User behavior path data acquisition and path division method, electronic equipment and computer storage medium
By automatically collecting user event data and constructing a Sankey diagram, the problem of time-consuming, labor-intensive and inaccurate user operation path analysis in existing technologies is solved, and fast and accurate user path diagram construction is achieved.
Patent Information
- Application Number
- CN202510967427.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-14
- Publication Date
- 2025-10-17
AI Technical Summary
The user operation path analysis method based on preset behavior paths in the existing technology is time-consuming, labor-intensive and inaccurate, cannot clearly display the user's true behavior path, and is interfered with by irrelevant data.
By automatically collecting user event data, using Sankey diagrams and adjacency lists to build user path diagrams, using event serialization records and reasonable session divisions, calculating PV/SV indicators, constructing Sankey diagrams and displaying user behavior paths.
It achieves fast and accurate user behavior path analysis, scientifically and rationally constructs user path maps, reduces manual intervention, and improves data collection efficiency and the accuracy of path display.
Smart Images
Figure CN120803873A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of display screen operation behavior, and particularly relates to a user behavior path data collection and path division method, an electronic device and a computer storage medium. BACKGROUND
[0002] With the increasingly refined operation requirements of Internet intelligent display screens, it is necessary to analyze the path distribution of users when using the intelligent display screen system, to mine the usage habits of users, to find mainstream processes, to find ignored value points, and to find user loss points, which are key means for enterprises to find data growth points.
[0003] Path analysis generally refers to fine tracking and recording of each operation behavior of a user, and obtaining detailed operation behavior path characteristics of the user, conversion characteristics of each step, and sources and destinations of each step through analysis and mining on this basis, thereby helping enterprises to analyze a series of operation behaviors of the user on the intelligent display screen terminal.
[0004] However, the existing solution generally calculates the behavior path of the user by adopting a manual configuration funnel mode based on a preset behavior path, and the path diagram obtained by calculation is only the step-by-step browsing volume of each path node on the preset path, which cannot clearly and intuitively show the real behavior path of the user, and is disturbed by some irrelevant data, resulting in inaccurate user behavior path, time-consuming and laborious. SUMMARY
[0005] To solve the problems in the prior art, the present application aims to provide a user behavior path data collection and path division method, an electronic device and a computer storage medium, and the present application realizes scientific and reasonable user path diagram construction.
[0006] To achieve the above-mentioned purpose, the technical solution adopted by the present application is as follows: a user behavior path data collection and path division method, comprising the following steps:
[0007] Step 1: automatically collecting data when a user triggers an event;
[0008] Step 2: performing session division on a series of interactive behaviors occurring on a system or a page within a specified time period, and distinguishing and identifying different paths;
[0009] Step 3: performing deduplication on adjacent pages of the distinguished and identified paths, and obtaining at least one level of pages before and after each page;
[0010] Step 4: counting the access times PV and session times SV of positive and negative paths of each level of pages, wherein the access times PV refer to the number of accesses within a period of time, and the session times SV refer to the number of sessions in which the corresponding access path appears.
[0011] Step 5, constructing a Sankey diagram to show the path of the user.
[0012] As a further improvement of the present application, in step 1, when automatically collecting data, the id_previous field is carried in the next event to mark the previous event, which is used to record the sequence of event occurrence, and complete the event serialization record.
[0013] As a further improvement of the present application, in step 2, according to the preset time interval, when the time interval of two interactive behaviors exceeds the preset time interval, the two interactive behaviors are divided into different paths.
[0014] As a further improvement of the present application, in step 3, the at least one level of page before and after each page in each session is obtained by window function analysis.
[0015] As a further improvement of the present application, in step 4, the access frequency PV and the session frequency SV of the path are calculated respectively to obtain the divided session data of the intermediate state; then, according to the session data, the access frequency PV and the session frequency SV are calculated according to each level group to obtain the access frequency PV and the session frequency SV of each level path, and finally all the results are combined.
[0016] As a further improvement of the present application, in step 5, a node is taken as a starting point or an end point to construct the behavior path diagram of the user, which can be converted into an adjacency list of a weighted tree, and the Sankey diagram is divided according to levels, and each two levels are represented by an adjacency list.
[0017] As a further improvement of the present application, the following four indexes are calculated for each link of the Sankey diagram: the pv / sv of each node, the conversion rate of each node, the pv / sv between nodes, and the conversion rate between nodes, and the specific index definitions are as follows:
[0018] Node pv / sv = the pv / sv sum of the current node in the current level;
[0019] Node conversion rate = (node pv / sv) / (path starting node pv / sv);
[0020] Inter-node pv / sv = the pv / sv of the flow from the previous level node to the current node;
[0021] Inter-node conversion rate = (inter-node pv / sv) / (previous level node pv / sv);
[0022] According to the user input, the above indexes are calculated and stored as the key information of the Sankey diagram node.
