated with sub-events to obtain sub-event nodes.

[0079] The working principle of the above technical solution is as follows: the system first collects various operation behavior information of users in the application in real time or regularly, including but not limited to clicks, slides, inputs, dwell time, etc. According to the preset behavior types (such as browsing, searching, purchasing, sharing, etc.), the user operation behavior information is classified and processed to form structured behavior classification information. For each behavior classification information, edge computing technology is used to perform calculations at or near the source where it is generated to reduce data transmission delays and reduce the pressure on cloud servers. The behavior weight coefficient of each behavior classification information is calculated. This coefficient reflects the importance or influence of the behavior in the overall behavior of the user. In the same preset time period (such as one day, one week or one month), the behavior weight coefficients of various behavior classification information are summarized. The event weight coefficient within the time period is calculated. This coefficient is used to measure the overall trend or key nodes of user behavior in the entire time period. According to the event weight coefficient, the user operation behavior information of the preset time period is weighted and labeled. Behaviors or behavior sequences with higher weights are regarded as key event nodes, which represent key turning points or important activities in user behavior. For each key event node, the internal behavior classification information is further analyzed and sub-events are labeled. Sub-event nodes are refinements within key events and are used to more accurately describe the user's specific behavior in the key event.

[0080] The technical effects of the above technical solution are: through edge computing technology, the real-time processing and analysis of user operation behavior information is realized, the delay of data transmission is reduced, and the response speed of the system is improved. Using complex calculation formulas and weighting mechanisms, the importance of user behavior can be more accurately evaluated, and key events and their sub-events can be identified. This provides strong support for subsequent personalized recommendations, user behavior predictions, etc. Edge computing reduces the burden on cloud servers, allowing the system to use computing resources more efficiently. At the same time, distributed data storage and processing also improve the scalability and fault tolerance of the system. By deeply analyzing user behavior data, developers can understand user needs and behavior habits more accurately, thereby optimizing the functional design, interface layout and other aspects of the application and improving the user experience. Based on the analysis results of user behavior data, companies can make more scientific and reasonable business decisions, such as product pricing and marketing strategy adjustments, to achieve better business benefits.

[0081] In one embodiment of the present invention, S2 includes:

[0082] Obtain data of multiple sub-event nodes and their key event nodes and upload them to the cloud;

[0083] Distribute and store each sub-event node of the key event node through the cloud to obtain the event storage node;

[0084] Connect the sub-event nodes of each behavior classification of the event storage node to obtain the sub-event behavior path of the corresponding behavior classification;

[0085] Connect the key event nodes of each event storage node to obtain the key event behavior path;

[0086] Aggregate the sub-event behavior path with the key event behavior path to obtain the aggregate path, such as Figure 2 shown.

[0087] The working principle of the above technical solution is as follows: the data of multiple sub-event nodes and their corresponding key event nodes collected by the mobile terminal or edge device are packaged and securely transmitted to the cloud server. The cloud server receives the data and performs preliminary verification and cleaning to ensure the integrity and accuracy of the data. The cloud server uses a distributed storage system (such as Hadoop, NoSQL database, etc.) to store the received data. The sub-event nodes of each key event node are independently stored as event storage nodes, which are distributed in different storage units to improve the reliability and access efficiency of the data. For each event storage node, the system connects the sub-event nodes belonging to the same behavior classification according to its internal behavior classification information to form a sub-event behavior path of the corresponding behavior classification. The system also connects the key event nodes of each event storage node to form a key event behavior path. These paths reflect the specific behavior process of users under different behavior classifications and the association between key events. The system aggregates the sub-event behavior path with the key event behavior path to form a complete aggregate path. This aggregate path not only contains the detailed behavior process of users under different behavior classifications, but also reveals the position and role of key events in the entire user behavior.

