Data processing and user incentive method based on integral system
By building a user portrait labeling system and utilizing knowledge graphs to optimize the points rules, we have solved the problem of the disconnect between incentive strategies and user needs in the traditional points system, achieved precise operations and adaptive capabilities of the points system, and improved user engagement and operational efficiency.
Patent Information
- Application Number
- CN202510853888.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-24
- Publication Date
- 2025-10-17
AI Technical Summary
The traditional points system lacks an in-depth understanding of user behavior and the ability to provide differentiated incentives, resulting in insufficient incentives for high-value users and low participation among ordinary users. The long-term solidification of rules makes it difficult to adapt to market changes, the incentive strategy is out of touch with user needs, and there is a lack of a quantitative evaluation mechanism, making it difficult to form a closed-loop optimization.
By collecting multi-dimensional information about users, we build a user portrait labeling system, use knowledge graphs to integrate user portrait data, set up relationship links between users, labels and points rules, dynamically optimize rules based on the frequency of points increase and decrease, continuously monitor points changes and evaluate incentive effects, and form a closed-loop optimization mechanism.
It achieves precise operations, enhances user identification and participation, improves the scientific nature of rule-making and operational efficiency, gives the points system adaptive capabilities, avoids user loss caused by lagging rules, and promotes the replication and promotion of successful experiences.
Smart Images
Figure CN120807043A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of user motivation, in particular to a data processing and user motivation method based on an integral system. BACKGROUND
[0002] Under the background of increasingly fierce digital business competition, the integral system, as an important means of user motivation, is widely used in e-commerce, finance, service and other fields.
[0003] The traditional integral system often adopts fixed rules to calculate points based on a single dimension (such as consumption amount), lacking in-depth understanding of user behavior and differentiated motivation ability. This "one-size-fits-all" approach has significant drawbacks: on the one hand, it cannot accurately match the personalized needs of different user groups, resulting in insufficient motivation for high-value users and low participation for ordinary users; on the other hand, long-term fixed rules are difficult to adapt to market changes and dynamic evolution of user behavior, making the integral system gradually lose its appeal and even become an operational burden for enterprises. With the development of big data and artificial intelligence technology, some enterprises try to optimize the integral rules through user behavior analysis, but the existing solutions still have limitations. For example, rule adjustment based on simple data statistics lacks systematic construction of user portraits, making it difficult to tap into user potential needs; relying on human experience to set rules leads to low efficiency and easy deviation; at the same time, isolated data processing methods cannot effectively integrate multi-dimensional information, making it difficult to establish a deep connection between users, behavior and integral rules. In addition, the traditional integral system lacks a quantitative evaluation mechanism, making it difficult to scientifically measure the actual effect of rule adjustment and form a closed-loop optimization, resulting in a disconnect between motivation strategies and user needs, ultimately weakening the appeal of the integral system to users and its role in promoting business growth.
[0004] Therefore, it is necessary to provide a data processing and user motivation method based on an integral system to solve the problem of disconnect between motivation strategies and user needs in the prior art. SUMMARY
[0005] In view of this, the present application provides a data processing and user motivation method based on an integral system, aiming to solve the problem of disconnect between motivation strategies and user needs in the prior art.
[0006] The present application provides a data processing and user motivation method based on an integral system, comprising:
[0007] Collecting multi-dimensional information of users, and constructing a user portrait label system according to the multi-dimensional information of the users;
[0008] Integrating user portrait data in the user portrait label system through a knowledge graph, and setting relationship links between users, labels and integral rules;
[0009] Collect the frequency of increase and decrease of the user's points, judge whether the user's point rules need to be optimized according to the frequency of increase and decrease of the points, if it is judged that optimization is needed, analyze the association between the tags through the knowledge graph, mine the potential matching points between the user and the point rules, and adjust the point rules of the user according to the potential matching points;
[0010] Continuously monitor the user point change after the point rule adjustment, obtain an incentive effect index according to the user point change, and judge whether the preset incentive target is reached according to the incentive effect index;
[0011] If it is judged that the preset incentive target has been reached, similar users are obtained from the knowledge graph, and the adjusted point rules are pushed to the similar users, and if it is judged that the preset incentive target has not been reached, the potential matching points are mined again, and the point rules of the user are adjusted again.