[0023] As a further improvement of the present application, the node depth and the node at each level recorded in the path table are finally filtered from the data in the path table to screen out the complete path, eliminate the intermediate path, and organize the path data into a tree structure, while the relevant indicators calculated are associated with the node connection line, and finally presented to the front end for user path graph display.
[0024] The application also discloses an electronic device, comprising a memory, a processor and a computer program stored in the memory and executed by the processor.
[0025] The processor is connected with the memory, and a user realizes the user behavior path data collection and path division method as described above by executing the computer program.
[0026] The application also discloses a computer storage medium, which stores a computer program; the computer program is executed to realize the user behavior path data collection and path division method as described above.
[0027] In order to ensure effective recording of user behavior events and formation of available user behavior event sequences, the application designs an event collection optimization scheme of a data collection scheme to optimize the target with minimum change cost, establishes user behavior event sequences, lays a foundation for construction of a user behavior path model, and makes it possible. The application optimizes the data collection scheme of the intelligent display screen, and designs a reasonable preset time period strategy to establish a basic data construction model of each node in the user behavior path graph, so as to realize basic data accumulation of the user behavior path graph.
[0028] The application has the following beneficial effects:
[0029] The application has the following beneficial effects: 1. Data collection: a kind of event sequence construction method is designed for the data collection scheme of the intelligent display screen, which records the previous event of the current event by using a special mark, to complete time sequence recording; 2. Session division and strategy: a series of user behavior operation sequences of the intelligent display screen are divided according to a reasonable event interval, so as to divide into different behavior paths; 3. Construction of user path graph: a Sankey diagram is used as a display method of the user path graph; After the above three aspects are optimized and constructed, the existing data collection system is fully and quickly utilized, a user path analysis method is mined and constructed, and the user path construction of the full intelligent display screen is realized; BRIEF DESCRIPTION OF DRAWINGS
[0030] Figure 1 It is a schematic diagram of event sequencing of the data collection end in the embodiment of the application;
[0031] Figure 2 A schematic diagram of session division in the embodiment of the present application;
[0032] Figure 3 A schematic diagram of four-level page records before and after session 0 stage of a user in the embodiment of the present application;
[0033] Figure 4 A schematic diagram of partial session path records of a user and a user b in the embodiment of the present application;
[0034] Figure 5 A PV and UV calculation example diagram of two path segments of a user in the embodiment of the present application;
[0035] Figure 6 A schematic diagram of a sand chart-adjacency list in the embodiment of the present application;
[0036] Figure 7 A schematic diagram of a sand chart-weighted tree in the embodiment of the present application. DETAILED DESCRIPTION
[0037] The embodiments of the present application will be described in detail below with reference to the accompanying drawings.
[0038] Embodiment 1
[0039] Starting from the actual problem that an event sequence cannot be formed from the existing collection scheme, the present embodiment designs a scheme for forming an event sequence, which can realize the connection of user behavior events to form an optimal decision for analysis; at the same time, through big data analysis technology, the preset time period is re-divided, so as to obtain the most suitable time period, and a session is defined by the time period. After seeking a time period more suitable for the user of the intelligent display screen, the event sequence is divided, so as to obtain a session more suitable for the series behavior of the user of the intelligent display screen; a user behavior path data collection and path division method, comprising:
[0040] (1) Automatic data collection:
[0041] The existing intelligent terminal data collection collects data and reports according to certain rules, assembles data and transmits to the background when the user triggers the corresponding event.
[0042] The present embodiment designs an event sequence formation method based on the existing intelligent display screen terminal data collection architecture, that is, an id_previous field is carried in the next event to mark the previous event, so as to record the sequence of event occurrence, and the event serialization record is completed as shown in Figure 1 .
[0043] (2) Session division:
[0044] Session, in data analysis, refers to a series of interactions that occur on a system or page within a specified time period. Session segmentation aims to distinguish and identify different paths, that is, to set a reasonable time interval. When the interval between two behaviors exceeds the set time interval, it is considered that the two behaviors do not belong to the same path.
[0045] The user behavior path model is based on various event ids to obtain corresponding element ids for page element path analysis. Taking the behavior events of a user on a certain day as an example, session segmentation is performed. The behavior events of a user on a certain day are E1, E2, E3,..., the corresponding pages are P1, P2, P3,..., the event occurrence times are T1, T2, T3,..., and the set session interval is ti, as shown in Figure 2 .