[0088] The technical effect of the above technical solution is: through distributed storage and path construction technology, the originally scattered user behavior data is integrated into an orderly and visual path map, so that developers can intuitively understand the user's behavior patterns and key events. The distributed storage system provides efficient data query and analysis capabilities, allowing developers to quickly retrieve user behavior data under a specific time period or behavior classification, and conduct in-depth statistical analysis and data mining. The distributed storage system can make full use of the computing resources and storage resources in the cloud to achieve efficient management and fast access to data. At the same time, with the increase in the number of users and the growth of data volume, the system can easily expand storage and computing capabilities to meet business needs. Based on the analysis results of the aggregated path, enterprises can make more accurate marketing strategies, optimize product functions, improve user experience and other decisions to achieve better business benefits and user satisfaction. Since the data is processed in the cloud, the system can receive and process new user behavior data in real time, and perform flexible path construction and aggregate analysis according to demand. This enables the system to respond quickly to market changes and changes in user needs.

[0089] In one embodiment of the present invention, S3 includes:

[0090] Obtain application performance data on the aggregation path, and calculate the key event anomaly coefficient of the key event node of the application program based on the application performance data combined with the event weight coefficient;

[0091] The calculation formula of the key event abnormality coefficient is:

[0092]

[0093] Among them, T gj is the key event anomaly coefficient of the key event node, G is the total number of application performance data types within the preset time period of the key event node, and r se is the actual performance data of the e-th application performance data type, r yes is the preset performance data of the e-th application performance data type;

[0094] Anomaly marking each key event node according to the key event anomaly coefficient to obtain an abnormal key event node;

[0095] Obtaining sub-event node information of the abnormal key event node, and calculating the sub-event abnormality coefficient of each sub-event node of the abnormal key event node according to the application performance information combined with the sub-event information;

[0096] The calculation formula of the sub-event abnormal coefficient is:

[0097]

[0098] Among them, T z is the sub-event anomaly coefficient of the sub-event node, m is the total number of application performance data types within the preset time period of the sub-event node, and r so is the actual performance data of the oth application performance data type, r yos It is the preset performance data of the oth application performance data type; the calculated data correspond one by one according to the type.

[0099] Each sub-event node is annotated as abnormal according to the sub-event abnormal coefficient to obtain an abnormal sub-event node.

[0100] The working principle of the above technical solution is as follows: the system needs to collect the performance data of the application on the aggregation path from each monitoring point or log source. These data may include key indicators such as response time, throughput, error rate, etc. For each key event node, the system calculates its weight coefficient, which reflects the degree of influence of the event node on the performance of the entire application. Using the collected application performance data and event weight coefficient, the key event abnormality coefficient of each key event node is calculated according to the given formula. This coefficient is a quantitative indicator used to evaluate the current abnormality of the event node. According to the calculated key event abnormality coefficient, the system will mark each key event node as abnormal. When the abnormality coefficient exceeds the preset threshold, the event node is marked as abnormal. For the key event node marked as abnormal, the system further analyzes its child event nodes. By obtaining detailed information of the child event node (such as child event type, occurrence time, performance data, etc.), combined with the application performance data of the parent node, the child event abnormality coefficient of each child event node is calculated. According to the child event abnormality coefficient, each child event node is marked as abnormal. This helps to accurately locate the source of the problem and provide a basis for subsequent troubleshooting and optimization.

[0101] The technical effect of the above technical solution is: through layer-by-layer analysis (from key event nodes to sub-event nodes), the performance bottleneck or abnormal source in the application can be more accurately located, thereby improving the efficiency of problem solving. The system can collect and analyze application performance data in real time, promptly discover potential performance problems, and avoid the expansion of problems. Based on quantitative data analysis and weight coefficients, the system can more scientifically evaluate the degree of abnormality of event nodes and provide strong support for decision-making. By identifying key event nodes and abnormal sub-event nodes, enterprises can optimize resource allocation in a more targeted manner and improve resource utilization efficiency.

[0102] Timely discovery and resolution of application performance issues can significantly improve user experience and enhance user satisfaction and loyalty.

[0103] In one embodiment of the present invention, the S4 includes:

[0104] Obtaining node information of each abnormal sub-event node, adjusting the node information, and obtaining an adjusted sub-event abnormality coefficient;

[0105] Obtaining the number of abnormal sub-event nodes according to the adjusted sub-event abnormal coefficient;

[0106] Determine whether to calculate the corresponding key event abnormality coefficient according to the number of abnormal sub-event nodes;

[0107] The number of abnormal sub-event nodes is compared with a preset number threshold, and when the number of abnormal sub-event nodes is greater than the preset number threshold, the corresponding key event abnormality coefficient is calculated;

[0108] Then, the number of abnormal key event nodes after adjustment is obtained. When there is no abnormal key event node, the calculation and adjustment of the abnormal coefficient is stopped.

