Data processing method and apparatus

Abstract

Embodiments of the present application disclose a data processing method and device. The main technical scheme comprises: based on user behavior sequence data, a user behavior path graph is constructed, the nodes in the user behavior path graph are used to represent each function module in the target application program, the edges in the user behavior path graph are used to represent the jump path of the user between each function module, the attributes of the nodes include the access behavior metric value of the corresponding function module, and the attributes of the edges include the jump behavior metric value of the user between the corresponding function modules; the user behavior path graph is analyzed to determine the abnormal node in the user behavior path graph; and the data of the user behavior path graph and the data of the abnormal node are sent to the user terminal to enable the user terminal to display the user behavior path graph and the abnormal node in a visual manner. The present application can effectively identify the abnormal function existing in the target application program.

Description

Technical Field

[0001] This application relates to the field of Internet application technology, and in particular to a data processing method and apparatus. Background Technology

[0002] When analyzing user behavior data from applications, the focus is typically on the number of times users access each functional module to understand how users interact with them. However, simply analyzing module access volume is insufficient to effectively identify abnormal functionalities within an application. Therefore, effectively identifying abnormal functionalities within an application has become a pressing technical challenge. Summary of the Invention

[0003] In view of this, this application provides a data processing method and apparatus to effectively identify abnormal functions in a target application.

[0004] This application provides the following solution: Firstly, a data processing method is provided, applied to the server side. This data processing method includes: constructing a user behavior path graph based on user behavior sequence data, where nodes in the user behavior path graph represent functional modules in the target application, edges in the user behavior path graph represent the user's jump path between functional modules, node attributes include access behavior metrics of the corresponding functional module, and edge attributes include jump behavior metrics of the user between the corresponding functional modules; analyzing the user behavior path graph to identify abnormal nodes in the user behavior path graph; and sending the user behavior path graph data and abnormal node data to the user terminal so that the user terminal can visualize the user behavior path graph and abnormal nodes.

[0005] Optionally, before constructing the user behavior path map based on the user behavior sequence data, the data processing method further includes: in response to detecting the presence of discontinuous data in the user behavior data, concatenating the discontinuous data to target continuous data related to the discontinuous data in the user behavior data to obtain user behavior sequence data.

[0006] Optionally, concatenating discontinuous data with target continuous data related to the discontinuous data in the user's behavior data to obtain user behavior sequence data includes: determining multiple candidate continuous data in the user's behavior data based on preset rules; determining target continuous data based on the matching degree between each candidate continuous data and the discontinuous data; concatenating the discontinuous data with the target continuous data to obtain user behavior sequence data; wherein the preset rules include at least one of the following: the time interval between the generation time of the discontinuous data and the generation time of the candidate continuous data is less than a first time interval threshold; the similarity between the task object corresponding to the discontinuous data and the task object corresponding to the candidate continuous data is greater than a similarity threshold; the continuity between the first operation event corresponding to the discontinuous data and the second operation event corresponding to the candidate continuous data is greater than a continuity threshold; the matching degree between the candidate continuous data and the discontinuous data is determined by at least one of the following: the time interval between the generation time of the discontinuous data and the generation time of the candidate continuous data, the similarity between the task object corresponding to the discontinuous data and the task object corresponding to the candidate continuous data, and the continuity between the first operation event corresponding to the discontinuous data and the second operation event corresponding to the candidate continuous data.

[0007] Optionally, detecting discontinuous data in user behavior data includes at least one of the following: determining that discontinuous data exists in user behavior data in response to detecting that the session identifier information corresponding to multiple consecutive behavior data of the same user is different; determining that discontinuous data exists in user behavior data in response to detecting that the device identifiers corresponding to multiple behavior data of the user are different, and the time interval between the generation times of the multiple behavior data is less than a second time interval threshold and / or the task objects corresponding to the multiple behavior data are the same; determining that discontinuous data exists in user behavior data in response to detecting an event in user behavior data where the user closes a function module and then restarts the function module within a first preset time period; or determining that discontinuous data exists in user behavior data in response to detecting an event in user behavior data where the response duration is greater than a second preset time period.

[0008] Optionally, after concatenating discontinuous data to target continuous data related to the discontinuous data in the user's behavior data to obtain user behavior sequence data, the data processing method further includes: adding concatenation identification information to the user behavior sequence data, the concatenation identification information being used to identify the confidence level of the user behavior sequence data; and constructing a user behavior path graph based on the user behavior sequence data, including: in response to detecting that the confidence level of the user behavior sequence data is greater than a confidence threshold, constructing a user behavior path graph based on the user behavior sequence data.

[0009] Optionally, the user behavior path graph is analyzed to determine abnormal nodes in the user behavior path graph, including: determining abnormal nodes in the user behavior path graph based on at least one of the following: access behavior metric value corresponding to the node in the user behavior path graph, connectivity of the node in the user behavior path graph, and jump behavior metric value of the edge in the user behavior path graph.

[0010] Optionally, based on at least one of the following: access behavior metrics corresponding to nodes in the user behavior path graph, connectivity of nodes in the user behavior path graph, and jump behavior metrics of edges in the user behavior path graph, abnormal nodes in the user behavior path graph are determined, including: if the access volume corresponding to the first node is greater than or equal to the access volume threshold and the access duration is less than or equal to the access duration threshold, then the first node is determined to be an abnormal node; if the second node is a node corresponding to a non-endpoint functional module and there is no successor node of the second node in the target user behavior path where the second node is located, then the second node is determined to be an abnormal node; if the out-degree of the third node is greater than or equal to a preset out-degree threshold and the jump frequency corresponding to the outgoing edge of the third node is less than or equal to the preset out-degree threshold, then the abnormal node is determined to be an abnormal node. If the jump frequency of the third node is equal to the first jump frequency threshold, then the third node is determined to be an abnormal node. The out-degree of the third node is used to represent the number of edges originating from the third node, and the out-edges of the third node are used to represent the edges originating from the third node. If the in-degree of the fourth node is greater than or equal to the preset in-degree threshold and the jump frequency corresponding to the in-edge of the fourth node is less than or equal to the second jump frequency threshold, then the fourth node is determined to be an abnormal node. The in-degree of the fourth node is used to represent the number of edges pointing to the fourth node, and the in-edges of the fourth node are used to represent the edges pointing to the fourth node. If the fifth node is an edgeless node and the access behavior metric corresponding to the fifth node is greater than or equal to the preset metric threshold, then the fifth node is determined to be an abnormal node.

[0011] Optionally, it also includes: determining an optimization strategy for the abnormal node based on the abnormal node's abnormality type and the task description information of the corresponding functional module.

[0012] Optionally, constructing a user behavior path map based on user behavior sequence data includes: constructing a user behavior path map for each user type based on the user behavior sequence data of each user type according to a preset user type; the data processing method also includes: adjusting the functional modules in the target application of the user type based on the user behavior path map of the user type.

[0013] Optionally, the user behavior path graph is analyzed to identify abnormal nodes in the user behavior path graph, including: analyzing the user behavior path graph of users corresponding to each user type to identify abnormal nodes in the user behavior path graph of users corresponding to each user type; and sending the user behavior path graph data and abnormal node data to the user terminal so that the user terminal can visualize the user behavior path graph and abnormal nodes, including: sending the user behavior path graph data of users corresponding to each user type and the abnormal node data in the user behavior path graph of users corresponding to each user type to the user terminal so that the user terminal can visualize the user behavior path graph of users corresponding to each user type and the abnormal nodes in the user behavior path graph of users corresponding to each user type.

[0014] Optionally, the method further includes: analyzing the user behavior path graph to determine a first type of path and a second type of path corresponding to completing the same target task; using the first type of path to generate an optimization strategy for the second type of path; wherein the first type of path satisfies at least one of the following: the completion rate of the target task is greater than or equal to a completion rate threshold, the time to complete the target task is less than or equal to a preset time threshold, and the proportion of abnormal nodes in the path is less than or equal to a preset proportion value; the second type of path satisfies at least one of the following: the completion rate of the target task is less than a completion rate threshold, the time to complete the target task is greater than a preset time threshold, and the proportion of abnormal nodes in the path is greater than a preset proportion value.

[0015] Secondly, a data processing method is provided for application on the user end. This data processing method includes: acquiring data from a user behavior path graph and data from abnormal nodes in the user behavior path graph. The user behavior path graph is constructed based on user behavior sequence data. Nodes in the user behavior path graph represent functional modules in the target application, edges in the user behavior path graph represent the user's jump paths between functional modules, and abnormal nodes are determined by analyzing the user behavior path graph. On the user interface, nodes in the user behavior path graph are displayed using a first style, edges in the user behavior path graph are displayed using a second style, and abnormal nodes are displayed using a third style. The first style represents the user's access behavior metrics for each functional module, and the second style represents the user's jump behavior metrics between functional modules.

[0016] Optionally, the nodes in the user behavior path graph are displayed using a first style, including: using the color of the node to represent the duration of the user's access to the functional module, and / or using the size of the node to represent the number of times the user accesses the functional module; the edges in the user behavior path graph are displayed using a second style, including: using the size of the edge to represent the frequency of the user's jump between various functional modules.

[0017] Optionally, it further includes: in response to an event that detects a selected node in the user behavior path graph, displaying at least one of the following: functional description information of the functional module corresponding to the node, the value of the access behavior metric corresponding to the node, the associated nodes of the node, and the value of the jump behavior metric between the associated nodes of the node; and / or, in response to an event that detects a selected edge in the user behavior path graph, displaying the value of the jump behavior metric corresponding to the edge and / or the associated nodes of the edge.

[0018] Thirdly, a data processing device is provided, applied on a server side. This data processing device includes: a construction unit configured to construct a user behavior path graph based on user behavior sequence data, wherein nodes in the user behavior path graph represent functional modules in a target application, edges in the user behavior path graph represent user jump paths between functional modules, node attributes include access behavior metrics of the corresponding functional module, and edge attributes include user jump behavior metrics between the corresponding functional modules; an analysis unit configured to analyze the user behavior path graph and identify abnormal nodes in the user behavior path graph; and a sending unit configured to send the user behavior path graph data and abnormal node data to the user terminal for the user terminal to visualize the user behavior path graph and abnormal nodes.

[0019] Fourthly, a data processing device is provided for use on a user terminal. The data processing device includes: an acquisition unit configured to acquire data of a user behavior path graph and data of abnormal nodes in the user behavior path graph. The user behavior path graph is constructed based on user behavior sequence data. Nodes in the user behavior path graph represent functional modules in a target application, and edges in the user behavior path graph represent the user's jump paths between functional modules. Abnormal nodes are determined through analysis of the user behavior path graph. A display unit configured to display nodes in the user behavior path graph on a user interface using a first style, edges in the user behavior path graph using a second style, and abnormal nodes in a third style. The first style represents the user's access behavior metrics to each functional module, and the second style represents the user's jump behavior metrics between functional modules.

[0020] Fifthly, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed, implements the steps of the data processing method described in the first or second aspect.

[0021] A sixth aspect provides an electronic device comprising: one or more processors; and a memory associated with the one or more processors, the memory being used to store program instructions that, when read and executed by the one or more processors, perform the steps of the data processing method of the first or second aspect described above.

[0022] In a seventh aspect, a computer program product is provided, comprising a computer program that, when executed, implements the steps of the data processing method described in the first or second aspect.

[0023] According to the specific embodiments provided in this application, the following technical effects are disclosed: 1) This application constructs a user behavior path graph that simultaneously reflects access behavior metrics and redirection behavior metrics, and analyzes the user behavior path graph to identify abnormal nodes. This allows the visualization to simultaneously present the usage of functional modules and the flow between functional modules, and to locate abnormal functional modules in the target application. This effectively identifies abnormal functions in the target application, thereby improving the efficiency and accuracy of optimizing the target application.