[0012] Further, when the user multi-dimensional information is collected and the user portrait tag system is constructed according to the user multi-dimensional information, it includes:
[0013] With the help of the user registration form and the real-name authentication interface, the user type, industry field and registered capital are collected, the basic portrait framework of the user is built, and the user behavior information is collected; wherein the user behavior information includes transaction amount, transaction frequency, sharing times, browsing time and usage proportion;
[0014] The tags in the user portrait tag system are divided into four types, including value stratification tags, behavior preference tags and system tendency tags;
[0015] The value stratification tags are used to divide the users into different levels according to the transaction amount and transaction frequency of the users on the platform;
[0016] The behavior preference tags are used to collect the behavior habits of the users on the platform, and the users are divided into different levels according to the behavior habits; wherein the behavior habits include sharing times and browsing time;
[0017] The system tendency tags are used to record the usage proportion of the users to different subsystems of the platform.
[0018] Further, when the user portrait data in the user portrait tag system is integrated through the knowledge graph, and the relationship link between the user, the tag and the point rule is set, it includes:
[0019] The user basic portrait framework, each type of tag and the point rule are converted into entities in the knowledge graph, and each entity is given a unique identifier, wherein the user entity includes user ID and registration time, the tag entity includes tag ID, tag type and tag description, and the point rule entity includes rule ID and rule content;
[0020] Constructing a relationship link between entities, explicitly each user has a combination of labels and the current user's credit rules.
[0021] Further, the collection of user's credit increase and decrease frequency, according to the credit increase and decrease frequency to determine whether the user's credit rules need to be optimized, comprising:
[0022] Set the minimum value of increase and decrease frequency, if the credit increase and decrease frequency is less than the minimum value of increase and decrease frequency, it is judged that the user's credit rules need to be optimized;
[0023] Otherwise, it is judged that the user's credit rules do not need to be optimized.
[0024] Further, if it is judged that optimization is needed, the association between labels is analyzed through knowledge graph, and the potential matching point between the user and the credit rule is mined.
[0025] Using the relationship link in the knowledge graph, the association between different labels is analyzed, and the corresponding user of the label with association is obtained;
[0026] Obtain the user behavior information of the corresponding user, and calculate the similarity between the user behavior information of the corresponding user and the user behavior information of the user to be optimized;
[0027] Set a similarity threshold, and the credit rule corresponding to the user with a similarity greater than or equal to the similarity threshold is taken as a potential matching point.
[0028] Further, the credit rule of the user is adjusted for the potential matching point, comprising:
[0029] The credit rule of the corresponding user is added to the credit rule of the user to be optimized;
[0030] Delete the original credit rule in the credit rule to be optimized which conflicts with the newly added credit rule.
[0031] Further, the user credit change after the credit rule is adjusted is continuously monitored, and the incentive effect index is obtained according to the user credit change.
[0032] Obtain the user credit change value within a predetermined time after the credit rule is adjusted and the user historical credit change value within a historical interval time before the credit rule is adjusted;
[0033] If the user credit change value is greater than twice the user historical credit change value, the incentive effect index is a first index;
[0034] If the user score change value is less than or equal to twice the user score change value and greater than the user historical score change value, the incentive effect index is a second index;
[0035] If the user score change value is less than or equal to the user historical score change value, the incentive effect index is a third index;
[0036] Wherein, the first index > the second index > 1 > the third index > 0.
[0037] Further, when judging whether the preset incentive target is reached according to the incentive effect index, it includes:
[0038] A preset incentive target index, if the incentive effect index is greater than or equal to the incentive target index, it is judged that the preset incentive target is reached;
[0039] If the incentive effect index is less than the incentive target index, it is judged that the preset incentive target is not reached.
[0040] Further, if it is judged that the preset incentive target is reached, the similar user is obtained from the knowledge graph, and the adjusted score rule is pushed to the similar user, including:
[0041] Obtain a to-be-optimized user in the knowledge graph, and obtain user behavior information of the to-be-optimized user, denoted as first behavior information;
[0042] Calculate the behavior similarity between the first behavior information and the user behavior information of the user who has reached the preset incentive target;
[0043] Set a similarity threshold value, if the behavior similarity is greater than or equal to the similarity threshold value, mark the to-be-optimized user as a similar user, and push the adjusted score rule to the similar user.