[0109] The working principle of the above technical solution is as follows: for each abnormal sub-event node, the system first obtains its node information, which may include the occurrence time, duration, scope of influence, related resource usage, etc. The system adjusts the node information according to the preset rules. After adjustment, the abnormal coefficient of the sub-event node is recalculated to obtain the adjusted sub-event abnormal coefficient. According to the adjusted sub-event abnormal coefficient, the system counts the number of abnormal sub-event nodes under the current key event. The system compares the number of abnormal sub-event nodes with the preset number threshold. This threshold is used to determine whether the current key event is "abnormal" enough to further calculate its key event abnormal coefficient. If the number of abnormal sub-event nodes is greater than the preset number threshold, it means that the key event is likely to have significant performance problems or abnormal behavior, and a deeper analysis is required. When the conditions are met, the system will calculate the corresponding key event abnormal coefficient based on the adjusted sub-event abnormal coefficient and other relevant information (such as key event weight, historical data, etc.). This calculation process may involve complex mathematical models or machine learning algorithms to ensure the accuracy and reliability of the abnormal coefficient. After calculating the abnormal coefficient of the key event, the system may further evaluate or adjust the key event itself based on the coefficient (for example, adjust the weight, reclassify, adjust the operation data, etc.). If necessary, the system will also update the number of abnormal key event nodes based on the adjustment results. If no abnormal key event nodes are found during the inspection process, or the predetermined analysis depth has been reached, the system will stop the calculation and adjustment process of the abnormal coefficient.

[0110] The technical effect of the above technical solution is: through the adjustment of node information and multiple iterative calculations, the system can more accurately evaluate the degree of abnormality of sub-events and key events, and reduce false positives and missed reports. By setting the number threshold, the system can avoid in-depth analysis of each subtle anomaly, thereby saving computing resources and time. The process can be adjusted and optimized according to different application scenarios and needs, such as by modifying the threshold, adjusting the node information adjustment rules, or improving the abnormality coefficient calculation algorithm. The system can collect and analyze data in real time, and dynamically adjust the calculation and annotation results of the abnormality coefficient according to changes in the data to ensure real-time monitoring and response to application performance. By accurately locating abnormal events and sub-events, the system can provide developers and operation and maintenance personnel with more targeted troubleshooting and optimization suggestions, thereby improving the efficiency of problem solving.

[0111] In one embodiment of the present invention, the system comprises:

[0112] The weight node acquisition module is used to obtain user operation behavior information, classify the behavior information and calculate the behavior weight coefficient, and then calculate the event weight coefficient corresponding to the preset time period, and obtain key event nodes and sub-event nodes;

[0113] The event path acquisition module is used to perform distributed storage of event node information, obtain event storage nodes, obtain sub-event behavior paths and key event behavior paths according to the event storage nodes, aggregate the paths, and obtain aggregated paths;

[0114] The abnormal calculation and analysis module is used to calculate the key event abnormality coefficient of the key event node of the application program, obtain the abnormal key event node, and then calculate the sub-event abnormality coefficient to obtain the abnormal sub-event node;

[0115] The abnormal adjustment judgment module is used to adjust the node information of the abnormal sub-event node, obtain the number of abnormal sub-event nodes and the number of abnormal key event nodes after adjustment, determine whether to continue adjustment, and obtain the adjustment result.