[0024] 2) By detecting and splicing discontinuous data, this application can improve the integrity of user behavior sequence data, so that the user behavior path map constructed subsequently can fully reflect the user's behavior chain, thereby improving the data integrity and data validity of the user behavior path map.

[0025] 3) By introducing preset rules and matching degree calculations based on dimensions such as time interval, task object similarity and operation event continuity, this application can improve the accuracy of determining target continuous data from multiple candidate continuous data, thereby improving the integrity, accuracy and reliability of the spliced ​​user behavior sequence data.

[0026] 4) This application detects discontinuous data in user behavior data through multiple methods, and can automatically and comprehensively identify behavior data breaks caused by various reasons such as session changes, device switching, brief exits, and response timeouts, thereby improving the accuracy and comprehensiveness of subsequent spliced ​​user behavior sequence data.

[0027] 5) By adding confidence indicators to the spliced ​​user behavior sequence data and filtering low-confidence user behavior sequence data when constructing the user behavior path map, this application can effectively reduce unreliable data introduced by incorrect splicing, thereby ensuring the data quality of the constructed user behavior path map and improving the accuracy of subsequent analysis of the user behavior path map.

[0028] 6) This application identifies abnormal nodes by combining multiple dimensions such as node access behavior metrics, node connectivity, and edge jump behavior metrics, thereby achieving multi-dimensional anomaly detection of user behavior path graphs and improving the comprehensiveness and accuracy of abnormal node identification.

[0029] 7) This application can accurately identify typical abnormal functions such as high traffic but poor conversion, path interruption, high out-degree and low jump, high in-degree and low jump, and isolated high access by using the access volume and access duration of the node, the connectivity of the node, the out-degree and jump frequency of the out-edge of the starting node, the in-degree and jump frequency of the pointing node, and edgeless nodes and their corresponding access behavior metrics. This can distinguish different types of abnormal bottlenecks and further improve the accuracy of the targeted optimization measures generated subsequently.

[0030] 8) By associating the anomaly type of the abnormal node with the corresponding task description information, this application can generate targeted and implementable functional optimization strategies, realizing a closed loop from behavior monitoring to insight output, and greatly improving the efficiency of generating optimization decisions.

[0031] 9) This application constructs user behavior path diagrams for users according to user type, and adjusts the functional modules in the target application used by users of different user types accordingly. This achieves precise stratification and differentiated optimization of user behavior for users of different user types, and enables the formulation of optimization strategies for users of different user types, thereby improving the effectiveness of optimization strategies.

[0032] 10) This application identifies and compares the first type of path (i.e., efficient path) and the second type of path (i.e. inefficient path) when completing the same target task, thereby using the first type of path as a reference to generate an optimization strategy for the second type of path, which can effectively simplify the user's operation when completing the target task and improve the user's completion rate and efficiency of the target task.

[0033] 11) This application reflects the user's access behavior metrics to the corresponding functional modules through the style of nodes, and reflects the user's jumping behavior metrics between functional modules through the style of edges. It can realize the rapid acquisition of the breadth, depth, etc. of functional modules and the flow frequency between functional modules by using visual perception, which significantly improves the efficiency of identifying the overall distribution and flow characteristics of user behavior; and by using different styles to display abnormal nodes, it can highlight abnormal nodes in the user behavior path graph, which improves the efficiency of locating abnormal functional modules in the target application.

[0034] 12) By interacting with nodes or edges in the user behavior path graph, this application can instantly obtain the precise measurement value and contextual relationship of the selected element, thereby simplifying the operation of obtaining relevant information from the user behavior path graph and improving the readability of the user behavior path graph.

[0035] Of course, any product implementing this application does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0036] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0037] Figure 1 This is a schematic diagram of the system architecture applicable to the embodiments of this application.

[0038] Figure 2 This is a schematic flowchart illustrating a data processing method provided in an embodiment of this application.

[0039] Figure 3 Another illustrative flowchart of the data processing method provided in the embodiments of this application.

[0040] Figure 4 This is a schematic diagram illustrating the display of a user behavior path graph on a user interface, as provided in an embodiment of this application.

[0041] Figure 5 This is a schematic block diagram of a data processing apparatus provided in an embodiment of this application.

[0042] Figure 6 Another schematic block diagram of the data processing apparatus provided in the embodiments of this application.

[0043] Figure 7 A schematic block diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0044] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0045] The terminology used in the embodiments of this application is for the purpose of describing particular embodiments only and is not intended to be limiting of this application. The singular forms “a,” “the,” and “the” used in the embodiments of this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise.

[0046] It should be understood that the term "and / or" used in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0047] Depending on the context, the word "if" as used here can be interpreted as "when," "when," "in response to determination," or "in response to detection." Similarly, depending on the context, the phrase "if determination" or "if detection (of the stated condition or event)" can be interpreted as "when determination," "in response to determination," "when detection (of the stated condition or event)," or "in response to detection (of the stated condition or event)."

[0048] When analyzing user behavior, a common approach to quantitatively assess the user experience of each functional module within an application is to analyze the access volume of those modules. This analysis evaluates user engagement with each module and identifies any abnormal modules within the application. However, the inventors discovered that simply relying on user access volume is insufficient to accurately identify abnormal functional modules. By using multi-dimensional data, such as user access volume, access duration, and the frequency of user navigation between modules, it becomes possible to effectively identify and quickly locate abnormal functional modules.

[0049] In view of this, this application provides a new approach. To facilitate understanding of this application, the system architecture on which this application is based will first be described. Figure 1 This is a schematic diagram of the system architecture applicable to the embodiments of this application, such as... Figure 1 As shown, the system architecture may include: user terminal, terminal device and server terminal.

[0050] In this embodiment, the user terminal is set on the terminal device. The user terminal involved in this application can be a client running on the terminal device, a mini-program, or a web application running through a browser.

[0051] Terminal devices can include, but are not limited to, smart mobile terminals, wearable devices, PCs (Personal Computers), and smart home devices. Smart mobile devices can include devices such as mobile phones, tablets, laptops, PDAs (Personal Digital Assistants), and connected car terminals. Wearable devices can include devices such as smartwatches, smart glasses, smart bracelets, VR (Virtual Reality) devices, AR (Augmented Reality) devices, and mixed reality devices (devices that support both virtual and augmented reality). Smart home devices can include devices such as smart TVs and smart refrigerators with displays.

[0052] The server can be a single server, a server cluster consisting of multiple servers, or a cloud server. A cloud server, also known as a cloud computing server or cloud host, is a host product in the cloud computing service system, designed to address the shortcomings of traditional physical hosts and Virtual Private Servers (VPS) services, such as high management difficulty and weak service scalability.

[0053] One feasible approach is for the server to obtain user behavior sequence data through a user behavior log system and construct a user behavior path graph based on this data. Nodes in the path graph represent functional modules within the target application, and edges represent the user's navigation paths between these modules. Node attributes can include access behavior metrics for the corresponding functional module, and edge attributes can include navigation behavior metrics for the user between those modules. The server can also analyze the path graph to identify abnormal nodes and send the path graph data and abnormal node data to the user terminal. The user terminal can send a request to the server to display the path graph and abnormal nodes, receive the data from the server, and render the path graph and abnormal nodes on the terminal device's screen. The nodes in the path graph can be displayed using a first style, the edges using a second style, and the abnormal nodes using a third style.

[0054] The method involved in this application allows the construction and analysis of the user behavior path graph to be completed by the server side. The server side then sends the user behavior path graph data and abnormal node data to the user side, and the user side displays the user behavior path graph and abnormal nodes.

[0055] It should be understood that Figure 1The number of client terminals, terminal devices, and servers shown is merely illustrative. Depending on implementation needs, there can be any number of client terminals, terminal devices, and servers.

[0056] Figure 2 A schematic flowchart illustrating a data processing method provided in an embodiment of this application, the method can be performed by... Figure 1 The server-side execution in the system shown. For example... Figure 2 As shown, the data processing method may include the following steps: Step 201: Based on user behavior sequence data, construct a user behavior path graph. Nodes in the user behavior path graph represent functional modules in the target application, and edges in the user behavior path graph represent the user's jump path between functional modules. The attributes of nodes include the access behavior metric of the corresponding functional module, and the attributes of edges include the jump behavior metric of the user between the corresponding functional modules.

[0057] Step 202: Analyze the user behavior path graph to identify abnormal nodes in the user behavior path graph.

[0058] Step 203: Send the user behavior path graph data and abnormal node data to the user terminal so that the user terminal can visualize the user behavior path graph and abnormal nodes.

[0059] As can be seen from the above process, this application constructs a user behavior path graph that simultaneously reflects access behavior metrics and redirection behavior metrics, and analyzes the user behavior path graph to identify abnormal nodes. This allows the visualization to simultaneously present the usage of functional modules and the flow between functional modules, and to locate abnormal functional modules in the target application. This effectively identifies abnormal functions in the target application, thereby improving the efficiency and accuracy of optimizing the target application.

[0060] The following describes in detail each step of the above process and the effects that can be further produced, with reference to the embodiments.

[0061] First, in conjunction with the embodiments, the above step 201, namely "constructing a user behavior path graph based on user behavior sequence data, wherein the nodes in the user behavior path graph are used to represent the functional modules in the target application, the edges in the user behavior path graph are used to represent the user's jump path between the functional modules, the attributes of the nodes include the access behavior metric value of the corresponding functional module, and the attributes of the edges include the jump behavior metric value of the user between the corresponding functional modules", will be described in detail.

[0062] User behavior path graphs are used to represent user behavior through a graph structure. A user behavior path graph includes multiple nodes and edges connecting the nodes.

[0063] In the user behavior path diagram, nodes represent functional modules within the target application, such as "Promotion Management," "Batch Creation," "Standard Creation," and "Targeted Templates" in an advertising application. Each node's attributes include access behavior metrics for the corresponding functional module. These metrics might include visit count (representing the cumulative number of times a user visits the module, such as page views (PV), unique visitors (UV, representing the number of different users accessing the module, used to measure the module's user reach), average number of visits per user, time on page (representing the average time a user spends within the module, used to measure the depth of user engagement with the module), and average time spent on page. By utilizing these node attributes, the breadth of functional module usage and the depth of user engagement can be simultaneously presented when visualizing the user behavior path diagram.

[0064] The edges in the user behavior path graph represent the user's jump paths between various functional modules, that is, the user's behavioral trajectory from one functional module to another. Each edge's attributes include metrics related to the user's jump behavior between the functional modules corresponding to that edge. These metrics may include jump frequency (the total number of times a user jumps between the functional modules corresponding to that edge within a specific time period), jump count (the total number of times a user jumps between the functional modules corresponding to that edge), flow rate (the proportion of jumps from the starting node to the pointing node relative to the total number of jumps from all outgoing edges of that starting node, used to measure the user's preference for a specific behavioral path among the starting nodes of that edge), average jump duration (the average time a user takes to jump from the starting node to the pointing node), and jump success rate (the proportion of successful jumps from the starting node to the pointing node relative to the total number of jumps), etc.

[0065] It is understandable that the attributes of nodes and edges in the user behavior path graph can be flexibly set according to actual needs, and may include, but are not limited to, the metric types given above. The above-mentioned multiple types of metrics together constitute a multi-dimensional description of user behavior, which significantly improves the comprehensive perception and identification capabilities of user behavior, flow patterns, and experience bottlenecks (bottleneck situations that occur when users use various functional modules, i.e., abnormal nodes).

[0066] To construct a user behavior path map, it is necessary to obtain user behavior sequence data, which can be obtained through collection and cleaning.