[0044] Further, if it is judged that the preset incentive target is not reached, the potential matching point is mined again, and the score rule of the user is adjusted again, including:
[0045] Set an upper limit value of the adjustment times, if the total adjustment times of the to-be-optimized user exceed the upper limit value of the adjustment times, the adjustment is not performed again.
[0046] Compared with the prior art, the present application has the beneficial effects that: the present application constructs an image label system by collecting multi-dimensional information of users, breaks through the traditional single-dimensional user recognition mode, can comprehensively capture the attribute characteristics, behavior habits and preference tendencies of users on the platform, and provides a solid data foundation for precise operation. This deep insight makes users no longer be classified in general, but have personalized label images, which helps the platform to develop incentive strategies that meet the needs of users, and improves the users' sense of identity and participation in the point system. Secondly, the knowledge graph is used to integrate user portrait data and build relationship links, realizing the intelligent association of users, labels and point rules. The powerful semantic analysis and reasoning ability of the knowledge graph can mine the potential logical relationship between data, so that the formulation and adjustment of point rules are no longer dependent on subjective experience, but based on objective data analysis results, which not only improves the scientificity of rule formulation, but also quickly locates the adaptive rules in massive data, significantly improving the operation efficiency. Thirdly, the mechanism of dynamically optimizing rules based on point increase and decrease frequency gives the point system self-adaptive ability. When abnormal point increase and decrease are monitored, the rule optimization process is triggered actively, potential matching points are mined through the knowledge graph, and the rules are adjusted to ensure that the point rules always fit the user behavior changes and avoid user loss due to rule lag. At the same time, the changed points after adjustment are continuously monitored and the optimization effect is evaluated by the incentive effect index, forming a closed-loop system of "monitoring-adjusting-evaluating-reoptimizing" to realize the continuous iteration and upgrading of the point system. In addition, the verified rules are pushed to similar users to promote the rapid replication and promotion of successful experience, further amplify the incentive value of the point system, and enhance the platform user stickiness and market competitiveness. BRIEF DESCRIPTION OF DRAWINGS
[0047] Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are for purposes of illustration only and are not intended to limit the present application thereto. Moreover, the use of the same reference symbols in different drawings indicates similar or identical items.
[0048] Figure 1 The flowchart of the data processing and user incentive method based on the point system provided by the embodiments of the present application. DETAILED DESCRIPTION
[0049] Exemplary embodiments of the present disclosure will be described in detail below with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it is to be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the present disclosure to those skilled in the art. It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict. The present application will be described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.
[0050] In some embodiments of the present application, referring to Figure 1 The present embodiments provide a data processing and user incentive method based on an integral system, including the following steps:
[0051] S100, collecting multi-dimensional information of a user, and constructing a user portrait label system according to the multi-dimensional information of the user;
[0052] S200, integrating user portrait data in the user portrait label system through a knowledge graph, and setting a relationship link between the user, the label and the integral rule;
[0053] S300, collecting the integral increase and decrease frequency of the user, judging whether the integral rule of the user needs to be optimized according to the integral increase and decrease frequency, if it is judged that the integral rule needs to be optimized, analyzing the association between the labels through the knowledge graph, mining potential matching points between the user and the integral rule, and adjusting the integral rule of the user according to the potential matching points;
[0054] S400, continuously monitoring the user integral change after the integral rule is adjusted, obtaining an incentive effect index according to the user integral change, and judging whether a preset incentive target is reached according to the incentive effect index;
[0055] S500, if it is judged that the preset incentive target is reached, obtaining similar users from the knowledge graph, and pushing the adjusted integral rule to the similar users, if it is judged that the preset incentive target is not reached, mining the potential matching points again, and adjusting the integral rule of the user again.
