User behavior burying point and user classification method based on event driving

Abstract

The invention relates to a user behavior burying point and analysis method based on event driving, and the method comprises the following steps: designing an atomic event, a combined event and a composite event as hierarchical event models, determining the identification parameter of each event, classifying each event into a key event, a non-core event and a basic behavior, performing different burying points on each event according to the type of the event to obtain an original behavior data set acquired by mixed burying points; associating the identification parameters of the original behavior data set to the same identification parameters through a matching algorithm, and adopting the original behavior data set after Kafka buffer association to obtain a structured behavior data set after unified identity association; and analyzing the structured behavior data set from multiple dimensions to obtain user behavior characteristics and product experience, and providing differentiated guidance based on the user behavior characteristics and the product experience. Compared with the prior art, the method has the advantages of shortening the service demand response time and the like.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of user behavior tracking, and in particular to a user behavior tracking and user classification method based on event driving. BACKGROUND

[0002] Tracking is a technical means of collecting user behavior data in Internet products. By implanting statistical code in the code, the specific operation of the user in the product is tracked. The core role of tracking: collecting user behavior data, such as clicking buttons, browsing pages, completing transactions, etc. Provide the basis for product optimization, understand user preferences and product pain points through data analysis. Support operational decision-making, such as evaluating the effectiveness of marketing activities and optimizing user conversion paths.

[0003] The current mainstream tracking technology types mainly include code tracking, visual tracking, full tracking and no tracking. Code tracking is implemented by manually embedding code, which is flexible and accurate but has high development cost; visual tracking is completed through interface configuration, which is suitable for non-technical personnel but has limited coverage scenarios; full tracking automatically collects all user behavior data, which is comprehensive but has large data volume and redundancy; no tracking is implemented through SDK, which is simple but lacks deep customization capabilities. Each type of tracking has its own advantages and disadvantages, and needs to be selected comprehensively according to business requirements and data precision requirements.

[0004] From the perspective of data standardization, due to the lack of unified minimum operation unit definition and identification parameter specification of atomic events, the behavior data collected by different platforms and different tracking methods will be in a chaotic format, which makes it impossible to directly correlate cross-platform data. Additional manpower is needed to align and complete the data format during subsequent analysis, which not only increases the cost of data processing, but also may cause some key data to be unusable due to format differences, damaging data integrity. At the same time, when the business requirements change, the tracking code needs to be redeveloped for the complete business process after the change, and it needs to wait for the product to be released before it takes effect. This not only significantly prolongs the response time of business requirements, but also increases the risk of system iteration due to frequent development and release, and cannot quickly adapt to sudden changes in market or user demand. SUMMARY

[0005] The purpose of the present application is to shorten the response time of business requirements and provide a user behavior tracking and user classification method based on event driving.

[0006] The purpose of the present application can be achieved by the following technical solutions: A user behavior tracking and user classification method based on event driving, the method comprising the following steps: Atomic events, combined events, and composite events are designed as hierarchical event models. Identification parameters for each event are determined. The hierarchical event models and identification parameters are combined to form the data format of each event. Each event is classified into key events, non-core events, and basic behaviors. Different event tracking points are applied to each event according to its type to obtain the original behavior dataset collected by mixed tracking points. The original behavior dataset includes the data collected for each event and the corresponding identification parameters and hierarchical event models. The identification parameters of the original behavior dataset are associated with the same identification parameters by a matching algorithm, and the associated original behavior dataset is buffered by Kafka to obtain a structured behavior dataset with unified identity association. By analyzing structured behavioral datasets from multiple dimensions, we can obtain user behavior characteristics and product experience, and provide differentiated guidance based on these characteristics and experience.

[0007] Furthermore, the atomic event is the smallest unit of operation, including click, page loading and redirection, and submission; the combined event is an atomic event that is connected in series to form a closed loop; and the composite event is a combination of events that are connected in series or in parallel.

[0008] Furthermore, the method also includes: Obtain the hierarchical event model configuration of the event. If the hierarchical event model configuration changes, construct a new hierarchical event model configuration based on atomic events, combined events, and composite events.