[0067] During the data collection phase, full-tracking tools and / or large form module behavior collection SDKs can be used to cover user behavior data across all functional modules of the target application. This behavior data records the event type (such as click, pause, redirect, submit, error, exit, etc.), timestamp, functional module identifier, user identifier, operation parameters, and operation results for each user operation. During collection, user identity identifiers and user types (such as user tier tags like large customers and small and medium-sized customers) can be automatically associated, and data can be reported to the data platform in real time. This supports millisecond-level data transmission in high-concurrency scenarios, ensuring data integrity and timeliness.

[0068] During the cleaning phase, preprocessing can be performed on the raw behavioral data, such as deduplication (removing duplicate data), completion (completing or marking data with missing key fields such as user ID and functional module ID), invalid data filtering (filtering abnormal behavioral data such as test users, crawler requests, and abnormal operations such as violent refresh and repeated clicks), and user type alignment (associating user type with behavioral data), to form a four-dimensional dataset in the form of [user type-behavioral data-functional module-timestamp].

[0069] The obtained behavioral data can be sorted based on user identifiers and timestamps to obtain user behavior sequence data. This user behavior sequence data reflects the complete link of a user's access to various functional modules in chronological order over a period of time. Based on this user behavior sequence data, and using each functional module as a node and the user's movement between functional modules as directed edges, a user behavior path graph can be constructed.

[0070] To further optimize data integrity and address the issue that discontinuity in user behavior data due to session interruptions, device switching, or other reasons affects the accuracy of the generated user behavior path map, in one embodiment of this application, before constructing the user behavior path map based on user behavior sequence data, in response to the detection of discontinuous data in the user behavior data, the discontinuous data is spliced ​​into the target continuous data related to the discontinuous data in the user behavior data to obtain user behavior sequence data.

[0071] Discontinuous data refers to data segments with breakpoints in user behavior data. Examples include changes in session identifiers due to network fluctuations, users switching devices to perform the same task, and users briefly exiting and then re-entering a functional module. By splicing discontinuous data into relevant continuous data within the user's behavior data when it is detected, the integrity of the user behavior sequence data can be significantly improved. This ensures that the final user behavior path graph constructed based on the user behavior sequence data reflects the complete user behavior chain, further enhancing the accuracy and reliability of identifying abnormal nodes using the user behavior path graph.

[0072] In one embodiment of this application, multiple methods for detecting discontinuous data are provided, that is, detecting the presence of discontinuous data in user behavior data, including at least one of the following: determining that discontinuous data exists in user behavior data in response to detecting that the session identifier information corresponding to multiple consecutive behavior data of the same user is different; determining that discontinuous data exists in user behavior data in response to detecting that the device identifiers corresponding to multiple behavior data of the user are different, and the time interval between the generation times of the multiple behavior data is less than a second time interval threshold and / or the task objects corresponding to the multiple behavior data are the same; determining that discontinuous data exists in user behavior data in response to detecting an event in user behavior data where the user closes a function module and then restarts the function module within a first preset time period; and determining that discontinuous data exists in user behavior data in response to detecting an event in user behavior data where the response duration is greater than a second preset time period.

[0073] The session ID is typically refreshed when a user closes a session, the session times out, or a session switches. If inconsistent Session IDs are detected across multiple consecutive behavioral data entries under the same user ID, a breakpoint is marked, indicating the presence of discontinuous data in the user's behavioral data. For example, user A is performing a task in a target application (this process corresponds to a Session ID), and suddenly the target application crashes and restarts (generating a new Session ID). If only the Session ID is used, it would be assumed that the original Session ID and the new Session ID correspond to user behaviors from different tasks, when in fact they belong to the same user behavior within the same task. Therefore, in this case, discontinuous data is considered to exist in the user's behavioral data.

[0074] In scenarios involving device or platform switching, if it is detected that multiple pieces of user behavior data under the same user ID have changed device IDs, and the time interval before and after the change is short, and the task object is consistent (e.g., both before and after the device ID change are for the same product purchase task), then a breakpoint is marked, indicating the presence of discontinuous data in the user's behavior data. For example, a user browses products and adds them to their cart in a target application on their mobile device, then logs into the target application on their computer and completes the payment for that product. In this case, the user's behavior data on the mobile and computer devices belongs to different device IDs, but is for the same user and the same task. Therefore, in this situation, the user's behavior data is considered to contain discontinuous data.

[0075] If a user closes a function module and then reopens it within a first preset time period, a breakpoint is marked, indicating that there is discontinuous data in the user's behavior data. For example, while performing a purchase task, a user finds insufficient funds when making payment on the payment page, closes the payment page to top up, and then reopens the payment page within a first preset time period (e.g., 2 minutes) to continue the payment. This re-entry behavior is usually strongly correlated with the task before closing the payment; therefore, in this case, the user's behavior data is considered to contain discontinuous data.

[0076] If a user's action is detected as unresponsive for an extended period, a breakpoint is marked, indicating that discontinuous data exists within the user's behavioral data. For example, when a user fills out a complex form on a form-filling page and clicks the save button, but the response times out, the user waits for a while, refreshes the form-filling page, starts filling out the form again, and saves successfully. In this case, although there are large gaps in the user's behavioral data before and after the wait, they belong to the same task. Therefore, in this situation, the user's behavioral data is considered to contain discontinuous data.

[0077] By using the above methods, more types of discontinuous data can be identified in user behavior data, thereby significantly improving the integrity and reliability of the spliced ​​user behavior sequence data, further enhancing the integrity and accuracy of the constructed user behavior path map, and providing an accurate data foundation for the subsequent identification of abnormal nodes.

[0078] In another embodiment of this application, non-continuous data can be spliced ​​into target continuous data related to the non-continuous data in the user's behavior data based on preset rules and matching degree calculation, so as to obtain user behavior sequence data. That is, concatenating discontinuous data with target continuous data related to the discontinuous data in the user's behavior data to obtain user behavior sequence data includes: determining multiple candidate continuous data in the user's behavior data based on preset rules; determining target continuous data according to the matching degree between each candidate continuous data and the discontinuous data; concatenating the discontinuous data with the target continuous data to obtain user behavior sequence data; wherein, the preset rules include at least one of the following: the time interval between the generation time of the discontinuous data and the generation time of the candidate continuous data is less than a first time interval threshold; the similarity between the task object corresponding to the discontinuous data and the task object corresponding to the candidate continuous data is greater than a similarity threshold; the continuity between the first operation event corresponding to the discontinuous data and the second operation event corresponding to the candidate continuous data is greater than a continuity threshold; the matching degree between the candidate continuous data and the discontinuous data is determined by at least one of the following: the time interval between the generation time of the discontinuous data and the generation time of the candidate continuous data, the similarity between the task object corresponding to the discontinuous data and the task object corresponding to the candidate continuous data, and the continuity between the first operation event corresponding to the discontinuous data and the second operation event corresponding to the candidate continuous data.

[0079] For each detected non-continuous data point, all behavioral data of the same user can be traversed to filter out candidate connected data that satisfy at least one preset rule. The preset rule may include a time proximity rule, meaning the time interval between the generation time of the non-continuous data and the generation time of the candidate continuous data is less than a first time interval threshold. For example, if the first time interval threshold is 10 minutes and the non-continuous data was generated at 10:00, then behavioral data between 9:55 and 10:05 can be considered candidate continuous data, while behavioral data after 16:00 can be considered candidate continuous data.

[0080] The preset rules can also include task object similarity rules, that is, the similarity between the task object corresponding to non-continuous data and the task object corresponding to candidate continuous data is greater than the similarity threshold. For example, when a user performs a product purchase task on the computer, the network fluctuates, and the order ID of the product purchase task is "12345". Then the user switches to the mobile phone to continue performing the product purchase task, and the order ID on the mobile phone is also "12345". Since the task objects are completely identical, the user's behavior data on the mobile phone can be used as candidate continuous data.

[0081] The preset rules can also include operation event continuity rules, that is, the degree of continuity between the first operation event corresponding to non-continuous data and the second operation event corresponding to candidate continuous data is greater than the continuity threshold. For example, if the user operation sequence is [click upload] -> [report error] -> [blank], which generates non-continuous data, if the user's operation sequence of [blank] -> [click retry] is detected, since clicking retry is the most common continuous operation after reporting an error, the degree of continuity is extremely high. Therefore, the behavioral data corresponding to the operation sequence of [blank] -> [click retry] can be used as candidate continuous data.

[0082] Using the aforementioned preset rules, candidate continuous data that may be associated with non-continuous data can be initially screened. To further identify the target continuous data most relevant to the non-continuous data among the candidate continuous data, a matching degree calculation can be performed on each candidate continuous data. Specifically, the shorter the time interval between the generation time of the non-continuous data and the generation time of the candidate continuous data, the higher the matching degree between the non-continuous data and the candidate continuous data; the greater the similarity between the task object corresponding to the non-continuous data and the task object corresponding to the candidate continuous data, the higher the matching degree between the non-continuous data and the candidate continuous data; and the higher the continuity between the first operation event corresponding to the non-continuous data and the second operation event corresponding to the candidate continuous data, the higher the matching degree between the non-continuous data and the candidate continuous data.

[0083] Optionally, the matching degree can be calculated based on a weighted summation of multiple dimensions, and the candidate continuous data with the highest matching degree can be selected as the target continuous data to be concatenated with the non-continuous data to obtain user behavior sequence data.

[0084] The above methods significantly improve the accuracy and completeness of the spliced ​​user behavior sequence data, further enhancing the completeness and accuracy of the constructed user behavior path graph, and providing an accurate data foundation for subsequent identification of abnormal nodes. Furthermore, the preset rules enhance the interpretability and controllability of the spliced ​​user behavior sequence data, facilitating retrospective review of the splicing criteria.

[0085] In another embodiment of this application, to address the problem of user behavior path graph distortion caused by unreliable data introduced due to data splicing errors, confidence management can be applied to the spliced ​​user behavior sequence data. That is, after splicing discontinuous data to target continuous data related to the discontinuous data in the user's behavior data to obtain user behavior sequence data, splicing identification information is added to the user behavior sequence data. This splicing identification information is used to identify the confidence level of the user behavior sequence data. Based on the user behavior sequence data, a user behavior path graph is constructed, including: in response to detecting that the confidence level of the user behavior sequence data is greater than a confidence threshold, constructing the user behavior path graph based on the user behavior sequence data.

[0086] The splicing identification information may include the following: Whether the data has been spliced ​​or not. The confidence level of spliced ​​user behavior sequence data is lower than that of unspliced ​​user behavior sequence data.

[0087] The degree of splicing, a value from 0 to 1 (or 0 to 100), represents the proportion of spliced ​​data in the user behavior sequence data. The higher the degree of splicing, the lower the confidence level of the user behavior sequence data.

[0088] The splicing criteria type records the preset rules used in this splicing process. The more preset rules used in the splicing, the higher the confidence level of the user behavior sequence data.

[0089] By using this splicing identifier information, the confidence level of user behavior sequence data can be determined. For example, the three types of information mentioned above can be weighted and summed to determine the confidence level of the user behavior sequence. The confidence level of the spliced ​​user behavior sequence data is quantified by using the splicing identifier information, and a confidence threshold is set. Only high-confidence user behavior sequence data with a confidence level higher than this threshold is allowed to enter the subsequent user behavior path graph construction process, filtering out low-confidence spliced ​​user behavior sequence data. This ensures the reliability of the data used in user behavior path graph construction and further improves the accuracy of subsequent abnormal node identification.

[0090] The following describes step 202, namely "analyzing the user behavior path graph and identifying abnormal nodes in the user behavior path graph", in detail with reference to the embodiments.

[0091] Here, abnormal nodes can be understood as points of difficulty in the user's experience when using the target application. Analyzing the user behavior path graph to identify abnormal nodes can be done based on pre-set analysis rules or by using machine learning or graph neural networks (GNNs) for in-depth analysis of the user behavior path graph.