[0056] It can be understood that the present application constructs an image label system by collecting multi-dimensional information of users, breaks through the traditional single-dimensional user recognition mode, can comprehensively capture the attribute characteristics, behavior habits and preference tendency of users on the platform, and provides a solid data foundation for precise operation. This deep insight makes users no longer be classified in general, but have personalized label images, which helps the platform to develop incentive strategies that meet the needs of users, and improves the identification and participation of users to the point system. Secondly, the knowledge graph is used to integrate user portrait data and build relationship links, realizing the intelligent association of users, labels and point rules. The powerful semantic analysis and reasoning ability of the knowledge graph can mine the potential logical relationship between data, so that the formulation and adjustment of point rules are no longer dependent on subjective experience, but based on objective data analysis results. This not only improves the scientificity of rule formulation, but also quickly locates the adaptive rules in massive data, significantly improving the operation efficiency. Furthermore, the mechanism of dynamically optimizing rules based on point increase and decrease frequency gives the point system self-adaptive ability. When abnormal point increase and decrease are monitored, the rule optimization process is triggered, potential matching points are mined through the knowledge graph, and the rules are adjusted to ensure that the point rules always adapt to user behavior changes and avoid user loss due to rule lag. At the same time, the changed points after adjustment are continuously monitored and the optimization effect is evaluated by the incentive effect index, forming a closed-loop system of "monitoring-adjusting-evaluating-reoptimizing" to realize the continuous iteration and upgrading of the point system. In addition, the verified rules are pushed to similar users to promote the rapid replication and promotion of successful experience, further amplify the incentive value of the point system, and enhance the platform user stickiness and market competitiveness.
[0057] In some embodiments of the present application, when collecting multi-dimensional information of users and constructing a user portrait label system according to the multi-dimensional information of users, it includes:
[0058] With the help of user registration forms and real-name authentication interfaces, user types, industry fields and registered capital are collected to build a basic portrait framework of users and collect user behavior information; wherein the user behavior information includes transaction amount, transaction frequency, sharing times, browsing time and usage proportion;
[0059] The labels in the user portrait label system are divided into four types, including value stratification labels, behavior preference labels and system tendency labels;
[0060] The value stratification label is used to divide users into different levels according to the transaction amount and transaction frequency of users on the platform;
[0061] The behavior preference label is used to collect the behavior habits of users on the platform, and divide users into different levels according to the behavior habits; wherein the behavior habits include sharing times and browsing time;
[0062] The system tendency label is used for recording the proportion of the use of different subsystems of the platform by the user.
[0063] It can be understood that in the user portrait label system construction process, collecting multi-dimensional information and systematic classification has significant advantages. Collecting basic attributes through the registration form and real-name authentication interface can quickly build a user basic portrait framework, clearly identify the identity characteristics and industry attributes of the user, and provide bottom support for subsequent fine operation; and the collection of multi-type behavior information such as transaction amount, frequency, and sharing behavior can deeply mine the behavior patterns and preference tendencies of the user, so that the portrait extends from static attributes to dynamic behavior analysis. The labels are divided into value stratification, behavior preference, and system tendency, etc. The structured sorting of user characteristics is realized, which can identify high-value users through value stratification labels, provide targeted exclusive incentives, capture user active habits through behavior preference labels, optimize the points acquisition path, and at the same time, master the user function use preference through the system tendency label, and reasonably allocate points resources. This multi-dimensional and type-specific portrait construction method can comprehensively and accurately depict user characteristics, provide detailed data basis for dynamic optimization and precise incentive of points rules, and effectively improve the pertinence and effectiveness of user incentives.
[0064] For example, taking a comprehensive enterprise service platform as an example, the platform integrates financing leasing, centralized procurement, after-sales maintenance and other subsystems. When implementing user portrait construction, the registration form is used to obtain basic information such as user type (enterprise / personal), industry field (manufacturing / service industry, etc.) and registered capital, and the user attributes are quickly distinguished. At the same time, the platform records the user behavior in real time: the manufacturing enterprise user A completes 3 large equipment leasing transactions in the financing leasing system within a month, with a cumulative amount of 5 million yuan, and shares procurement experience 2-3 times in the platform every week, and 80% of the operation time is concentrated in the centralized procurement system. Based on these data, the platform gives user A labels such as “high net worth user” (value stratification label), “high frequency sharer” (behavior preference label) and “centralized procurement system preference” (system tendency label). According to these labels, the platform adjusts the points rules for user A, such as giving double points in financing leasing transactions, giving exclusive points for sharing behavior, and opening the right to exchange rare resources in priority in the centralized procurement system.