[0116] The working principle of the above technical solution is as follows: the system first collects various operation behavior information of users in the application, such as clicks, slides, dwell time, etc. These behavior information are classified to identify different types of user activities, such as browsing, searching, purchasing, etc. The weight coefficient of each behavior type is calculated to reflect its importance to the overall user behavior. Based on the behavior weight coefficient, the weight coefficient of each event within the preset time period is further calculated, which helps to identify which application usage events have a greater impact on the overall performance or user experience of the application. Key event nodes and their sub-event nodes are identified, which represent the key paths or important turning points of user behavior in the application. Event node information is distributedly stored to improve the efficiency and scalability of data processing. Based on the stored event nodes, sub-event behavior paths and key event behavior paths are constructed and aggregated to form a complete user behavior path map. The abnormal coefficient of the key event node is calculated, and the abnormal key event node is identified by comparing the difference between the actual behavior and the expected or historical average behavior. The sub-event abnormal coefficient is calculated to further identify the specific abnormal sub-event node. The identified abnormal sub-event node is adjusted for information, which may be refreshing the operation data or refreshing the running data. Monitor the effect after adjustment, evaluate whether the adjustment is effective by comparing the number of abnormal sub-event nodes and key event nodes before and after adjustment, and decide whether to continue the adjustment.

[0117] The technical effect of the above technical solution is: by accurately identifying abnormal points in user behavior and making targeted adjustments, the user's satisfaction and efficiency in the process of using the application can be significantly improved. Based on the results of big data analysis, application developers can formulate product strategies more scientifically, such as adjusting functional layout, optimizing recommendation algorithms, etc., to better meet user needs. By real-time monitoring of abnormal events in user behavior and handling them in a timely manner, the risk of application crashes or performance degradation can be effectively reduced, and the stability and reliability of the application can be improved. Distributed storage and efficient data processing mechanisms enable the system to respond quickly to data changes. Based on in-depth analysis of user behavior data, the application can provide a more personalized service experience, such as personalized recommendations, customized content, etc., thereby enhancing user stickiness and loyalty.

[0118] In one embodiment of the present invention, the weight node acquisition module includes:

[0119] A behavior classification module is used to obtain user operation behavior information, classify the user operation behavior information according to preset behavior types, and obtain behavior classification information;

[0120] A classification weight calculation module is used to perform edge calculation on each behavior classification information and calculate the behavior weight coefficient of each behavior classification information by edge calculation method;

[0121] The calculation formula of the behavior weight coefficient is:

[0122]

[0123] Among them, Q xw is the behavior weight coefficient, c is the number of behavior operation types that need to be calculated, Z ix is the average operation data of the i-th operation type, Z yxi is the preset operation data of the i-th operation type, n is the total number of behavior classification information, Z ia is the average operation data of the ith operation type of the ath behavior classification information, Z ya The preset operation data of the i-th operation type of the a-th behavior classification information;

[0124] An event weight calculation module is used to obtain the behavior weight coefficient of each behavior classification information in the same preset time period and calculate the event weight coefficient corresponding to the preset time period;

[0125] The calculation formula of the event weight coefficient is:

[0126]

[0127] Among them, Q sj is the event weight coefficient, s is the number of behavior classification information types within the preset time period, Q xwd is the behavior weight coefficient of the dth behavior classification information type within the preset time period, Q xwa is the behavior weight coefficient of the a-th behavior classification information;

[0128] The event node acquisition module is used to mark the user operation behavior information in a preset time period with events of corresponding weights according to the event weight coefficient to obtain key event nodes, and to mark each behavior classification of the key event nodes with sub-events to obtain sub-event nodes.

[0129] The working principle of the above technical solution is as follows: the system first collects various operation behavior information of users in the application in real time or regularly, including but not limited to clicks, slides, inputs, dwell time, etc. According to the preset behavior types (such as browsing, searching, purchasing, sharing, etc.), the user operation behavior information is classified and processed to form structured behavior classification information. For each behavior classification information, edge computing technology is used to perform calculations at or near the source where it is generated to reduce data transmission delays and reduce the pressure on cloud servers. The behavior weight coefficient of each behavior classification information is calculated. This coefficient reflects the importance or influence of the behavior in the overall behavior of the user. In the same preset time period (such as one day, one week or one month), the behavior weight coefficients of various behavior classification information are summarized. The event weight coefficient within the time period is calculated. This coefficient is used to measure the overall trend or key nodes of user behavior in the entire time period. According to the event weight coefficient, the user operation behavior information of the preset time period is weighted and labeled. Behaviors or behavior sequences with higher weights are regarded as key event nodes, which represent key turning points or important activities in user behavior. For each key event node, the internal behavior classification information is further analyzed and sub-events are labeled. Sub-event nodes are refinements within key events and are used to more accurately describe the user's specific behavior in the key event.