[0092] In one embodiment of this application, analyzing a user behavior path graph to determine abnormal nodes in the user behavior path graph includes: determining abnormal nodes in the user behavior path graph based on at least one of the following: access behavior metric value corresponding to a node in the user behavior path graph, connectivity of the node in the user behavior path graph, and jump behavior metric value of the edge in the user behavior path graph.

[0093] Among them, access behavior metrics are used to describe how users use the corresponding functional modules through multi-dimensional data. Some abnormal nodes can be identified by the relative relationship between two or more access behavior metrics of the node itself. For example, if the functional module corresponding to a certain node has high access volume and low dwell time, it can be considered that the functional module corresponding to that node has problems with misleading entry points or insufficient content value.

[0094] The connectivity of a node in the user path graph describes the structural characteristics of the node's connections with other nodes. A normal functional module should have nodes in the user behavior path graph with out-degree (reachable downstream functional modules) and in-degree (originating upstream functional modules) that match its function and hierarchy. Violation of this connectivity characteristic may indicate an anomaly in the node.

[0095] Redirect behavior metrics are used to describe the user's flow between the functional modules corresponding to an edge using multi-dimensional data. Edge redirection behavior metrics can reflect the flow between nodes. When a node appears normal (e.g., high traffic), but its associated edges exhibit an abnormally low redirection frequency, it may indicate an anomaly in that node or its downstream nodes. For example, the "Submit for Review" node may have a normal number of inbound edges (users arrive from multiple entry points), but the redirection frequency of all its outbound edges (leading to pages such as "Review Successful," "Review Failed," and "Please Modify") is close to zero. This suggests that the "Submit for Review" node itself may have a serious error or be functionally frozen, preventing users from proceeding to the next step.

[0096] This multi-dimensional analysis method can identify various types of abnormal nodes, including deeply hidden ones, and improve the accuracy of abnormal node identification, thereby providing more comprehensive and accurate optimization strategies for abnormal nodes.

[0097] To provide clear and quantifiable criteria for identifying abnormal nodes in the aforementioned multi-dimensional analysis methods, in one embodiment of this application, abnormal nodes in the user behavior path graph are determined based on at least one of the following: the access behavior metric corresponding to a node in the user behavior path graph, the connectivity of the node in the user behavior path graph, and the jump behavior metric of the edge in the user behavior path graph. This includes: if the access volume corresponding to the first node is greater than or equal to an access volume threshold and the access duration is less than or equal to an access duration threshold, then the first node is determined to be an abnormal node; if the second node is a node corresponding to a non-endpoint functional module and there is no successor node to the second node in the target user behavior path where the second node is located, then the second node is determined to be an abnormal node; if the out-degree of the third node is greater than or equal to a preset out-degree threshold... If the third node's out-degree threshold is greater than or equal to the first jump frequency threshold and the jump frequency corresponding to the outgoing edge of the third node is less than or equal to the first jump frequency threshold, then the third node is determined to be an abnormal node. The outgoing degree of the third node is used to represent the number of edges originating from the third node, and the outgoing edges of the third node are used to represent the edges originating from the third node. If the fourth node's in-degree threshold is greater than or equal to the preset in-degree threshold and the jump frequency corresponding to the incoming edge of the fourth node is less than or equal to the second jump frequency threshold, then the fourth node is determined to be an abnormal node. The ingoing degree of the fourth node is used to represent the number of edges pointing to the fourth node, and the incoming edges of the fourth node are used to represent the edges pointing to the fourth node. If the fifth node is an edgeless node and the access behavior metric value corresponding to the fifth node is greater than or equal to the preset metric value threshold, then the fifth node is determined to be an abnormal node.

[0098] The types of abnormal nodes can include high-traffic, low-stay type. These abnormal nodes can be determined by combining the node's traffic and visit duration. Specifically, if the traffic of the first node is greater than or equal to a preset traffic threshold (e.g., 1000 times / day), and the visit duration of the first node is less than or equal to a preset visit duration threshold (e.g., average stay less than 5 seconds), it indicates a large influx of users into the functional module corresponding to that first node, but almost no stay. Possible reasons include misleading entry text, page loading failure, or low-value content. Therefore, the first node is determined to be an abnormal node.

[0099] Abnormal nodes can also include path interruption types, which can be determined using node connectivity. That is, if the second node is a non-terminal functional module (i.e., according to product design, this functional module should have downstream functional modules), and there is no successor node to this second node in the target user behavior path (i.e., the functional module corresponding to the second node has no downstream functional modules), then it means that after the user reaches the functional module corresponding to the second node, they cannot perform any subsequent operations, and the behavior path is interrupted. This usually indicates problems such as a missing "Next" button, a failed navigation logic, or a page freeze. Therefore, the second node is determined to be an abnormal node.

[0100] Abnormal nodes can also include those with out-degree blocking. These abnormal nodes can be determined by combining the node's out-degree and the jump frequency of its outgoing edges. That is, if the out-degree of a third node is greater than or equal to a preset out-degree threshold (e.g., out-degree ≥ 5, indicating the functional module corresponding to the third node can reach multiple downstream functional modules), and the jump frequency of all outgoing edges of the third node is less than or equal to a preset first jump frequency threshold (e.g., the average daily jump count of each outgoing edge is ≤ 10 times), then the functional module corresponding to the third node itself has an anomaly (such as errors or performance bottlenecks), preventing users from jumping to other functional modules. Therefore, the third node is determined to be an abnormal node.

[0101] Abnormal nodes can also include those with in-degree blocking anomalies. These are determined by combining the node's in-degree and the jump frequency corresponding to its incoming edges. Specifically, if the in-degree of the fourth node is greater than or equal to a preset in-degree threshold (e.g., in-degree ≥ 5, indicating multiple upstream functional modules can jump to the functional module corresponding to the fourth node), and the jump frequency corresponding to all incoming edges of the fourth node is less than or equal to a preset second jump frequency threshold (e.g., the average daily jump count of each incoming edge is ≤ 10 times), then it means that many nodes should theoretically be able to jump to the fourth node, but in reality, almost no one successfully reaches it. This indicates that the functional module corresponding to the fourth node itself has an anomaly (e.g., page crash, permission verification failure, link failure). Therefore, the fourth node is determined to be an abnormal node.

[0102] The types of abnormal nodes can also include isolated high-access anomalies. This type of abnormal node can be determined using the access behavior metric of edgeless nodes. That is, if the fifth node is an edgeless node (i.e., both its out-degree and in-degree are zero, completely isolated) and the access behavior metric corresponding to the fifth node is greater than or equal to a preset metric threshold (e.g., daily average access volume ≥ 500), it indicates that an isolated functional module without any redirection relationship has generated significant access volume. This usually means that users access it through direct URL access, bookmarks, or external redirection, but the functional module lacks internal navigation paths. It may be an important functional module but lacks internal guidance. Therefore, the fifth node is determined to be an abnormal node.

[0103] By using the above multi-dimensional abnormal node determination criteria, we can not only accurately locate the cause of abnormality, but also improve the accuracy and interpretability of abnormal node identification. This provides a more comprehensive and accurate data foundation for generating optimization strategies for abnormal nodes, thereby improving the accuracy and targeting of the optimization strategies.

[0104] In one embodiment of this application, an optimization strategy for the abnormal node can also be determined based on the abnormal node's abnormality type and the task description information of the corresponding functional module.

[0105] After identifying abnormal nodes, we can further combine the node type of the abnormal node with the task description information of its corresponding functional module to generate targeted and executable optimization strategies, thus realizing a closed loop from identifying abnormalities to optimization.

[0106] In one alternative implementation, a rule base can be pre-established to associate and store the anomaly types and specific anomaly causes of abnormal nodes.

[0107] For example, the anomaly of high traffic but low dwell time can be caused by factors such as misleading entry text, mismatch between user expectations and actual content, slow page loading speed, low page content value, or page functions being prioritized or disabled, preventing users from performing the expected operations.

[0108] The path interruption exception can be caused by reasons such as the lack of an entry point to the downstream functional module, errors in the backend response interface causing no response after user operation, page freezing, or the path endpoint not being marked correctly even though the user has reached the endpoint.

[0109] Outbound blocking anomaly can be caused by serious functional defects in the functional module itself (such as infinite loops or errors), overly stringent triggering conditions for all exits within the functional module, or unreachable downstream functional modules (such as broken links).

[0110] An inbound blocking exception can be caused by reasons such as the functional module itself crashing or frequently reporting errors, all paths leading to the functional module having link errors or permission verification failures, or the functional module being intentionally or unintentionally set to be inaccessible.

[0111] Isolated high traffic anomalies can be caused by reasons such as users directly accessing a functional module through external links (promotions, bookmarks) but there is no internal navigation, or the functional module being an independent module but lacking internal guidance to continue browsing.

[0112] In another optional implementation, a functional module metadata database can be set up for each node. This database records the identification information and corresponding task description information of each node. The task description information is used to describe the core business objectives and typical user operations of the functional module. If an abnormal node is identified, its corresponding task description information can be obtained through the identification information of the abnormal node.

[0113] For example, the task description information could be that users can view, modify, copy, and delete created advertising creatives in this function module. The core operations include: filtering creatives, modifying copy / materials, previewing effects, and submitting for review.

[0114] By combining the mapping relationship between exception types and exception causes, and the task description information of the corresponding functional modules of the exception nodes, specific optimization strategies can be generated. For example, if the exception node is the Ad Creative Manager, the exception type is high traffic but low dwell time exception, the exception cause is misleading entry text, and the task description information is that the user modifies an already created ad creative in this functional module, with the core operation being modifying the copy / materials, then the generated optimization strategy could be: optimize the response time and interaction flow of the "Preview" function, and consider providing thumbnail previews directly on the list page. The optimization strategy can be a descriptive text in natural language, for example: The core operation of this Ad Creative Manager is "Preview," but users spend very little time on it. This may be due to performance issues with the "Preview" function (slow loading or inability to display correctly), or an unclear preview entry point, preventing users from quickly performing this operation. It is recommended to check the response time and interaction flow of the "Preview" function and consider providing thumbnail previews directly on the list page.

[0115] Optionally, after generating optimization strategies for all abnormal nodes, the optimization strategies can be sorted according to the priority of the abnormality type and integrated into a diagnostic report to clearly display the executable optimization strategies for all abnormal nodes.

[0116] To address the differentiated behavioral data of users of different user types, this application can further construct user behavior path diagrams for different user types and adjust the functional modules in the target applications used by users of different user types. Specifically, in one embodiment of this application, constructing a user behavior path diagram based on user behavior sequence data includes: constructing a user behavior path diagram for each user type based on the user behavior sequence data of each user type, according to a preset user type; the data processing method further includes: adjusting the functional modules in the target applications of the user types based on the user behavior path diagrams of the user types.

[0117] User types can be identified through user hierarchical tags, such as large customer type and small and medium-sized customer type. User behavior sequence data is split into independent datasets corresponding to different user types, and a user behavior path map is independently constructed for each user type. Based on the differentiated user behaviors between the user behavior path maps of users of different user types, the functional module configuration of the target application for the corresponding user group is adjusted in a targeted manner to achieve refined, accurate, personalized and effective functional optimization.

[0118] User types can be pre-defined based on business rules (such as registration duration, activity level, etc.) or machine learning models (such as user value segmentation models). During the collection or preprocessing phase, each user's behavioral data can be associated with its corresponding user type. User behavior sequence data is acquired and split into multiple independent datasets based on user type. Then, for each dataset, a user behavior path diagram corresponding to that user type is constructed. Different user types may have different needs, therefore, the required functional modules when performing corresponding tasks using the target application may also differ. Thus, based on the user behavior path diagrams corresponding to different user types, functional modules can be adjusted in a targeted manner.