[0065] In some embodiments of the present application, when the user portrait data in the user portrait label system is integrated through the knowledge graph, and the relationship link between the user, the label and the points rule is set, it includes:
[0066] The user base portrait framework, various types of labels, and point rules are converted into entities in the knowledge graph, and each entity is given a unique identifier, wherein the user entity includes a user ID and a registration time, the label entity includes a label ID, a label type, and a label description, and the point rule entity includes a rule ID and a rule content;
[0067] The relationship links between entities are constructed to clearly define the label combination possessed by each user and the point rule of the current user.
[0068] It can be understood that by integrating user portrait data and constructing relationship links through the knowledge graph, scattered user information, labels, and point rules can be converted into a structured semantic network, significantly improving data correlation efficiency and rule matching accuracy. Defining the user base portrait, labels, and point rules as independent entities and giving them unique identifiers enables standardized storage and management of data, and constructing relationship links between entities breaks through the logical path of "user characteristics-label combination-point rule", enabling real-time label-based dynamic retrieval of adaptive rules and avoiding the lag and subjectivity of traditional rule matching. This structured integration not only supports efficient graph query operations (such as quickly locating user groups with specific label combinations and their corresponding rules), but also can mine potential label-rule associations through the reasoning ability of the knowledge graph, providing semantic-level support for the intelligent optimization of point rules, and ultimately achieving the precision and automation of user incentive strategies.
[0069] Specifically, for example, taking a supply chain finance platform as an example, when user A (entity ID: U001, registration time 2024.01) is given the label combination of "high net worth user" (label ID: T005, type: value stratification, description: monthly transaction ≥ 1 million) and "factoring business preference" (label ID: T012, type: system inclination, description: factoring system points account for ≥ 70%), the knowledge graph automatically matches to the point rule with ID R023 (content: 15 points per 10,000 yuan of factoring business transaction amount) in the rule library through the predefined "user-owns-label" and "label-associated-rule" relationship links. At the same time, based on the association analysis of the knowledge graph, it is found that users with similar label combinations generally respond better to "ladder-type point reward" rules, so the R047 rule (content: point ratio increased to 20 / 10,000 yuan when single factoring transaction ≥ 500,000) is further associated, and the two rules are combined and pushed to user A.
[0070] In some embodiments of the present application, when the point increase / decrease frequency of the user is collected, and it is judged whether the point rule of the user needs to be optimized according to the point increase / decrease frequency, it includes:
[0071] A frequency increase / decrease minimum value is set, and if the integral increase / decrease frequency is less than the frequency increase / decrease minimum value, it is determined that the user's integral rules need to be optimized.
[0072] Otherwise, it is determined that the user's integral rules do not need to be optimized.
[0073] In some embodiments of the present application, if it is determined that optimization is needed, the potential matching points between the user and the integral rules are mined by analyzing the association between the labels in the knowledge graph.
[0074] The association between different labels is analyzed using the relationship links in the knowledge graph to obtain corresponding users of the labels with association;
[0075] User behavior information of the corresponding users is obtained, and the user behavior information of the corresponding users is calculated for similarity with the user behavior information of the user to be optimized;
[0076] A similarity threshold is set, and the integral rules corresponding to the users with a similarity greater than or equal to the similarity threshold are taken as potential matching points.
[0077] In some embodiments of the present application, when the integral rules of the user are adjusted for the potential matching points, it includes:
[0078] The integral rules of the corresponding users are added to the integral rules of the user to be optimized;
[0079] The original integral rules that conflict with the newly added integral rules in the integral rules to be optimized are deleted.
[0080] It can be understood that by setting the integral increase / decrease frequency minimum value to trigger the rule optimization mechanism, the low response state of the user to the existing integral system can be captured in time, and the loss of user activity due to rule lag can be avoided. By analyzing the label association in the knowledge graph and calculating the user behavior similarity, the user group with similar characteristics can be accurately positioned, the effective integral rules verified by practice are taken as potential matching points, and the pertinence and effectiveness of rule adjustment are significantly improved. When the rules are adjusted, the new rules and conflict rule deletion operations are executed simultaneously, which ensures the logical consistency of the integral system and avoids rule redundancy or contradiction affecting the user experience.