[0046] (3) Adjacent page deduplication and acquisition of the previous and next four levels of each page:
[0047] Different behavior operations may correspond to the same page. After session segmentation, adjacent page deduplication is performed to filter out the same pages.
[0048] Through window function analysis, the previous and next four levels of each page in each session are obtained. The sid is spliced according to the user identifier ID and session number, as shown in Figure 3 , Figure 3 The page column in the above logic is the current page, and the empty page is represented by -1. According to the above logic, all records are calculated.
[0049] (4) Statistics of positive and negative path PV / SV and PV / SV of each level path:
[0050] PV, which stands for Page View, refers to the number of visits in a certain period of time. In this model, it refers to the number of visits in a certain period of time. SV, which stands for Session View, refers to the number of sessions. In this model, it refers to the number of sessions that have visited the path. For example, there are path one: A→B→C→D→A→B and path two: A→B→D, then the PV of A→B is 2+1=3, and the SV is 1+1=2.
[0051] For a certain page, the positive and negative paths are counted respectively, where: Figure 3 is taken as an example: next corresponds to the positive path, and previous corresponds to the negative path.
[0052] According to the previous calculation logic and user behavior data, the following temporary data table can be obtained:
[0053]
[0054] For a better understanding, we can extract the row data of mark_id=m5 from the above table. This is the result data of a forward path (path_direction=1). Figure 3 It is represented as a path from left to right, which corresponds to Figure 4 There are two sets of session data for user a in the first row, session=S0, and the third row, session=S2. This path appears twice in the path record of user a (see Figure 5 (marked in red box), so: user a's forward path has SV=2, PV=2. Then, from the above table, we can extract the row data with mark_id=m6, whose path_direction=1 is also the forward path result data, and its path corresponds to Figure 4 The second row of session=S1 data for user a in the , this path occurs twice in the same session (see Figure 5 Orange box), so SV=1, PV=2.
[0055] Calculate the path's PV and SV (deduplicated by SID) to obtain intermediate session data. Then, based on this data, calculate the sum of PV and SV by page_id_lv1 to obtain the PV and SV of the first-level path. Similarly, calculate the PV and SV of the second, third, fourth, and fifth-level paths, and merge all the results.
[0056] (5) Constructing a Sankey diagram:
[0057] Sankey diagrams, as a representation of user path graphs, can be simplified to a graph compression storage problem. This model uses adjacency lists for storage. Adjacency lists are a commonly used graph compression storage structure. They use linked lists to store nodes and edges in a graph while ignoring non-existent edges between nodes, thereby compressing the matrix. To meet the needs of the model, the vertex and edge node structures were modified.
[0058] According to the requirement description, to meet the construction of the user's behavior path diagram with a certain node as the starting point (or end point), it can be converted into an adjacency list to construct a weighted tree, such as Figure 6 and Figure 7 shown.
[0059] Figure 6 This is the adjacency list used in the model. Vertex represents the node structure, storing node information such as the node name (name), node code (code), related information (info), and a pointer to the edge list (Edges). Edge represents the edge structure, storing the weight of the current edge, endpoint information, and a pointer to the next edge at the same node. Each vertex points to an edge list (Edge).
[0060] In the Sankey diagram, the same name with different conversion rates appears in different levels, which is a ring of the path and cannot be regarded as a repeated node according to the name, and does not form a loop. If the entire Sankey diagram is represented by an adjacency list, such nodes will be treated as the same node, so that a loop appears in the image. Therefore, the embodiment divides the Sankey diagram according to the level, and represents each two levels by an adjacency list, as shown in Figure 7 Level 1 represents the nodes of level 1 and the edges pointing to level 2, Level 2 represents the nodes of level 2 and the edges pointing to level 3, and so on.
[0061] Embodiment 2
[0062] A user behavior path data collection and path division method, comprising:
[0063] 1. Automatic collection scheme:
[0064] After sufficient data analysis and research on the existing behavior events reported by the intelligent display screen terminal, it is found that the existing data collection strategy cannot meet the perfection of the event sequence, therefore, the following optimization points are proposed:
[0065] (1) Emphasize the normative definition and description of behavior events, define and describe all existing events, and unify the description of all types of events.
[0066]
[0067] (2) On the basis of the original collection strategy, add the data collection field event_id_previous, which is used to mark the previous event of this event, and report the event attribute data when triggering the specified event, to complete the serialization of the event.
[0068] As shown in Figure 1 , it is the event attribute structure diagram of the collection event provided by the embodiment of the application, wherein event_id_previous is the previous event attribute field, and the event attribute data structure can be applied to each type of event, and finally serves as part of the system application collection device. Figure 1
[0069] 2. Session division:
[0070] The user interaction behavior of the intelligent display screen has its particularity, and the division of the session should meet the user operation behavior of the intelligent display screen.