[0119] For example, feature modules can be enabled or disabled. That is, a dedicated advanced feature module can be enabled for a user group corresponding to a certain user type, while the feature module can be hidden for a user group corresponding to another user type to avoid confusion. For example, the "API batch integration" feature module can be enabled for large customer type users, while the feature module can be hidden for small and medium-sized customer type users.

[0120] You can also adjust the visibility or location of function entry points. That is, for high-frequency function modules corresponding to a certain user type, place their entry points in a more conspicuous and convenient location. For example, for large customer type users, highlight the "Batch Creation" entry point on the homepage, and for small and medium-sized customer type users, highlight the "Newcomer Guide" or "Standard Creation" entry point.

[0121] It also allows for customization of interaction processes or interface layouts, that is, designing different interaction processes for the same functional module based on the usage habits of user groups corresponding to different user types. For example, the "Create Plan" functional module provides an "Advanced Settings" tab for large customer users and a simplified "Quick Create" form for small and medium-sized customer users.

[0122] It can also generate personalized content recommendations, that is, within the functional modules, different content is dynamically displayed according to the characteristics of user groups corresponding to different user types. For example, the "Design Inspiration" functional module displays "Large Brand Cases" for large clients and "Small Brand Cases" for small and medium-sized clients.

[0123] To achieve precise behavioral insights and differentiated experience optimization for different types of user groups, this application can also segment users according to user type and construct and analyze their corresponding user behavior path maps, which can automatically identify and visualize the exclusive experience checkpoints for each type of user group.

[0124] For each user type and the corresponding user behavior path graph, abnormal node analysis can be performed separately. The analysis process includes, but is not limited to, identifying abnormal nodes based on dimensions such as node access behavior metrics, node connectivity, and edge jump behavior metrics, using preset rules or AI models (such as GNN).

[0125] Because different user types exhibit varying behavioral patterns, the types and locations of anomalous nodes in their user behavior pathmaps may also differ. For example, for "large clients," the user behavior pathmap can focus on analyzing the access depth and redirection points of high-frequency functional modules such as "promotion management" and "batch creation," identifying anomalous nodes such as "high access volume but low dwell time." Conversely, for "small and medium-sized clients," the user behavior pathmap should focus on analyzing introductory functional modules such as "standard creation" and "targeted templates," identifying anomalous nodes such as "path interruption" or "high introductory depth but low redirection," reflecting drop-off points in onboarding and feature exploration.

[0126] Based on the analysis results of the user behavior path diagrams corresponding to each user type, abnormal nodes are identified in the user behavior path diagrams corresponding to each user type. For example, the node "Batch Creation" is marked as an abnormal node with "high exit degree and low jump" in the user behavior path diagram of "Large Customers", but it may not be marked as an abnormal node in the user behavior path diagram of "Small and Medium Customers" due to low access volume.

[0127] To achieve differentiated comparative displays on the user end, user behavior path graphs for each user type, along with data on abnormal nodes within those graphs, can be sent to the user end. In other words, the data sent to the user end includes: complete data (node ​​and edge attributes) of the user behavior path graph for a specific user type (e.g., "large customer"), and information on identified abnormal nodes within that user type's user behavior path graph. The user end can receive complete analysis results for different user groups, such as "large customers" and "small and medium-sized customers," laying the data foundation for subsequent comparative visualization.

[0128] When displaying layered visualizations on the user side, the client can receive user behavior path graphs for each user type from the server, as well as data on abnormal nodes within those graphs. The user interface can then use methods such as switching views or side-by-side views to display the user behavior path graphs for different user types separately. Figure 4 As shown in the diagram. For example, the user interface can provide a drop-down selector that allows users to switch between options such as "All Users," "Large Customers," and "Small and Medium Customers" to view a user behavior path map for a specific user group.

[0129] When displaying the user behavior path diagram of a specific user type on the client side, styles such as red highlighting and flashing can be used to highlight abnormal nodes in the user behavior path diagram of that user type.

[0130] Through the aforementioned layered construction, analysis, and visualization methods, we can intuitively compare the differences in user behavior paths for different user types, such as "large customers" and "small and medium-sized customers," and accurately pinpoint the unique experience bottlenecks of specific user groups. This facilitates understanding the reasons for these bottlenecks, such as "why large customers have a high churn rate after creating batch modules" or "why small and medium-sized customers always exit the standard creation module." This allows us to generate targeted optimization strategies, such as optimizing batch operation paths for large customers and strengthening introductory module guidance for small and medium-sized customers, significantly improving the accuracy and effectiveness of optimization.

[0131] To further optimize the various functional modules in the target application, in one embodiment of this application, the user behavior path graph can be analyzed to determine a first type of path and a second type of path corresponding to completing the same target task. The first type of path is used to generate an optimization strategy for the second type of path. The first type of path satisfies at least one of the following: the completion rate of the target task is greater than or equal to a completion rate threshold; the time taken to complete the target task is less than or equal to a preset time threshold; and the proportion of abnormal nodes in the path is less than or equal to a preset proportion value. The second type of path satisfies at least one of the following: the completion rate of the target task is less than a completion rate threshold; the time taken to complete the target task is greater than a preset time threshold; and the proportion of abnormal nodes in the path is greater than a preset proportion value.

[0132] The first type of path, also known as the efficient behavior path, and the second type of path, also known as the inefficient behavior path, refer to the different user behavior paths that may exist when different users perform the same target task. Based on the target task (such as generating advertising creatives), all user behavior paths in the user behavior path graph that start from the task's start event (e.g., clicking "Start Creating") and end with the task's success event (e.g., clicking "Submit Advertising Creative") are defined as candidate paths for that target task. A corresponding path ID can be generated for each candidate path, and all its node sequences (functional modules) and edge sequences (inter-module jumps) can be recorded. Then, based on preset efficiency metrics, an efficiency value is calculated for each candidate path, and according to a preset efficiency threshold, each candidate path is classified as either a first-type path or a second-type path.

[0133] The preset efficiency indicators may include the completion rate of the target task (whether the jump to the node corresponding to the endpoint functional module is successful), the time taken to complete the target task, and the proportion of abnormal nodes in the path.

[0134] For example, if the completion rate of the target task of candidate path A is 95%, the time to complete the target task is 3 minutes, and the proportion of abnormal nodes is 5%, then candidate path A can be considered to meet the efficiency index of high efficiency, and thus candidate path A can be classified as a first type of path.

[0135] If the completion rate of the target task of candidate path B is 30%, the time to complete the target task is 15 minutes, and the proportion of abnormal nodes is 40%, then candidate path A does not meet the efficiency index, and candidate path B can be classified as a second type of path.

[0136] The completion rate of the target task of candidate path C is 60%, the time to complete the target task is 8 minutes, and the proportion of abnormal nodes is 10%. Based on the actual situation, candidate path C can be classified as a first type path or a second type path, or other efficiency indicators can be used to continue to calculate the efficiency value of candidate path C.

[0137] Alternatively, the efficiency value of each candidate path can be determined by weighted summation of multiple efficiency indicators.

[0138] After identifying the first type of path, in-depth analysis can be performed to uncover its common characteristics (e.g., using clustering algorithms). For example, the direction of the analysis can include: node sequence patterns, whether there are common, core, and essential node sequences in the efficient behavior path (e.g., uploading materials -> previewing -> submitting for review), and these essential node sequences can be used as the most efficient user behavior path.

[0139] The mining direction can also include: key node features, that is, in the efficient behavior path, the nodes that almost all users visit, and the nodes that almost no one visits and can be skipped.

[0140] The exploration direction can also include: jump patterns, that is, nodes that users quickly jump to or briefly stay at in the efficient behavior path.

[0141] The exploration direction can also include: abnormal avoidance mode, that is, the way to avoid abnormal nodes that frequently occur in inefficient behavior paths in efficient behavior paths. For example, efficient behavior paths may avoid the abnormal node "standard creation" and directly use the "batch creation" node.

[0142] By using the first type of path, optimization strategies can be generated for the second type of path, thereby instructing users who follow inefficient paths to switch to efficient behavioral paths, significantly improving the efficiency and success rate of users who follow inefficient behavioral paths in completing their target tasks.

[0143] In step 203 above, namely "sending the user behavior path graph data and abnormal node data to the user terminal so that the user terminal can visualize the user behavior path graph and abnormal nodes", the server can send the constructed user behavior path graph data (including nodes, edges and corresponding attributes) and abnormal node data to the user terminal. The user terminal can use the received data to render a visual view on the user interface.

[0144] Figure 3 This is another illustrative flowchart of the data processing method provided in the embodiments of this application. The method can be... Figure 1 The user-side execution in the system shown is as follows. Figure 3 As shown, the data processing method may include the following steps: Step 301: Obtain the data of the user behavior path graph and the data of abnormal nodes in the user behavior path graph. The user behavior path graph is constructed based on user behavior sequence data. The nodes in the user behavior path graph are used to represent the functional modules in the target application, and the edges in the user behavior path graph are used to represent the user's jump path between the functional modules. Abnormal nodes are determined by analyzing the user behavior path graph.

[0145] Step 302: Display the nodes in the user behavior path graph using the first style, the edges in the user behavior path graph using the second style, and the abnormal nodes using the third style on the user interface. The first style is used to represent the user's access behavior measurement value to each functional module, and the second style is used to represent the user's jump behavior measurement value between each functional module.

[0146] As can be seen from the above process, this application constructs a user behavior path graph that simultaneously reflects access behavior metrics and redirection behavior metrics, and analyzes the user behavior path graph to identify abnormal nodes. This allows the visualization to simultaneously present the usage of functional modules and the flow between functional modules, and to locate abnormal functional modules in the target application. This effectively identifies abnormal functions in the target application, thereby improving the efficiency and accuracy of optimizing the target application.

[0147] The following describes in detail each step of the above process and the effects that can be produced, with reference to the embodiments.

[0148] In step 301 above, namely, "obtaining data from the user behavior path graph and data from abnormal nodes in the user behavior path graph, wherein the user behavior path graph is constructed based on user behavior sequence data, nodes in the user behavior path graph represent various functional modules in the target application, edges in the user behavior path graph represent the user's jump paths between various functional modules, and abnormal nodes are determined through analysis of the user behavior path graph," the server can construct the user behavior path graph based on the user behavior sequence data, analyze the user behavior path graph, determine abnormal nodes in the user behavior path graph, and send the constructed user behavior path graph data (including nodes, edges, and corresponding attributes) and abnormal node data to the user terminal. The user terminal can use the received data to render a visual view on the user interface.

[0149] The following describes in detail step 302, namely, "displaying nodes in the user behavior path graph using a first style, displaying edges in the user behavior path graph using a second style, and displaying abnormal nodes using a third style on the user interface, wherein the first style is used to represent the user's access behavior measurement value to each functional module, and the second style is used to represent the user's jump behavior measurement value between each functional module," with reference to the embodiments.

[0150] In one embodiment of this application, nodes in the user behavior path graph are displayed using a first style, including: using the color of the node to represent the duration of the user's access to the functional module, and / or using the size of the node to represent the number of times the user accesses the functional module; edges in the user behavior path graph are displayed using a second style, including: using the size of the edge to represent the frequency of the user's jump between various functional modules.

[0151] The first style includes at least one visual attribute of the node's color, size, transparency, and shape; the second style includes at least one visual attribute of the edge's color, size, transparency, and shape; the third style includes at least one visual attribute of the abnormal node's color, size, transparency, and shape, and the third style differs from at least some of the visual attributes of the first and second styles. This significantly improves the efficiency of identifying the overall distribution and flow characteristics of user behavior; and by using different styles to display abnormal nodes, abnormal nodes can be highlighted in the user behavior path graph, improving the efficiency of locating abnormal functional modules in the target application.