[0009] Furthermore, the identification parameters include user ID and device ID.

[0010] Furthermore, the specific steps for implementing different tracking points for each event based on its type are as follows: For critical business events, code tracking is implemented, and SDK code is embedded to collect business parameters. For non-core events, visual tracking is performed, and interface elements are selected through the management platform to automatically monitor and collect business parameters. For basic event execution without code tracking, the SDK collects all basic user behaviors.

[0011] Furthermore, the specific steps to obtain a structured behavior dataset with unified identity association by using Kafka to buffer and associate the original behavior dataset are as follows: The original behavior dataset after association is buffered using Kafka. The original behavior dataset after association is preprocessed according to the priority scheduling order to obtain a structured behavior dataset after unified identity association.

[0012] Furthermore, the preprocessing involves cleaning up outliers such as those with disordered timestamps, missing parameters, and incorrect data formats.

[0013] Furthermore, user behavior characteristics include: operating habits, operating skills, and reaction ability.

[0014] Furthermore, the product experience is defined by: identifying drop-off points through funnel analysis, identifying inefficient paths through path friction detection, and obtaining version conversion rates through A / B testing.

[0015] Furthermore, the specific steps for providing differentiated guidance based on user behavior characteristics and product experience are as follows: Users are categorized into experts, novices, and high-risk users based on their behavioral characteristics and product experience, and different guidance is provided for each category.

[0016] Compared with the prior art, the present invention has the following beneficial effects: By designing a hierarchical event model composed of atomic events, combined events, and composite events, and clarifying the identification parameters of each event to form a unified data format, this method supports the dynamic generation of new hierarchical event models based on changes in the configuration of the hierarchical event model. This method has significant advantages in multiple dimensions: From the perspective of data standardization, atomic events, as the smallest unit of operation, clearly define the core behaviors such as clicks, page loading and redirection, and submissions. Combined with unified identification parameters, a standardized data format is constructed, which can completely solve the problems of chaotic data formats and inconsistent parameters for different platforms and behaviors in traditional event tracking. This ensures that all collected data has a consistent structure and identifiability, laying the foundation for subsequent cross-scenario and cross-platform data association and analysis. Combined events form business loops by connecting atomic events, while composite events cover complex business processes by combining events in series or parallel. This hierarchical design allows the event model to accurately match business scenarios of varying complexity. When business processes are adjusted, such as adding operation steps or changing the process order, resulting in changes to the hierarchical event model configuration, there is no need to redevelop the event tracking code. Instead, a new hierarchical event model can be dynamically reconstructed based on existing atomic and combined events, greatly shortening the response time to business requirements and avoiding the drawbacks of traditional event tracking that require re-release due to business changes. At the same time, the hierarchical event structure ensures that data is no longer an isolated operation record. The fine-grained recording of atomic events makes it possible to trace every basic user behavior, while combined and composite events can restore the complete business chain. Combined with dynamic adaptation capabilities, this ensures that data collection can still fully cover the new chain after business process changes. This provides continuous, complete, and accurate data support for subsequent in-depth analysis such as user behavior path analysis, business funnel conversion calculation, and product experience bottleneck identification, thereby improving the accuracy of user segmentation and the scientific nature of product iteration decisions. Attached Figure Description

[0017] Figure 1 This is a flowchart of the present invention. Detailed Implementation

[0018] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, providing detailed implementation methods and specific operation processes; however, the scope of protection of the present invention is not limited to the following embodiments. The present invention proposes an event-driven user behavior tracking and user classification method, the method comprising the following steps: Atomic events, combined events, and composite events are designed as hierarchical event models. Identification parameters for each event are determined. The hierarchical event models and identification parameters are combined to form the data format of each event. Each event is classified into key events, non-core events, and basic behaviors. Different event tracking points are applied to each event according to its type to obtain the original behavior dataset collected by mixed tracking points. The original behavior dataset includes the data collected for each event and the corresponding identification parameters and hierarchical event models. The identification parameters of the original behavior dataset are associated with the same identification parameters by a matching algorithm, and the associated original behavior dataset is buffered by Kafka to obtain a structured behavior dataset with unified identity association. By analyzing structured behavioral datasets from multiple dimensions, we can obtain user behavior characteristics and product experience, and provide differentiated guidance based on these characteristics and experience. Figure 1 This is a flowchart of the present invention.