[0071] (1) The setting of the session time interval needs to be analyzed according to the user use history of the intelligent display screen to obtain a more reasonable time interval.
[0072] From the clustering analysis of the users with active behavior every month in the past year, the most common time interval data of the group is obtained; the conclusion is verified to the users with boot-up behavior in the past year, and the time interval dimension is dispersed, and finally, the data set is sorted and the reasonable session time interval is selected from the data set.
[0073] (2) According to the time interval obtained by the above steps, the operation behavior of all users is divided into sessions, and the division logic and division method are divided according to Figure 2 The behavior events of a user on the same day are taken as an example to divide the session, and the behavior events of a user on the same day are E1, E2, E3,..., the pages corresponding to the events are P1, P2, P3,..., and the time of the events is T1, T2, T3,..., and the set session interval is ti. For the same user, the interval time between the operation pages corresponding to the two behavior events is greater than the set event interval ti, and the two sessions are divided.
[0074] The above session division is performed on the objects to be analyzed, and the data is prepared for subsequent path analysis.
[0075] 3. Path calculation:
[0076] (1) Before the next calculation and analysis of the sessions divided in the above steps, the adjacent pages are de-duplicated to obtain the session segment after standard processing. Using window function analysis, the previous four pages of each page in each session are obtained, wherein sid is spliced according to the user identification ID and session number, as shown in Figure 3 , the page column in Figure 3 is the current page, and the empty page is represented by -1.
[0077] According to the above logic, the previous four levels of all divided sessions and pages are calculated.
[0078] (2) Statistics of positive path PV / SV:
[0079] According to the business requirements, the PV / SV of the path and the PV / SV of each level of the path need to be counted, and the positive path and negative path of a page on the path are counted in Figure 3
[0080] According to sid (i.e. session number) deduplication, the pv and sv of the path are calculated respectively to obtain the session data of the intermediate state. Then, according to the intermediate state data, the sum of pv and sv is calculated according to page_id_lv1 grouping to obtain the pv and sv of the first-level path, and then the pv and sv of the second, third, fourth and fifth-level paths are calculated respectively in a similar manner, and all results are combined.
[0081] Obtain behavior data of a user of an intelligent display screen and at least one user operation process, wherein the user operation process comprises: event identification, user behavior activity, event attribute, event reporting content information, previous event identification, and mapping relationship between executed events; based on the event data of the user, an event sequence of user behavior can be recorded in the user operation process; based on the event index sequence of the user, a complete behavior path of the whole user can be constituted, and the behavior path of the user in a certain period of time is split based on the result; based on the segmented split of the behavior path, a user behavior path analysis of the intelligent display screen based on clustering can be finally formed.
[0082] Obtain the obtained user behavior data, including: intelligent display screen opening relationship behavior data; intelligent display screen portal browsing data; intelligent display screen application startup and application exit data; intelligent display screen playback data; intelligent display screen element click data; intelligent display screen voice wake-up data; intelligent display screen corresponding remote control key click data.
[0083] The intelligent display screen application startup and exit data based on the obtained user behavior data, including: intelligent display screen system Launcher startup and exit data; intelligent display screen system self-provided application startup and exit data, such as: system settings, global search, etc.; intelligent display screen system third-party application startup and exit data, such as: live TV, application store, video application, audio application, etc.
[0084] Obtain the intelligent display screen playback data based on the obtained user behavior data, including: intelligent display screen system video playback data; intelligent display screen system audio playback data; intelligent display screen system picture playback data; intelligent display screen system program source playback data; intelligent display screen system screen projection playback data;
[0085] Obtain the intelligent display screen element click data based on the obtained user behavior data, including: intelligent display screen Launcher system poster click data; intelligent display screen Launcher system special topic click data; intelligent display screen Launcher system component click data; intelligent display screen Launcher system background broadcast video click data; intelligent display screen system screen saver click data; intelligent display screen system pop-up prompt information data.
[0086] Acquiring voice wake-up data based on the acquired user behavior data, including: far-field voice wake-up data of the smart display screen; voice key wake-up data of the remote controller paired with the smart display screen.