[0081] Specifically, for example, taking a cross-border e-commerce platform as an example, the platform monitors that the integral increase and decrease frequency of user A in the past 30 days is 2.5 times / week, which is lower than the preset minimum value of 5 times / week, triggering the rule optimization process. The knowledge graph analyzes the labels of user A such as "high-frequency overseas shopping user" and "luxury product preference", retrieves 100 similar users with a label similarity of ≥0.8, and finds that the integral rule of user B "luxury product category consumption of 5000 yuan returns 500 points" has a response rate of 60% in this group. Then add this rule to the integral rule library of user A, and detect that the original rule of user A "ordinary goods over 1000 yuan return 100 points" exists category coverage conflict with the new rule, and automatically delete the old rule.
[0082] In some embodiments of the present application, when the user integral change value is greater than twice the user historical integral change value, the incentive effect index is a first index.
[0083] Obtain the user integral change value within a preset time after the integral rule is adjusted and the user historical integral change value within the historical interval time before the integral rule is adjusted;
[0084] If the user integral change value is greater than twice the user historical integral change value, the incentive effect index is a first index.
[0085] If the user integral change value is less than or equal to twice the user integral change value and greater than one times the user historical integral change value, the incentive effect index is a second index.
[0086] If the user integral change value is less than or equal to one times the user historical integral change value, the incentive effect index is a third index.
[0087] Wherein, the first index> the second index>1>the third index>0.
[0088] In some embodiments of the present application, when the user integral change value is greater than twice the user historical integral change value, the incentive effect index is a first index.
[0089] A preset incentive target index, if the incentive effect index is greater than or equal to the incentive target index, it is judged that the preset incentive target is reached;
[0090] If the incentive effect index is less than the incentive target index, it is judged that the preset incentive target is not reached.
[0091] It can be understood that by setting the explicit incentive effect index calculation rule and the preset incentive target index, the effect of the adjusted integral rule can be quantitatively evaluated, and an objective and measurable standard is provided for the optimization of the integral system. The method compares the user integral change value before and after the rule adjustment, divides the incentive effect index into different levels in a multiple relationship, and intuitively reflects the influence degree of the rule adjustment on the user behavior. At the same time, the preset incentive target index is used as a baseline to measure whether the optimization is successful or not, which can clearly judge whether the adjusted integral rule reaches the expected target, and then decide whether to further optimize or promote successful experience, effectively avoid the uncertainty brought by subjective judgment, and ensure that the integral system always iterates towards the direction of improving user activity and incentive effect.
[0092] Specifically, taking a membership shopping platform as an example, the integral rule is adjusted for a high-value user group, and the original "1 yuan of consumption accumulates 1 point" is adjusted to "1 yuan of consumption accumulates 1.5 points, and an additional 50% of points is awarded for weekend consumption". Within 30 days after the rule adjustment, it is monitored that the integral change value of user A is 1500 points, and the historical integral change value in the same period before the adjustment is 600 points. After calculation, the integral change value (1500 points) of user A is more than twice the historical integral change value (600 points), so the incentive effect index of the user is rated as the first index. Since the platform preset incentive target index is 1.2, and the first index is significantly higher than the target value, it is determined that the integral rule adjustment reaches the preset incentive target, and this rule is pushed to other similar high-value users through the knowledge graph.
[0093] In some embodiments of the present application, if it is judged that the preset incentive target has been reached, the similar users are obtained from the knowledge graph, and the adjusted integral rule is pushed to the similar users, comprising:
[0094] Obtain a to-be-optimized user in the knowledge graph, and obtain user behavior information of the to-be-optimized user, denoted as first behavior information;
[0095] Calculate the behavior similarity between the first behavior information and the user behavior information of the user who has reached the preset incentive target;
[0096] Set a similarity threshold value, if the behavior similarity is greater than or equal to the similarity threshold value, mark the to-be-optimized user as a similar user, and push the adjusted integral rule to the similar user.
[0097] In some embodiments of the present application, if it is judged that the preset incentive target has not been reached, the potential matching points are further mined, and the integral rule of the user is adjusted again, comprising:
[0098] Set an upper limit value of the adjustment times, if the total adjustment times of the to-be-optimized user exceed the upper limit value of the adjustment times, the adjustment is not performed again.