[0019] Atomic events are the smallest units of operation, including clicks, page loading and redirection, and submissions. Combined events are atomic events that are connected in series to form a closed loop. Composite events are combined events that are connected in series or in parallel.

[0020] The hierarchical event model is designed as follows: Atomic events: Define the smallest unit of operation (such as click, page loading and redirection, submission, etc.), and describe it in a standardized way using atomic event ID and parameters.

[0021] Combined events: These connect the above atomic events to form a closed business loop (e.g., "Pullover Event": Pullover menu → Pullover page loading → Pullover data query → Amount verification input → Pullover operation → Pullover details page). Dynamic configuration of atomic event sequence rules is supported to adapt to changes in business processes.

[0022] Composite events: These are combinations of the above events, either sequentially or in parallel, forming a business loop (e.g., a "composite transaction event": order placement event → third-party redirection → order return event → detailed query event). Dynamic configuration of composite event sequence rules is supported to adapt to changes and restructuring of business processes.

[0023] Meanwhile, this invention standardizes cross-platform parameter naming (such as userId, sessionId, atomicId, eventId, etc.) to ensure data consistency and easy identification. It also controls parameter specifications.

[0024] The method also includes: Obtain the hierarchical event model configuration of the event. If the hierarchical event model configuration changes, construct a new hierarchical event model configuration based on atomic events, combined events, and composite events.

[0025] The identification parameters include user ID and device ID.

[0026] The specific steps for implementing different tracking points for each event based on its type are as follows: For critical business events, code tracking is implemented, and SDK code is embedded to collect business parameters. For non-core events, visual tracking is performed, and interface elements are selected through the management platform to automatically monitor and collect business parameters. For basic event execution without code tracking, the SDK collects all basic user behaviors.

[0027] Among them, code tracking: key business events (such as clicks, redirects, etc.) are embedded in SDK code to accurately collect business parameters.

[0028] Visualized event tracking: Non-core events (such as page views and area clicks) can be automatically monitored by selecting interface elements through the management platform, reducing development costs and highlighting key events.

[0029] No-code tracking assistance: Fully collect basic user behaviors (scrolling, dwell time, etc.) for exploratory analysis and application.

[0030] This invention also performs dynamic event tracking management, and the configuration center supports real-time distribution of event collection rules, allowing for the addition / modification of event definitions without requiring a release.

[0031] Edge computing is used to preprocess data (such as filtering invalid clicks) to reduce server load.

[0032] The specific steps to obtain a structured behavior dataset after unified identity association by using Kafka to buffer the original behavior dataset after association are as follows: The raw behavioral dataset after association is buffered using Kafka. This dataset is then preprocessed according to priority scheduling to obtain a structured behavioral dataset with unified identity association. Specifically, unified identity association involves userId fusion, using multiple source identifiers such as device ID, mobile phone number, WeChat ID, and mini-program, combined with algorithms to match cross-platform behaviors.

[0033] The preprocessing involves cleaning up outliers such as incorrect timestamps, missing parameters, and incorrect data formats.

[0034] User behavior characteristics include: operating habits, operating skills, and reaction ability.

[0035] Among them, operating habits: Path frequency analysis (high-frequency path: A→B→C vs. abnormal path: A→C); Time preference models (such as heatmaps of active periods).

[0036] Operational level: Completion rate of key events (such as the frequency of abnormal operation prompts); Error rate statistics (e.g., duplicate form submissions, consecutive clicks); Bottlenecks in the interface interaction of heatmap recognition.

[0037] Reaction ability: Step interval duration (e.g., the delay from clicking to submit); Efficiency in completing complex tasks (such as the time spent on multi-step configuration). Product experience is measured by: churn points identified through funnel analysis, inefficient paths identified through path friction detection, and version conversion rates obtained through A / B testing.

[0038] Product experience quantitative evaluation: Funnel analysis: Identifying the loss process.