[0152] Node attributes include access behavior metrics for the corresponding functional module. These metrics can include visit volume, unique visitors, average number of visits per user, visit duration, and average visit duration per user. At least one access behavior metric can be reflected through the node's first style. For example, node size reflects the number of visits to the corresponding functional module: larger nodes indicate higher visit volume, and smaller nodes indicate lower visit volume. Node color reflects the duration of user visits to the corresponding functional module: darker colors indicate longer visit durations, and lighter colors indicate shorter visit durations. Node transparency reflects the number of unique visitors to the corresponding functional module: higher transparency indicates fewer unique visitors, and lower transparency indicates more unique visitors. And so on. It should be noted that the examples here are not intended to limit the types of node styles, access behavior metrics, or the mapping relationship between styles and access behavior metrics.

[0153] The attributes of the edges include metrics of user navigation behavior between corresponding functional modules. These metrics can include navigation frequency, number of navigations, throughput, navigation duration, and navigation success rate. At least one navigation behavior metric can be reflected through the edge's second style. For example, the thickness of the edge reflects the frequency of navigation between corresponding functional modules: a thicker edge indicates a higher frequency, and a thinner edge indicates a lower frequency. The length of the edge reflects the duration of navigation between corresponding functional modules: a longer edge indicates a longer navigation duration, and a shorter edge indicates a shorter navigation duration. The color of the edge reflects the success rate of navigation between corresponding functional modules: a darker color indicates a higher success rate, and a lighter color indicates a lower success rate. It should be noted that the examples here are not intended to limit the style of the edges, the type of navigation behavior metrics, or the mapping relationship between styles and navigation behavior metrics.

[0154] To highlight abnormal nodes in the user behavior path graph and facilitate quick location of abnormal nodes, abnormal nodes can be displayed using a third style that differs from the first and second styles. For example, adding a highlighted border to the abnormal node, changing the color of the abnormal node to a highlighted color, or changing the blinking effect of the abnormal node can make the abnormal node clearly distinguishable from normal nodes and edges.

[0155] Figure 4 This is a schematic diagram illustrating the display of a user behavior path graph on a user interface, as provided in an embodiment of this application. Figure 4 As shown, the user interface displays user hierarchical labels, including all users, large customer type users, and small and medium customer type users. By selecting the corresponding hierarchical label, the user behavior path diagram of the user corresponding to that hierarchical label can be displayed.

[0156] like Figure 4As shown in (a), a user behavior path diagram is displayed for all users. This user behavior path diagram includes nodes corresponding to the "Promotion Management" function module, the "Standard Creation" function module, the "Batch Creation" function module, the "Video Library" function module, the "Homepage" function module, the "Submit for Review" function module, the "Review Successful" function module, and the "Custom Report" function module. The edges connecting the nodes of the "Homepage" function module to the nodes of the "Promotion Management" function module, the "Promotion Management" function module to the nodes of the "Standard Creation" function module, the "Promotion Management" function module to the nodes of the "Batch Creation" function module, the bidirectional edges between the nodes of the "Standard Creation" function module and the nodes of the "Video Library" function module, the bidirectional edges between the nodes of the "Batch Creation" function module and the nodes of the "Video Library" function module, the edges connecting the nodes of the "Standard Creation" function module to the nodes of the "Submit for Review" function module, the "Batch Creation" function module to the nodes of the "Submit for Review" function module, and the edges connecting the nodes of the "Submit for Review" function module to the nodes of the "Review Successful" function module.

[0157] If in Figure 4 When selecting the "Large Customer" type user in the interface shown in (a), then as follows: Figure 4 As shown in Figure (b), a user behavior path diagram corresponding to large customer type users is displayed. This user behavior path diagram includes nodes corresponding to the "Promotion Management" function module, the "Standard Creation" function module, the "Video Library" function module, the "Homepage" function module, the "Submit for Review" function module, and the "Review Successful" function module. The diagram also includes edges pointing from the "Homepage" function module node to the "Promotion Management" function module node, edges pointing from the "Promotion Management" function module node to the "Standard Creation" function module node, bidirectional edges between the "Standard Creation" and "Video Library" function modules, edges pointing from the "Standard Creation" function module node to the "Submit for Review" function module node, and edges pointing from the "Submit for Review" function module node to the "Review Successful" function module node.

[0158] If in Figure 4 When selecting the small and medium-sized customer type user in the interface shown in (a), then as follows Figure 4As shown in (c), a user behavior path diagram corresponding to small and medium-sized customer types is displayed. This user behavior path diagram includes nodes corresponding to the "Promotion Management" function module, the "Batch Creation" function module, the "Video Library" function module, the "Homepage" function module, the "Submit for Review" function module, and the "Review Successful" function module. The diagram also includes edges pointing from the "Homepage" function module node to the "Promotion Management" function module node, edges pointing from the "Promotion Management" function module node to the "Batch Creation" function module node, bidirectional edges between the "Batch Creation" and "Video Library" function modules, edges pointing from the "Batch Creation" function module node to the "Submit for Review" function module node, and edges pointing from the "Submit for Review" function module node to the "Review Successful" function module node.

[0159] exist Figure 4 In this context, node size reflects user access frequency to corresponding functional modules; larger nodes indicate higher access frequency, and smaller nodes indicate lower access frequency. Node color reflects the duration of user visits to corresponding functional modules; darker colors indicate longer visits, and lighter colors indicate shorter visits. Edge thickness reflects the frequency of user navigation between functional modules; thicker edges indicate higher navigation frequency, and thinner edges indicate lower navigation frequency. Edge length reflects the duration of user navigation between functional modules; longer edges indicate longer navigation duration, and shorter edges indicate shorter navigation duration. Figure 4 In (a), since the node corresponding to the “Custom Report” function module is a borderless node and the size of the node is relatively large, it can be considered as an isolated high-access exception node. The border of the exception node can be thickened to distinguish it significantly from other nodes in the user behavior path graph.

[0160] Optionally, a bar chart can also be used to display each functional module or the functional module with a high access behavior metric value on the user interface, such as... Figure 4 As shown in (a), the user interface also displays the access duration for the "Promotion Management" module, the "Standard Creation" module, and the "Batch Creation" module using bar charts. The longer the bar, the longer the access duration. This method provides a more intuitive view of the access duration for each module.

[0161] Optionally, the user interface can also display the changes in the access behavior metrics of each functional module, such as the increase or decrease in the access behavior metrics of a certain functional module in the current period compared to the previous period, such as an increase of 3.2% in access duration and a decrease of 4.8% in access volume, etc.

[0162] Optionally, a prompt message can also be displayed on the user interface. This prompt message is used to alert functional modules whose access behavior metric values ​​are lower than preset values, making it easier to intuitively display functional modules with potential anomalies.

[0163] In one embodiment of this application, an interactive function with a user behavior path graph can also be provided. That is, in response to detecting an event that a node in the user behavior path graph is selected, at least one of the following is displayed: the functional description information of the functional module corresponding to the node, the numerical value of the access behavior metric corresponding to the node, the associated nodes of the node, and the numerical value of the jump behavior metric between the node and the associated nodes of the node; and / or, in response to detecting an event that an edge in the user behavior path graph is selected, the numerical value of the jump behavior metric corresponding to the edge and / or the associated nodes of the edge are displayed.

[0164] By interacting with the user behavior path map, detailed measurement information and contextual information of the selected elements in the user behavior path map can be dynamically displayed. This solves the limitation that visual styles can only present trends and relative magnitudes and cannot directly provide precise values ​​and relationships. This simplifies the operation of obtaining relevant information from the user behavior path map, improves the readability of the user behavior path map, and increases the efficiency of analyzing the user behavior path map.

[0165] Events involving selected nodes in the user behavior path graph can be triggered by a user clicking (or hovering) a node in the user behavior path graph on the user interface. The unique identifier of the selected node (such as a functional module ID) can be obtained. Using this identifier, the functional module's functional description information (e.g., "Standard Creation: Users can view, modify, copy, and delete created advertising creatives in this module"), the precise value of the access behavior metric associated with the node (e.g., "Daily average visits: 12,350 times; average dwell time: 8.2 seconds"), all associated nodes of the node (i.e., nodes directly connected to the node via edges), and the corresponding jump behavior metric values ​​(e.g., "Downstream nodes: Creative preview page (daily average jumps: 8,500 times), Submission for review page (daily average jumps: 950 times)") can be retrieved.

[0166] like Figure 4 (e) (can be) Figure 4 The interface shown in any of (a) to (e) is used here only as an example. Figure 4(Taking the interface shown in (e) as an example), when the node corresponding to the “Standard Creation” function module is selected, the user interface will display “Standard Creation: Users can view, modify, copy, and delete created advertising creatives in this module”, “Daily average visits: 12,350 times; Average dwell time per person: 8.2 seconds”, “Upstream node: Promotion management (Daily average number of redirects: 7,900 times); Downstream node: Submit for review (Daily average number of redirects: 950 times)”.

[0167] The event of selecting an edge in the user behavior path graph can be triggered by the user clicking (or hovering) on ​​an edge in the user behavior path graph on the user interface. The unique identifier of the selected edge (such as "functional module ID_A->functional module ID_B"), the precise value of the jump behavior metric associated with the edge (such as "daily jump frequency: 1200 times; jump success rate: 98.5%; average jump duration: 0.8 seconds"), and the starting node and pointing node of the edge (such as "starting node: standard creation, pointing node: submission for review").

[0168] Continue to refer to Figure 4 In the middle (e), when the node corresponding to the “Promotion Management” function module is selected and the edge pointing to the node corresponding to the “Standard Creation” function module is selected, the following can be displayed on the user interface: “Promotion Management -> Standard Creation”, “Daily jump frequency: 1200 times; Jump success rate: 98.5%; Average jump time: 0.8 seconds”, “Starting node: Promotion Management, Pointing node: Standard Creation”.

[0169] Floating windows, sidebars, information cards, etc., can be rendered on the user interface to dynamically display the information retrieved above.

[0170] In one alternative implementation, in response to an event that detects a selected node in the user behavior path graph, only the selected node, its associated nodes (including upstream and downstream nodes), and corresponding edges can be displayed (or highlighted), thereby focusing on the flow path of the selected node. See also... Figure 4 In response to the detection of selection Figure 4 The node corresponding to the "Promotion Management" function module in (a) is in Figure 4 The interface shown in (d) displays the nodes corresponding to the "Promotion Management" function module, the upstream node corresponding to the "Homepage" function module, and the downstream nodes: the nodes corresponding to the "Standard Creation" function module and the "Batch Creation" function module, as well as the edges pointing from the nodes corresponding to the "Homepage" function module to the nodes corresponding to the "Promotion Management" function module, the edges pointing from the nodes corresponding to the "Promotion Management" function module to the nodes corresponding to the "Standard Creation" function module, and the edges pointing from the nodes corresponding to the "Promotion Management" function module to the nodes corresponding to the "Batch Creation" function module. It does not display... Figure 4 Other nodes and edges in (a).

[0171] In another alternative implementation, in response to an event that detects a selected edge in the user behavior path graph, only the selected edge, along with its originating and pointing nodes, can be displayed (or highlighted), thereby focusing on the data of a single path of the selected edge. See also... Figure 4 In response to the detection of selection Figure 4 In (a), the edge pointing from the node corresponding to the "Promotion Management" function module to the node corresponding to the "Standard Creation" function module is in... Figure 4 (e) displays the nodes corresponding to the "Promotion Management" function module and the "Standard Creation" function module, as well as the edges pointing from the nodes of the "Promotion Management" function module to the nodes of the "Standard Creation" function module, but does not display... Figure 4 Other nodes and edges in (a).

[0172] Optionally, when a user selects an element, other previously displayed details can be automatically closed to avoid interface clutter. The display duration of the retrieved information can also be dynamically adjusted based on the duration of the selection event. In addition, the displayed details can follow the position of the selected element or be fixed to one side of the user interface (such as a sidebar) and can be manually closed by the user. Furthermore, the displayed content can be configured according to business needs. For example, only "measurement values" and "related nodes" can be displayed, while "functional description information" is not displayed.