[0099] It can be understood that, by acquiring similar users through the knowledge graph and pushing the optimized credit rules, the rapid reuse and large-scale promotion of successful incentive strategies can be realized, and setting an upper limit of the number of adjustments avoids the consumption of system resources by invalid optimization, forming a scientific rule iteration boundary control. Specifically, when a certain credit rule is verified to be effective in a specific user group, through the semantic retrieval ability of the knowledge graph and based on the similarity calculation of user behavior information, potential beneficiary users with the same behavior characteristics can be accurately located, avoiding resource waste caused by blind rule pushing; and the upper limit of the number of adjustments prevents excessive intervention of “difficult optimization users”, while ensuring user experience, improving the overall operation efficiency of the credit system.
[0100] Specifically, taking a local life service platform as an example, after the adjustment of the credit rule of user A (optimizing “10 points for consumption of 100 yuan in store” to “15 points for 100 yuan + 5 points for sharing stores”), the credit change value reaches 2000 points within 30 days, which is 3 times higher than before the adjustment, and the incentive effect index reaches the first index, exceeding the preset target index by 1.5. Through the knowledge graph, 100 users with a behavior similarity of ≥0.75 to user A (such as young groups who prefer to share and go to stores more than twice a week) are retrieved, and the optimized rule is pushed to these users. At the same time, for user B, the rule adjustment does not reach the incentive target for the first three times, and the upper limit mechanism (set to 3 times) is triggered automatically, stopping the automatic optimization of the rule, and switching to manual diagnosis combined with user portraits.
[0101] Those skilled in the art will appreciate that embodiments of the application can be provided as methods, systems or computer program products. Accordingly, the application can take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the application can take the form of a computer program product on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage devices, etc.) embodying computer-readable program code.
[0102] The application is described with reference to flowcharts and / or block diagrams according to the embodiments of the application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams can be implemented by computer program instructions, and the combination of flows and / or blocks in the flowcharts and / or block diagrams. These computer program instructions can be provided to a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing devices to produce a machine, so that the instructions executed by the computer or other programmable data processing devices produce a device that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 one flow or multiple flows and / or blocks Figure 1means for performing the function specified by the block or blocks.
[0103] These computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture including instructions which implement the flow Figure 1 one or more flows and / or blocks Figure 1 means for performing the function specified by the block or blocks.
[0104] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the flow Figure 1 one or more flows and / or blocks Figure 1 means for performing the function specified by the block or blocks.
[0105] Finally, it should be noted that the above-mentioned embodiments are merely intended to illustrate the technical solutions of the present application, but not to limit it. Although the present application has been described in detail with reference to the above-mentioned embodiments, those skilled in the art should understand that the specific embodiments of the present application can be modified or replaced by equivalents without departing from the spirit and scope of the present application, and any modifications or replacements not departing from the spirit and scope of the present application should be covered within the protection scope of the claims of the present application.
Claims
1. A data processing and user incentive method based on a points system, characterized in that: include: Collect multi-dimensional information of users and build a user portrait label system based on the multi-dimensional information of users; Integrate user portrait data in the user portrait tag system through the knowledge graph, and set the relationship link between users, tags and points rules; Collect the frequency of a user's points increase and decrease, and determine whether the user's points rules need to be optimized based on the frequency of points increase and decrease. If optimization is determined to be necessary, analyze the associations between tags through the knowledge graph to discover potential matching points between the user and the points rules, and adjust the user's points rules based on the potential matching points. Continuously monitor changes in user points after the adjustment of the points rules, obtain an incentive effect index based on the changes in user points, and determine whether the preset incentive target has been achieved based on the incentive effect index; If it is determined that the preset incentive target has been achieved, similar users are obtained from the knowledge graph and the adjusted points rules are pushed to the similar users. If it is determined that the preset incentive target has not been achieved, the potential matching points are mined again and the points rules for the user are adjusted again.
2. The data processing and user incentive method based on the points system according to claim 1, characterized in that: The collecting of multi-dimensional user information and constructing a user portrait label system based on the multi-dimensional user information includes: Leveraging user registration forms and real-name authentication interfaces, we collect user type, industry sector, and registered capital, build a basic user portrait framework, and collect user behavior information; this includes transaction amount, transaction frequency, number of shares, browsing time, and usage percentage. The labels in the user portrait label system are divided into four types, including value stratification labels, behavior preference labels, and system tendency labels; The value tiering labels are used to classify users into different levels based on their transaction amounts and transaction frequencies on the platform; The behavior preference tag is used to collect the user's behavior habits on the platform and classify users into different levels according to the behavior habits; wherein the behavior habits include the number of shares and the browsing time; The system preference tag is used to record the user usage ratio of different subsystems of the platform.