[0130] The technical effects of the above technical solution are: through edge computing technology, the real-time processing and analysis of user operation behavior information is realized, the delay of data transmission is reduced, and the response speed of the system is improved. Using complex calculation formulas and weighting mechanisms, the importance of user behavior can be more accurately evaluated, and key events and their sub-events can be identified. This provides strong support for subsequent personalized recommendations, user behavior predictions, etc. Edge computing reduces the burden on cloud servers, allowing the system to use computing resources more efficiently. At the same time, distributed data storage and processing also improve the scalability and fault tolerance of the system. By deeply analyzing user behavior data, developers can understand user needs and behavior habits more accurately, thereby optimizing the functional design, interface layout and other aspects of the application and improving the user experience. Based on the analysis results of user behavior data, companies can make more scientific and reasonable business decisions, such as product pricing and marketing strategy adjustments, to achieve better business benefits.

[0131] In one embodiment of the present invention, the event path acquisition module includes:

[0132] A distributed storage module is used to obtain data of multiple sub-event nodes and their key event nodes and upload them to the cloud;

[0133] Distribute and store each sub-event node of the key event node through the cloud to obtain the event storage node;

[0134] A path generation module is used to connect the sub-event nodes of each behavior classification of the event storage node to obtain the sub-event behavior path of the corresponding behavior classification;

[0135] Connect the key event nodes of each event storage node to obtain the key event behavior path;

[0136] The path aggregation module is used to aggregate the sub-event behavior path with the key event behavior path to obtain an aggregated path.

[0137] The working principle of the above technical solution is as follows: the data of multiple sub-event nodes and their corresponding key event nodes collected by the mobile terminal or edge device are packaged and securely transmitted to the cloud server. The cloud server receives the data and performs preliminary verification and cleaning to ensure the integrity and accuracy of the data. The cloud server uses a distributed storage system (such as Hadoop, NoSQL database, etc.) to store the received data. The sub-event nodes of each key event node are independently stored as event storage nodes, which are distributed in different storage units to improve the reliability and access efficiency of the data. For each event storage node, the system connects the sub-event nodes belonging to the same behavior classification according to its internal behavior classification information to form a sub-event behavior path of the corresponding behavior classification. The system also connects the key event nodes of each event storage node to form a key event behavior path. These paths reflect the specific behavior process of users under different behavior classifications and the association between key events. The system aggregates the sub-event behavior path with the key event behavior path to form a complete aggregate path. This aggregate path not only contains the detailed behavior process of users under different behavior classifications, but also reveals the position and role of key events in the entire user behavior.

[0138] The technical effect of the above technical solution is: through distributed storage and path construction technology, the originally scattered user behavior data is integrated into an orderly and visual path map, so that developers can intuitively understand the user's behavior patterns and key events. The distributed storage system provides efficient data query and analysis capabilities, allowing developers to quickly retrieve user behavior data under a specific time period or behavior classification, and conduct in-depth statistical analysis and data mining. The distributed storage system can make full use of the computing resources and storage resources in the cloud to achieve efficient management and fast access to data. At the same time, with the increase in the number of users and the growth of data volume, the system can easily expand storage and computing capabilities to meet business needs. Based on the analysis results of the aggregated path, enterprises can make more accurate marketing strategies, optimize product functions, improve user experience and other decisions to achieve better business benefits and user satisfaction. Since the data is processed in the cloud, the system can receive and process new user behavior data in real time, and perform flexible path construction and aggregate analysis according to demand. This enables the system to respond quickly to market changes and changes in user needs.

[0139] In one embodiment of the present invention, the abnormal calculation analysis module includes:

[0140] A key anomaly calculation module is used to obtain application performance data on the aggregation path, and calculate the key event anomaly coefficient of the key event node of the application program based on the application performance data combined with the event weight coefficient;

[0141] The calculation formula of the key event abnormality coefficient is:

[0142]

[0143] Among them, T gj is the key event anomaly coefficient of the key event node, G is the total number of application performance data types within the preset time period of the key event node, and r se is the actual performance data of the e-th application performance data type, r yes is the preset performance data of the e-th application performance data type;

[0144] A key node anomaly acquisition module is used to annotate each key event node according to the key event anomaly coefficient to obtain an abnormal key event node;