[0173] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0174] According to another embodiment, a data processing apparatus is provided. Figure 5 This is a schematic block diagram of a data processing apparatus provided in an embodiment of this application. The apparatus is disposed in... Figure 1 The server side in the illustrated architecture. For example... Figure 5 As shown, the data processing device 500 mainly includes: a construction unit 501, an analysis unit 502, and a transmission unit 503. The main functions of each component are as follows: Construction unit 501 is configured to construct a user behavior path graph based on user behavior sequence data. Nodes in the user behavior path graph represent functional modules in the target application, and edges in the user behavior path graph represent the user's jump path between functional modules. The attributes of nodes include the access behavior metric of the corresponding functional module, and the attributes of edges include the jump behavior metric of the user between the corresponding functional modules.

[0175] Analysis unit 502 is configured to analyze the user behavior path graph and identify abnormal nodes in the user behavior path graph.

[0176] The sending unit 503 is configured to send the user behavior path graph data and abnormal node data to the user terminal so that the user terminal can visualize the user behavior path graph and abnormal nodes.

[0177] Furthermore, the data processing apparatus 500 may also include: a splicing unit (not shown in the figure), configured to, before constructing a user behavior path map based on user behavior sequence data, in response to detecting the presence of discontinuous data in the user behavior data, splice the discontinuous data to target continuous data related to the discontinuous data in the user behavior data to obtain user behavior sequence data.

[0178] As one optional implementation, when the splicing unit splices discontinuous data to target continuous data related to the discontinuous data in the user's behavior data to obtain user behavior sequence data, it can be specifically configured to: determine multiple candidate continuous data in the user's behavior data based on preset rules; determine target continuous data according to the matching degree between each candidate continuous data and the discontinuous data; and splice the discontinuous data with the target continuous data to obtain user behavior sequence data; wherein the preset rules include at least one of the following: the time interval between the generation time of the discontinuous data and the generation time of the candidate continuous data is less than a first time interval threshold; the similarity between the task object corresponding to the discontinuous data and the task object corresponding to the candidate continuous data is greater than a similarity threshold; the continuity between the first operation event corresponding to the discontinuous data and the second operation event corresponding to the candidate continuous data is greater than a continuity threshold; the matching degree between the candidate continuous data and the discontinuous data is determined by at least one of the following: the time interval between the generation time of the discontinuous data and the generation time of the candidate continuous data, the similarity between the task object corresponding to the discontinuous data and the task object corresponding to the candidate continuous data, and the continuity between the first operation event corresponding to the discontinuous data and the second operation event corresponding to the candidate continuous data.

[0179] As one optional implementation, when the splicing unit detects discontinuous data in the user's behavior data, it can be specifically configured to at least one of the following: determining that discontinuous data exists in the user's behavior data in response to detecting that the session identifier information corresponding to multiple consecutive behavior data of the same user is different; determining that discontinuous data exists in the user's behavior data in response to detecting that the device identifiers corresponding to multiple behavior data of the user are different, and the time interval between the generation times of the multiple behavior data is less than a second time interval threshold and / or the task objects corresponding to the multiple behavior data are the same; determining that discontinuous data exists in the user's behavior data in response to detecting an event in the user's behavior data where the user closes a function module and then restarts the function module within a first preset time period; determining that discontinuous data exists in the user's behavior data in response to detecting an event in the user's behavior data where the response duration is greater than a second preset time period.

[0180] Furthermore, the splicing unit can be specifically configured to: after splicing discontinuous data to target continuous data related to the discontinuous data in the user's behavior data to obtain user behavior sequence data, add splicing identification information to the user behavior sequence data. The splicing identification information is used to identify the confidence level of the user behavior sequence data. Correspondingly, when constructing a user behavior path graph based on the user behavior sequence data, the construction unit 501 can be specifically configured to: in response to detecting that the confidence level of the user behavior sequence data is greater than the confidence level threshold, construct a user behavior path graph based on the user behavior sequence data.

[0181] As one of the optional implementation methods, when the analysis unit 502 analyzes the user behavior path graph and determines the abnormal nodes in the user behavior path graph, it can be specifically configured to: determine the abnormal nodes in the user behavior path graph based on at least one of the following: the access behavior metric value corresponding to the node in the user behavior path graph, the connectivity of the node in the user behavior path graph, and the jump behavior metric value of the edge in the user behavior path graph.

[0182] As one optional implementation, when the analysis unit 502 determines abnormal nodes in the user behavior path graph based on at least one of the access behavior metric value corresponding to a node in the user behavior path graph, the connectivity of the node in the user behavior path graph, and the jump behavior metric value of the edge in the user behavior path graph, it can be specifically configured as follows: if the access volume corresponding to the first node is greater than or equal to the access volume threshold and the access duration is less than or equal to the access duration threshold, then the first node is determined to be an abnormal node; if the second node is a node corresponding to a non-endpoint functional module and there is no successor node of the second node in the target user behavior path where the second node is located, then the second node is determined to be an abnormal node; if the out-degree of the third node is greater than or equal to the preset out-degree threshold and the third node... If the jump frequency corresponding to the outgoing edge of the third node is less than or equal to the first jump frequency threshold, then the third node is determined to be an abnormal node. The outgoing degree of the third node is used to represent the number of edges originating from the third node, and the outgoing edges of the third node are used to represent the edges originating from the third node. If the ingoing degree of the fourth node is greater than or equal to the preset ingoing degree threshold and the jump frequency corresponding to the incoming edge of the fourth node is less than or equal to the second jump frequency threshold, then the fourth node is determined to be an abnormal node. The ingoing degree of the fourth node is used to represent the number of edges pointing to the fourth node, and the incoming edges of the fourth node are used to represent the edges pointing to the fourth node. If the fifth node is an edgeless node and the access behavior metric value corresponding to the fifth node is greater than or equal to the preset metric value threshold, then the fifth node is determined to be an abnormal node.

[0183] Furthermore, the data processing device 500 may also include an optimization unit (not shown in the figure), configured to determine an optimization strategy for the abnormal node based on the abnormal node's abnormality type and the task description information of the functional module corresponding to the abnormal node.

[0184] As one of the optional implementation methods, when constructing a user behavior path diagram based on user behavior sequence data, the construction unit 501 can be specifically configured to: construct a user behavior path diagram for each user type based on the user behavior sequence data of the user corresponding to each user type according to a preset user type; correspondingly, the optimization unit can also be configured to adjust the functional modules in the target application of the user type corresponding to the user based on the user behavior path diagram of the user type corresponding to the user.

[0185] As one optional implementation, when the analysis unit 502 analyzes the user behavior path graph and determines the abnormal nodes in the user behavior path graph, it can be specifically configured to: analyze the user behavior path graph of users corresponding to each user type and determine the abnormal nodes in the user behavior path graph of users corresponding to each user type; when the sending unit 503 sends the user behavior path graph data and abnormal node data to the user terminal so that the user terminal can visualize the user behavior path graph and abnormal nodes, it can be specifically configured to: send the user behavior path graph data of users corresponding to each user type and the abnormal node data in the user behavior path graph of users corresponding to each user type to the user terminal so that the user terminal can visualize the user behavior path graph of users corresponding to each user type and the abnormal nodes in the user behavior path graph of users corresponding to each user type.

[0186] Furthermore, the optimization unit can also be configured to analyze the user behavior path graph, identify a first type of path and a second type of path corresponding to completing the same target task in the user behavior path graph; and generate an optimization strategy for the second type of path using the first type of path; wherein the first type of path satisfies at least one of the following: the completion rate of the target task is greater than or equal to a completion rate threshold, the time to complete the target task is less than or equal to a preset time threshold, and the proportion of abnormal nodes in the path is less than or equal to a preset proportion value; the second type of path satisfies at least one of the following: the completion rate of the target task is less than a completion rate threshold, the time to complete the target task is greater than a preset time threshold, and the proportion of abnormal nodes in the path is greater than a preset proportion value.

[0187] According to another embodiment, a data processing apparatus is provided. Figure 6 This is another schematic block diagram of the data processing apparatus provided in the embodiments of this application, the apparatus being disposed in... Figure 1 The user end in the illustrated architecture. (For example...) Figure 6 As shown, the data processing device 600 mainly includes an acquisition unit 601 and a display unit 602. The main functions of each component are as follows: The acquisition unit 601 is configured to acquire data of the user behavior path graph and data of abnormal nodes in the user behavior path graph. The user behavior path graph is constructed based on user behavior sequence data. The nodes in the user behavior path graph are used to represent the functional modules in the target application, and the edges in the user behavior path graph are used to represent the user's jump path between the functional modules. The abnormal nodes are determined by analyzing the user behavior path graph.

[0188] The display unit 602 is configured to display nodes in the user behavior path graph using a first style, edges in the user behavior path graph using a second style, and abnormal nodes using a third style on the user interface. The first style is used to represent the user's access behavior measurement value to each functional module, and the second style is used to represent the user's jump behavior measurement value between each functional module.

[0189] Furthermore, the nodes in the user behavior path graph are displayed using a first style, including: using the color of the node to represent the duration of the user's access to the functional module, and / or using the size of the node to represent the number of times the user accesses the functional module; the edges in the user behavior path graph are displayed using a second style, including: using the size of the edge to represent the frequency of the user's jump between various functional modules.

[0190] Furthermore, the display unit 602 is also configured to: in response to an event that a node in the user behavior path graph is selected, display at least one of the following: functional description information of the functional module corresponding to the node, the value of the access behavior metric corresponding to the node, the associated nodes of the node, and the value of the jump behavior metric between the associated nodes of the node; and / or, in response to an event that an edge in the user behavior path graph is selected, display the value of the jump behavior metric corresponding to the edge and / or the associated nodes of the edge.

[0191] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, for system or device embodiments, since they are basically similar to method embodiments, the description is relatively simple, and relevant parts can be referred to the description of the method embodiments. The system and device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0192] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of the relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0193] In addition, embodiments of this application also provide a computer-readable storage medium storing a computer program thereon, which, when executed, implements the steps of the method described in any of the foregoing method embodiments.

[0194] And an electronic device, comprising: One or more processors; and A memory associated with the one or more processors, the memory being used to store program instructions that, when read and executed by the one or more processors, perform the steps of the method described in any of the foregoing method embodiments.

[0195] This application also provides a computer program product, including a computer program that, when executed, implements the steps of the method described in any of the foregoing method embodiments.

[0196] in, Figure 7 This is a schematic block diagram of an electronic device provided in an embodiment of this application. Specifically, it may include a processor 710, a video display adapter 711, a disk drive 712, an input / output interface 713, a network interface 714, and a memory 720. The processor 710, video display adapter 711, disk drive 712, input / output interface 713, network interface 714, and memory 720 can be communicatively connected via a communication bus 730. The input / output interface 713 can also be referred to as an I / O interface 713.

[0197] The processor 710 can be implemented using a general-purpose CPU, microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits to execute relevant programs and implement the technical solution provided in this application.

[0198] The memory 720 can be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage device, dynamic storage device, etc. The memory 720 can store the operating system 721 for controlling the operation of the electronic device 1000, and the basic input / output system (BIOS) 722 for controlling the low-level operations of the electronic device 700. Additionally, it can store a web browser 723, a data storage management system 724, and data processing devices 500 and 600, etc. The aforementioned data processing devices 500 and 600 can be application programs that specifically implement the aforementioned steps in the embodiments of this application. In summary, when the technical solution provided in this application is implemented through software or firmware, the relevant program code is stored in the memory 720 and is called and executed by the processor 710.