[0039] Path friction detection: Identifies inefficient paths with excessive jumps and high backtracking rates.

[0040] A / B testing supports comparing version conversion rates based on tracking data.

[0041] A user profile and tagging system can be developed based on user behavior characteristics. Basic attributes: region, device type, and channel source.

[0042] Behavioral tags: such as "high frequency of use but low conversion rate" and "new users are prone to churn".

[0043] The specific steps for providing differentiated guidance based on user behavior characteristics and product experience are as follows: Users are categorized into experts, novices, and high-risk users based on their behavior and product experience. Different guidance is provided for each category, categorized by operational proficiency (expert / novice), with differentiated guidance (e.g., simplified beginner processes). High-risk user identification (e.g., multiple failed operations) triggers automated work orders in real time.

[0044] Furthermore, product iteration can be driven by user behavior characteristics and product experience: optimizing button placement through operation heatmaps; and compressing redundant steps based on event completion time (such as improving completion rate after optimizing a certain path). If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

Claims

1. An event-driven user behavior tracking and user classification method, characterized in that, The method comprises the following steps: The atomic event, the combination event and the composite event are designed as a hierarchical event model, the identification parameter of each event is determined, the hierarchical event model and the identification parameter are combined to form a data format of each event, each event is classified as a key event, a non-core event and a basic behavior, different burying points are performed on each event according to the type of the event, and an original behavior data set collected by mixed burying points is obtained, the original behavior data set comprises data collected by each event and corresponding identification parameters and hierarchical event models respectively; The identification parameters of the original behavior data set are associated to the same identification parameters by a matching algorithm, and the original behavior data set after association is buffered by using Kafka, so as to obtain a structured behavior data set after association of unified identities. The structured behavior data set is analyzed from multiple dimensions to obtain user behavior characteristics and product experience, and difference guidance is provided based on the user behavior characteristics and the product experience.

2. The user behavior event-driven tracking and user classification method according to claim 1, wherein, The atomic event is a minimum operation unit, comprising clicking, page loading and jumping and submitting, the combination event is an atomic event forming a closed loop in series, and the composite event is a combination event in series or parallel.

3. The user behavior event-driven tracking and user classification method of claim 2, wherein, The method further comprises the following steps: The hierarchical event model configuration of the event is acquired, and if the hierarchical event model configuration changes, a new hierarchical event model corresponding to the hierarchical event model configuration is constructed according to the atomic event, the combination event and the composite event.

4. The user behavior event-driven tracking and user classification method of claim 1, wherein, The identification parameter comprises a user ID and a device ID.

5. The user behavior event-driven tracking and user classification method according to claim 1, wherein, The specific steps of performing different burying points on each event according to the type of the event are as follows: Code burying is performed on the key business event, SDK code is embedded, and business parameters are collected; Visual burying is performed on the non-core event, interface element automatic monitoring is performed through a management platform, and business parameters are collected; No burying is performed on the basic event, and user basic behaviors are collected through SDK.

6. The user behavior event-driven tracking and user classification method of claim 1, wherein, The specific steps of buffering the original behavior data set after association by using Kafka to obtain the structured behavior data set after association of unified identities are as follows: The original behavior data set after association is preprocessed according to a priority scheduling order to obtain the structured behavior data set after association of unified identities.

7. The user behavior event-driven based on the method of burying points and user classification according to claim 6, characterized in that, The preprocessing is to clean up abnormal values such as time stamp disorder, parameter missing and data format error.

8. The user behavior event-driven tracking and user classification method of claim 1, wherein, The user behavior characteristics are operation habits, operation level and reaction ability.

9. The user behavior event-driven tracking and user classification method of claim 1, wherein, The product experience is a loss link positioned through funnel analysis, an inefficient path identified through path friction detection and a version conversion rate obtained through A / B testing.

10. The user behavior event-driven tracking and user classification method of claim 1, wherein, The specific steps of providing difference guidance based on the user behavior characteristics and the product experience are as follows: Users are divided into experts, novices and high-risk users through the user behavior characteristics and the product experience, and different guidance is set for the experts, the novices and the high-risk users respectively.

Images