[0145] A sub-anomaly calculation module, used to obtain sub-event node information of the abnormal key event node, and calculate the sub-event anomaly coefficient of each sub-event node of the abnormal key event node according to the application performance information combined with the sub-event information;

[0146] The calculation formula of the sub-event abnormal coefficient is:

[0147]

[0148] Among them, T z is the sub-event anomaly coefficient of the sub-event node, m is the total number of application performance data types within the preset time period of the sub-event node, and r so is the actual performance data of the oth application performance data type, r yos It is the preset performance data of the oth application performance data type; the calculated data correspond one by one according to the type.

[0149] The sub-node anomaly acquisition module is used to annotate each sub-event node according to the sub-event anomaly coefficient to obtain an abnormal sub-event node.

[0150] The working principle of the above technical solution is as follows: the system needs to collect the performance data of the application on the aggregation path from each monitoring point or log source. These data may include key indicators such as response time, throughput, error rate, etc. For each key event node, the system calculates its weight coefficient, which reflects the degree of influence of the event node on the performance of the entire application. Using the collected application performance data and event weight coefficient, the key event abnormality coefficient of each key event node is calculated according to the given formula. This coefficient is a quantitative indicator used to evaluate the current abnormality of the event node. According to the calculated key event abnormality coefficient, the system will mark each key event node as abnormal. When the abnormality coefficient exceeds the preset threshold, the event node is marked as abnormal. For the key event node marked as abnormal, the system further analyzes its child event nodes. By obtaining detailed information of the child event node (such as child event type, occurrence time, performance data, etc.), combined with the application performance data of the parent node, the child event abnormality coefficient of each child event node is calculated. According to the child event abnormality coefficient, each child event node is marked as abnormal. This helps to accurately locate the source of the problem and provide a basis for subsequent troubleshooting and optimization.

[0151] The technical effect of the above technical solution is: through layer-by-layer analysis (from key event nodes to sub-event nodes), the performance bottleneck or abnormal source in the application can be more accurately located, thereby improving the efficiency of problem solving. The system can collect and analyze application performance data in real time, promptly discover potential performance problems, and avoid the expansion of problems. Based on quantitative data analysis and weight coefficients, the system can more scientifically evaluate the degree of abnormality of event nodes and provide strong support for decision-making. By identifying key event nodes and abnormal sub-event nodes, enterprises can optimize resource allocation in a more targeted manner and improve resource utilization efficiency.

[0152] Timely discovery and resolution of application performance issues can significantly improve user experience and enhance user satisfaction and loyalty.

[0153] In one embodiment of the present invention, the abnormal adjustment judgment module includes:

[0154] An adjustment sub-node acquisition module is used to obtain node information of each abnormal sub-event node, adjust the node information, and obtain an adjusted sub-event abnormality coefficient;

[0155] Obtaining the number of abnormal sub-event nodes according to the adjusted sub-event abnormal coefficient;

[0156] The key node acquisition module is adjusted to determine whether to calculate the corresponding key event abnormality coefficient according to the number of abnormal sub-event nodes;

[0157] The number of abnormal sub-event nodes is compared with a preset number threshold, and when the number of abnormal sub-event nodes is greater than the preset number threshold, the corresponding key event abnormality coefficient is calculated;

[0158] Then, the number of abnormal key event nodes after adjustment is obtained. When there is no abnormal key event node, the calculation and adjustment of the abnormal coefficient is stopped.