[0087] 4, the construction of the Sankey diagram:
[0088] According to the Sankey diagram node display requirements, the following four indicators need to be calculated for each link of the Sankey diagram: pv / sv of each node, conversion rate of each node, pv / sv between nodes, and conversion rate between nodes. The specific indicator definitions are as follows:
[0089] Node pv / sv = pv / sv sum of the current node in the current level
[0090] Node conversion rate = (node pv / sv) / (path starting node pv / sv)
[0091] Inter-node pv / sv = pv / sv of the flow from the previous node to the current node
[0092] Inter-node conversion rate = (inter-node pv / sv) / (previous node pv / sv)
[0093] According to the user input, the above content is calculated as the key information of the Sankey diagram node storage to support subsequent data presentation.
[0094] According to the design idea of the embodiment, based on the user behavior record, a complete reconstructed adjacency list is constructed. According to the design of the model, the Sankey diagram is divided into levels, and each two levels are represented by an adjacency list, as shown in Figure 6 and Figure 7 .
[0095] The implementation is realized by constructing a weighted path tree. The user path is finally converted into a weighted tree adjacency table to obtain the Sankey diagram under the complete path.
[0096] Combining the presentation requirements of the user path diagram, the complete path is finally selected from the path table data by filtering out the intermediate path, and the path data is organized into a tree structure. At the same time, the related indicators calculated are associated with the node connection line, and finally presented to the front end for user path diagram display.
[0097] The above embodiments only express the specific implementation of the present application, and the description is more specific and detailed, but it cannot be understood as a limitation on the scope of the present application. It should be noted that for ordinary skilled persons in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are within the scope of the present application.
Claims
1. A method for collecting user behavior path data and dividing paths, characterized in that: The following steps are involved: Step 1: When the user triggers an event, data is automatically collected; Step 2: Divide the interactive behaviors that occurred on the system or page within a specified time period into sessions and identify different paths; Step 3: De-duplicate adjacent pages of the identified path and obtain at least one level of pages before and after each page; Step 4: Count the number of visits (PV) and the number of sessions (SV) for the positive and negative paths of each level of page. PV refers to the number of visits within a period of time, and SV refers to the number of sessions that have occurred in the corresponding access path. Step 5: Construct a Sankey diagram to display the user's path.
2. The user behavior path data collection and path division method according to claim 1 is characterized in that: In step 1, when automatically collecting data, the id_previous field is carried in the next event to mark the previous event, which is used to record the order in which the events occurred and complete the event serialization record.
3. The user behavior path data collection and path division method according to claim 1 is characterized in that: In step 2, according to a preset time interval, when the time interval between two interactive behaviors exceeds the preset time interval, the two interactive behaviors are divided into different paths.
4. The user behavior path data collection and path division method according to claim 1 is characterized in that: In step 3, at least one level of pages before and after each page in each session is obtained through window function analysis.
5. The user behavior path data collection and path division method according to claim 1 is characterized in that: In step 4, the number of visits PV and the number of sessions SV of the path are calculated separately to obtain the divided session data of the intermediate state; then, based on the session data, the number of visits PV and the number of sessions SV are calculated according to each level of grouping to obtain the number of visits PV and the number of sessions SV of each level of path, and finally all the results are merged.
6. The method for collecting user behavior path data and dividing paths according to claim 1, characterized in that: In step 5, a user's behavior path diagram is constructed with a certain node as the starting point or end point, which can be converted into an adjacency list for constructing a weighted tree, and the Sankey diagram is divided into levels, with each two levels represented by an adjacency list.
7. The method for collecting user behavior path data and dividing paths according to claim 6, characterized in that: The following four indicators are calculated for each link of the Sankey diagram: pv / sv of each node, conversion rate of each node, pv / sv between nodes, and conversion rate between nodes. The specific indicators are defined as follows: Node pv / sv = the sum of the pv / sv of the current node in the current level; Node conversion rate = (node pv / sv) / (path starting node pv / sv); PV / SV between nodes = PV / SV from the previous node to the current node; Inter-node conversion rate = (inter-node pv / sv) / (previous-level node pv / sv); Based on the user input, the above indicators are calculated and stored as the key information of the Sankey diagram node.
8. The method for collecting user behavior path data and dividing paths according to claim 7, characterized in that: The depth of each node and each level of nodes recorded in the path table are finally filtered out from the data in the path table, the intermediate paths are eliminated, and the path data is organized into a tree structure. At the same time, the calculated relevant indicators are associated with the node connection lines, and finally presented to the front end for displaying the user path map.
9. An electronic device, characterized in that: include: a memory, a processor, and a computer program stored on the memory and executed by the processor; The processor is connected to the memory, and the user implements the user behavior path data collection and path division method as described in any one of claims 1 to 8 by executing the computer program.
10. A computer storage medium, characterized in that The computer storage medium stores a computer program; after the computer program is executed, the user behavior path data collection and path division method according to any one of claims 1 to 8 is implemented.