[0199] Input / output interface 713 is used to connect input / output modules to realize information input and output. Input / output modules can be configured as components in the device (not shown in the figure) or externally connected to the device to provide corresponding functions. Input devices may include keyboards, mice, touch screens, microphones, various sensors, etc., and output devices may include displays, speakers, vibrators, indicator lights, etc.

[0200] Network interface 714 is used to connect a communication module (not shown in the figure) to enable communication between this device and other devices. The communication module can communicate via wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0201] Bus 730 includes a pathway for transmitting information between various components of the device, such as processor 710, video display adapter 711, disk drive 712, input / output interface 713, network interface 714, and memory 720.

[0202] It should be noted that although the above-described device only shows the processor 710, video display adapter 711, disk drive 712, input / output interface 713, network interface 714, memory 720, bus 730, etc., in specific implementations, the device may also include other components necessary for normal operation. Furthermore, those skilled in the art will understand that the above-described device may only include the components necessary for implementing the solution of this application, and does not necessarily include all the components shown in the figures.

[0203] As can be seen from the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a computer program product. This computer program product can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in various embodiments or some parts of the embodiments of this application.

[0204] The technical solutions provided in this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A data processing method, characterized in that, Applied to the server side, the data processing method includes: Based on user behavior sequence data, a user behavior path graph is constructed. The nodes in the user behavior path graph are used to represent the functional modules in the target application, and the edges in the user behavior path graph are used to represent the user's jump path between the functional modules. The attributes of the nodes include the access behavior metric of the corresponding functional module, and the attributes of the edges include the jump behavior metric of the user between the corresponding functional modules. Analyze the user behavior path graph to identify abnormal nodes in the user behavior path graph; The user behavior path graph data and the abnormal node data are sent to the user terminal so that the user terminal can visualize the user behavior path graph and the abnormal node.

2. The data processing method according to claim 1, characterized in that, Before constructing the user behavior path graph based on user behavior sequence data, the data processing method further includes: In response to the detection of discontinuous data in the user's behavior data, the discontinuous data is concatenated to the target continuous data related to the discontinuous data in the user's behavior data to obtain the user behavior sequence data.

3. The data processing method according to claim 2, characterized in that, The step of concatenating the discontinuous data into target continuous data related to the discontinuous data in the user's behavior data to obtain the user behavior sequence data includes: Based on preset rules, multiple candidate continuous data are determined from the user's behavior data; The target continuous data is determined based on the matching degree between each candidate continuous data and the non-continuous data among the plurality of candidate continuous data; The discontinuous data is concatenated with the target continuous data to obtain the user behavior sequence data; The preset rules include at least one of the following: the time interval between the generation time of the non-continuous data and the generation time of the candidate continuous data is less than a first time interval threshold; the similarity between the task object corresponding to the non-continuous data and the task object corresponding to the candidate continuous data is greater than a similarity threshold; and the continuity between the first operation event corresponding to the non-continuous data and the second operation event corresponding to the candidate continuous data is greater than a continuity threshold. The matching degree between the candidate continuous data and the non-continuous data is determined by at least one of the following: the time interval between the generation time of the non-continuous data and the generation time of the candidate continuous data; the similarity between the task object corresponding to the non-continuous data and the task object corresponding to the candidate continuous data; and the degree of continuity between the first operation event corresponding to the non-continuous data and the second operation event corresponding to the candidate continuous data.

4. The data processing method according to claim 2, characterized in that, The detection of discontinuous data in the user's behavior data includes at least one of the following: In response to detecting that the session identifier information corresponding to multiple consecutive behavioral data of the same user is different, it is determined that there is non-continuous data in the user's behavioral data; In response to the detection that multiple behavioral data of the user correspond to different device identifiers, and the time interval between the generation times of the multiple behavioral data is less than a second time interval threshold and / or the task objects corresponding to the multiple behavioral data are the same, it is determined that there is non-continuous data in the user's behavioral data. In response to an event that the user closes the function module and then restarts the function module within a first preset time period after detecting an event in the user's behavior data, it is determined that there is discontinuous data in the user's behavior data. In response to detecting an event in the user's behavior data with a response duration greater than a second preset duration, it is determined that there is discontinuous data in the user's behavior data.

5. The data processing method according to claim 2, characterized in that, After concatenating the discontinuous data into the target continuous data related to the discontinuous data in the user's behavior data to obtain the user behavior sequence data, the data processing method further includes: Add splicing identifier information to the user behavior sequence data, the splicing identifier information being used to identify the confidence level of the user behavior sequence data; The process of constructing a user behavior path graph based on user behavior sequence data includes: In response to detecting that the confidence level of the user behavior sequence data is greater than a confidence threshold, the user behavior path graph is constructed based on the user behavior sequence data.

6. The data processing method according to any one of claims 1-5, characterized in that, The step of analyzing the user behavior path graph and identifying abnormal nodes in the user behavior path graph includes: Abnormal nodes in the user behavior path graph are determined based on at least one of the following: the access behavior metric value corresponding to the node in the user behavior path graph, the connectivity of the node in the user behavior path graph, and the jump behavior metric value of the edge in the user behavior path graph.

7. The data processing method according to claim 6, characterized in that, The step of determining abnormal nodes in the user behavior path graph based on at least one of the following: access behavior metrics corresponding to nodes in the user behavior path graph, connectivity of nodes in the user behavior path graph, and jump behavior metrics of edges in the user behavior path graph, includes: If the number of visits to the first node is greater than or equal to the number of visits threshold and the access duration is less than or equal to the access duration threshold, then the first node is determined to be an abnormal node. If the second node is a node corresponding to a non-endpoint functional module and there is no successor node of the second node in the target user behavior path where the second node is located, the second node is determined to be an abnormal node. If the out-degree of the third node is greater than or equal to a preset out-degree threshold and the jump frequency corresponding to the outgoing edge of the third node is less than or equal to a first jump frequency threshold, then the third node is determined to be an abnormal node. The out-degree of the third node is used to represent the number of edges with the third node as the starting node, and the outgoing edge of the third node is used to represent the edges with the third node as the starting node. If the in-degree of the fourth node is greater than or equal to a preset in-degree threshold and the jump frequency corresponding to the incoming edge of the fourth node is less than or equal to a second jump frequency threshold, then the fourth node is determined to be an abnormal node. The in-degree of the fourth node is used to represent the number of edges pointing to the fourth node, and the incoming edge of the fourth node is used to represent the edges pointing to the fourth node. If the fifth node is a borderless node and the access behavior metric value corresponding to the fifth node is greater than or equal to a preset metric value threshold, then the fifth node is determined to be an abnormal node.

8. The data processing method according to claim 1, characterized in that, Also includes: Based on the anomaly type of the abnormal node and the task description information of the corresponding functional module of the abnormal node, an optimization strategy is determined for the abnormal node.

9. The data processing method according to any one of claims 1-5, characterized in that, The process of constructing a user behavior path graph based on user behavior sequence data includes: Based on the preset user types, and using the user behavior sequence data corresponding to each user type, a user behavior path diagram for each user type is constructed. The data processing method further includes: Based on the user behavior path diagram of the user corresponding to the user type, adjust the functional modules in the target application of the user corresponding to the user type.

10. The data processing method according to claim 9, characterized in that, The step of analyzing the user behavior path graph and identifying abnormal nodes in the user behavior path graph includes: Analyze the user behavior path graphs of users corresponding to each user type to identify abnormal nodes in the user behavior path graphs of users corresponding to each user type; The step of sending the user behavior path graph data and the abnormal node data to the user terminal, so that the user terminal can visualize the user behavior path graph and the abnormal node, includes: The user behavior path graph data corresponding to each user type, as well as the abnormal node data in the user behavior path graph corresponding to each user type, are sent to the user terminal so that the user terminal can visualize the user behavior path graph corresponding to each user type and the abnormal node data in the user behavior path graph corresponding to each user type.

11. The data processing method according to any one of claims 1-5, characterized in that, Also includes: The user behavior path graph is analyzed to determine the first type of path and the second type of path corresponding to the same target task. Using the first type of path, generate an optimization strategy for the second type of path; The first type of path satisfies at least one of the following: the completion rate of the target task is greater than or equal to the completion rate threshold, the time to complete the target task is less than or equal to the preset time threshold, and the proportion of abnormal nodes in the path is less than or equal to the preset proportion value. The second type of path satisfies at least one of the following: the completion rate of the target task is less than the completion rate threshold, the time taken to complete the target task is greater than the preset time threshold, and the proportion of abnormal nodes in the path is greater than the preset proportion value.

12. A data processing method, characterized in that, When applied to the user end, the data processing method includes: The user behavior path graph data and the abnormal node data in the user behavior path graph are obtained. The user behavior path graph is constructed based on user behavior sequence data. The nodes in the user behavior path graph are used to represent the functional modules in the target application. The edges in the user behavior path graph are used to represent the user's jump path between the functional modules. The abnormal nodes are determined by analyzing the user behavior path graph. The user interface displays nodes in the user behavior path graph using a first style, edges in the user behavior path graph using a second style, and abnormal nodes using a third style. The first style is used to represent the user's access behavior metric for each functional module, and the second style is used to represent the user's jump behavior metric between each functional module.

13. The data processing method according to claim 12, characterized in that, The first style of displaying nodes in the user behavior path graph includes: using the color of the node to represent the duration of the user's access to the functional module, and / or using the size of the node to represent the number of times the user accesses the functional module; The second style of displaying the edges in the user behavior path graph includes: using the size of the edges to represent the frequency of the user's jumps between the various functional modules.

14. The data processing method according to claim 12, characterized in that, Also includes: In response to detecting an event that a node in the user behavior path graph is selected, at least one of the following is displayed: the functional description information of the functional module corresponding to the node, the numerical value of the access behavior metric corresponding to the node, the associated nodes of the node, and the numerical value of the jump behavior metric between the node and the associated nodes of the node. And / or, In response to an event that detects the selection of an edge in the user behavior path graph, the numerical value of the jump behavior metric corresponding to the edge and / or the associated node of the edge are displayed.

15. A data processing apparatus, characterized in that, The data processing device, applied on the server side, includes: The building unit is configured to construct a user behavior path graph based on user behavior sequence data. The nodes in the user behavior path graph are used to represent the functional modules in the target application, and the edges in the user behavior path graph are used to represent the user's jump path between the functional modules. The attributes of the nodes include the access behavior metric of the corresponding functional module, and the attributes of the edges include the jump behavior metric of the user between the corresponding functional modules. The analysis unit is configured to analyze the user behavior path graph and identify abnormal nodes in the user behavior path graph. The sending unit is configured to send the data of the user behavior path map and the data of the abnormal nodes to the user terminal so that the user terminal can visualize the user behavior path map and the abnormal nodes.

16. A data processing apparatus, characterized in that, The data processing device, applied to the user end, includes: The acquisition unit is configured to acquire data of a user behavior path graph and data of abnormal nodes in the user behavior path graph. The user behavior path graph is constructed based on user behavior sequence data. The nodes in the user behavior path graph are used to represent various functional modules in the target application. The edges in the user behavior path graph are used to represent the user's jump path between the various functional modules. The abnormal nodes are determined by analyzing the user behavior path graph. The display unit is configured to display nodes in the user behavior path graph using a first style, edges in the user behavior path graph using a second style, and abnormal nodes using a third style on the user interface. The first style is used to represent the user's access behavior metric value to each functional module, and the second style is used to represent the user's jump behavior metric value between each functional module.

17. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed, it implements the steps of the data processing method according to any one of claims 1-14.

18. An electronic device, characterized in that, include: One or more processors; as well as A memory associated with the one or more processors, the memory being used to store program instructions that, when read and executed by the one or more processors, perform the steps of the data processing method according to any one of claims 1-14.

19. A computer program product, comprising a computer program, characterized in that, When the computer program is executed, it implements the steps of the data processing method according to any one of claims 1-14.

Images