[0159] The working principle of the above technical solution is as follows: for each abnormal sub-event node, the system first obtains its node information, which may include the occurrence time, duration, scope of influence, related resource usage, etc. The system adjusts the node information according to the preset rules. After adjustment, the abnormal coefficient of the sub-event node is recalculated to obtain the adjusted sub-event abnormal coefficient. According to the adjusted sub-event abnormal coefficient, the system counts the number of abnormal sub-event nodes under the current key event. The system compares the number of abnormal sub-event nodes with the preset number threshold. This threshold is used to determine whether the current key event is "abnormal" enough to further calculate its key event abnormal coefficient. If the number of abnormal sub-event nodes is greater than the preset number threshold, it means that the key event is likely to have significant performance problems or abnormal behavior, and a deeper analysis is required. When the conditions are met, the system will calculate the corresponding key event abnormal coefficient based on the adjusted sub-event abnormal coefficient and other relevant information (such as key event weight, historical data, etc.). This calculation process may involve complex mathematical models or machine learning algorithms to ensure the accuracy and reliability of the abnormal coefficient. After calculating the abnormal coefficient of the key event, the system may further evaluate or adjust the key event itself based on the coefficient (for example, adjust the weight, reclassify, adjust the operation data, etc.). If necessary, the system will also update the number of abnormal key event nodes based on the adjustment results. If no abnormal key event nodes are found during the inspection process, or the predetermined analysis depth has been reached, the system will stop the calculation and adjustment process of the abnormal coefficient.

[0160] The technical effect of the above technical solution is: through the adjustment of node information and multiple iterative calculations, the system can more accurately evaluate the degree of abnormality of sub-events and key events, and reduce false positives and missed reports. By setting the number threshold, the system can avoid in-depth analysis of each subtle anomaly, thereby saving computing resources and time. The process can be adjusted and optimized according to different application scenarios and needs, such as by modifying the threshold, adjusting the node information adjustment rules, or improving the abnormality coefficient calculation algorithm. The system can collect and analyze data in real time, and dynamically adjust the calculation and annotation results of the abnormality coefficient according to changes in the data to ensure real-time monitoring and response to application performance. By accurately locating abnormal events and sub-events, the system can provide developers and operation and maintenance personnel with more targeted troubleshooting and optimization suggestions, thereby improving the efficiency of problem solving.

[0161] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for analyzing mobile application usage data, characterized in that: The method comprises: S1. Obtain user operation behavior information, classify the behavior information and calculate the behavior weight coefficient, and then calculate the event weight coefficient corresponding to the preset time period to obtain key event nodes and sub-event nodes; S2. Distribute and store event node information to obtain event storage nodes, obtain sub-event behavior paths and key event behavior paths according to the event storage nodes, aggregate the paths, and obtain aggregated paths; S3, calculating the key event abnormality coefficient of the key event node of the application program to obtain the abnormal key event node, and then calculating the sub-event abnormality coefficient to obtain the abnormal sub-event node; S4. Adjust the node information of the abnormal sub-event nodes, obtain the number of abnormal sub-event nodes and the number of abnormal key event nodes after adjustment, determine whether to continue adjustment, and obtain the adjustment result.

2. According to claim 1, a method for analyzing mobile application usage data, characterized in that: The S1 includes: Acquire user operation behavior information, classify the user operation behavior information according to preset behavior categories, and obtain behavior classification information; Perform edge computing on each behavior classification information, and calculate the behavior weight coefficient of each behavior classification information through edge computing method; Obtaining the behavior weight coefficient of each behavior classification information in the same preset time period to calculate the event weight coefficient corresponding to the preset time period; According to the event weight coefficient, the user operation behavior information in the preset time period is annotated with events of corresponding weights to obtain key event nodes, and each behavior classification of the key event node is annotated with sub-events to obtain sub-event nodes.

3. According to claim 1, a method for analyzing mobile application usage data is characterized in that: The S2 includes: Obtain data of multiple sub-event nodes and their key event nodes and upload them to the cloud; Distribute and store each sub-event node of the key event node through the cloud to obtain the event storage node; Connect the sub-event nodes of each behavior classification of the event storage node to obtain the sub-event behavior path of the corresponding behavior classification; Connect the key event nodes of each event storage node to obtain the key event behavior path; Aggregate the sub-event behavior path with the key event behavior path to obtain the aggregate path.

4. According to claim 1, a method for analyzing mobile application usage data, characterized in that: The S3 includes: Obtain application performance data on the aggregation path, and calculate the key event anomaly coefficient of the key event node of the application program based on the application performance data combined with the event weight coefficient; Anomaly marking each key event node according to the key event anomaly coefficient to obtain an abnormal key event node; Obtaining sub-event node information of the abnormal key event node, and calculating the sub-event abnormality coefficient of each sub-event node of the abnormal key event node according to the application performance information combined with the sub-event information; Each sub-event node is annotated as abnormal according to the sub-event abnormal coefficient to obtain an abnormal sub-event node.