3. The data processing and user incentive method based on the points system according to claim 1, characterized in that: The process of integrating user portrait data in the user portrait tag system through the knowledge graph and setting the relationship link between users, tags, and points rules includes: Convert the user basic portrait framework, various types of tags, and points rules into entities in the knowledge graph, and assign a unique identifier to each entity. The user entity contains the user ID and registration time, the tag entity contains the tag ID, tag type, and tag description, and the points rule entity contains the rule ID and rule content. Build relationship links between entities, clarify the tag combinations of each user and the current user's points rules.
4. The data processing and user incentive method based on the points system according to claim 1, characterized in that: The collecting of the user's points increase and decrease frequency and determining whether the user's points rule needs to be optimized according to the points increase and decrease frequency include: Set a minimum frequency of increase and decrease. If the frequency of increase and decrease of points is less than the minimum frequency of increase and decrease, it is determined that the user's points rules need to be optimized. Otherwise, it is determined that there is no need to optimize the user's points rules.
5. The data processing and user incentive method based on the points system according to claim 4 is characterized in that: If it is determined that optimization is needed, the association between tags is analyzed through the knowledge graph to explore potential matching points between users and points rules, including: Utilize the relationship links in the knowledge graph to analyze the correlation between different tags and obtain the corresponding users of the related tags; Obtain user behavior information of the corresponding user, and calculate the similarity between the user behavior information of the corresponding user and the user behavior information of the user to be optimized; A similarity threshold is set, and the scoring rules corresponding to users whose similarity is greater than or equal to the similarity threshold are used as potential matching points.
6. The data processing and user incentive method based on the points system according to claim 5, characterized in that: The adjusting of the points rule for the user according to the potential matching point includes: Add the corresponding user's points rules to the points rules of the user that needs to be optimized; Delete the original integration rules that conflict with the newly added integration rules among the integration rules that need to be optimized.
7. The data processing and user incentive method based on the points system according to claim 1, characterized in that: The continuously monitoring the change of user points after the adjustment of the points rule and obtaining the incentive effect index according to the change of user points includes: Obtain the user points change value within the preset time after the points rule adjustment and the user's historical points change value within the historical interval before the points rule adjustment; If the user's points change value is greater than twice the user's historical points change value, the incentive effect index is the first index; If the user points change value is less than or equal to twice the user points change value, and greater than one times the user historical points change value, the incentive effect index is the second index; If the user points change value is less than or equal to one times the user's historical points change value, the incentive effect index is the third index; Among them, the first index > the second index > 1 > the third index > 0.
8. The data processing and user incentive method based on the points system according to claim 7, characterized in that: The step of determining whether a preset incentive target has been achieved according to the incentive effect index includes: A preset incentive target index, if the incentive effect index is greater than or equal to the incentive target index, it is determined that the preset incentive target has been achieved; If the incentive effect index is less than the incentive target index, it is determined that the preset incentive target has not been achieved.
9. The data processing and user incentive method based on the points system according to claim 8, characterized in that: If it is determined that the preset incentive target has been achieved, similar users are obtained from the knowledge graph, and the adjusted points rules are pushed to the similar users, including: Obtain the user to be optimized in the knowledge graph, and obtain the user behavior information of the user to be optimized, which is recorded as the first behavior information; Calculating the behavioral similarity between the first behavioral information and the user behavioral information of users who have achieved the preset incentive goal; A similarity threshold value is set. If the behavior similarity is greater than or equal to the similarity threshold value, the user to be optimized is marked as a similar user, and the adjusted points rule is pushed to the similar user.
10. The data processing and user incentive method based on the points system according to claim 9, characterized in that: If it is determined that the preset incentive target has not been achieved, the potential matching points are mined again and the points rule for the user is adjusted again, including: Set an upper limit for the number of adjustments. If the total number of adjustments for the user to be optimized exceeds the upper limit, no further adjustments will be made.