5. According to claim 1, a method for analyzing mobile application usage data, characterized in that: The S4 includes: Obtaining node information of each abnormal sub-event node, adjusting the node information, and obtaining an adjusted sub-event abnormality coefficient; Obtaining the number of abnormal sub-event nodes according to the adjusted sub-event abnormal coefficient; Determine whether to calculate the corresponding key event abnormality coefficient according to the number of abnormal sub-event nodes; Then the number of abnormal key event nodes after adjustment is obtained.

6. A mobile application usage data analysis system, characterized in that: The system comprises: The weight node acquisition module is used to obtain user operation behavior information, classify the behavior information and calculate the behavior weight coefficient, and then calculate the event weight coefficient corresponding to the preset time period, and obtain key event nodes and sub-event nodes; The event path acquisition module is used to perform distributed storage of event node information, obtain event storage nodes, obtain sub-event behavior paths and key event behavior paths according to the event storage nodes, aggregate the paths, and obtain aggregated paths; The abnormal calculation and analysis module is used to calculate the key event abnormality coefficient of the key event node of the application program, obtain the abnormal key event node, and then calculate the sub-event abnormality coefficient to obtain the abnormal sub-event node; The abnormal adjustment judgment module is used to adjust the node information of the abnormal sub-event node, obtain the number of abnormal sub-event nodes and the number of abnormal key event nodes after adjustment, determine whether to continue adjustment, and obtain the adjustment result.

7. A mobile application usage data analysis system according to claim 6, characterized in that: The weight node acquisition module includes: A behavior classification module is used to obtain user operation behavior information, classify the user operation behavior information according to preset behavior types, and obtain behavior classification information; A classification weight calculation module is used to perform edge calculation on each behavior classification information and calculate the behavior weight coefficient of each behavior classification information by edge calculation method; An event weight calculation module is used to obtain the behavior weight coefficient of each behavior classification information in the same preset time period and calculate the event weight coefficient corresponding to the preset time period; The event node acquisition module is used to mark the user operation behavior information in a preset time period with events of corresponding weights according to the event weight coefficient to obtain key event nodes, and to mark each behavior classification of the key event nodes with sub-events to obtain sub-event nodes.

8. A mobile application usage data analysis system according to claim 6, characterized in that: The event path acquisition module includes: A distributed storage module is used to obtain data of multiple sub-event nodes and their key event nodes and upload them to the cloud; Distribute and store each sub-event node of the key event node through the cloud to obtain the event storage node; A path generation module is used to connect the sub-event nodes of each behavior classification of the event storage node to obtain the sub-event behavior path of the corresponding behavior classification; Connect the key event nodes of each event storage node to obtain the key event behavior path; The path aggregation module is used to aggregate the sub-event behavior path with the key event behavior path to obtain an aggregated path.

9. A mobile application usage data analysis system according to claim 6, characterized in that: The abnormal calculation analysis module includes: A key anomaly calculation module is used to obtain application performance data on the aggregation path, and calculate the key event anomaly coefficient of the key event node of the application program based on the application performance data combined with the event weight coefficient; A key node anomaly acquisition module is used to annotate each key event node according to the key event anomaly coefficient to obtain an abnormal key event node; A sub-anomaly calculation module, used to obtain sub-event node information of the abnormal key event node, and calculate the sub-event anomaly coefficient of each sub-event node of the abnormal key event node according to the application performance information combined with the sub-event information; The sub-node anomaly acquisition module is used to annotate each sub-event node according to the sub-event anomaly coefficient to obtain an abnormal sub-event node.

10. A mobile application usage data analysis system according to claim 6, characterized in that: The abnormal adjustment judgment module includes: An adjustment sub-node acquisition module is used to obtain node information of each abnormal sub-event node, adjust the node information, and obtain an adjusted sub-event abnormality coefficient; Obtaining the number of abnormal sub-event nodes according to the adjusted sub-event abnormal coefficient; The key node acquisition module is adjusted to determine whether to calculate the corresponding key event abnormality coefficient according to the number of abnormal sub-event nodes; Then the number of abnormal key event nodes after adjustment is obtained.

Images