Member portrait construction and application method and system based on multi-source data

Abstract

The application provides a member portrait construction and application method and system based on multi-source data. The method comprises the following steps: collecting member static attribute data and real-time dynamic behavior data at a preset frequency, obtaining core feature quantitative data after processing the RFM behavior average data of a preset period and combining the real-time dynamic behavior data; fusing the core feature quantitative data and the static attribute data, generating a member label system through a hybrid algorithm model of statistical analysis and machine learning, and constructing a structured member consumption portrait based on the label system; continuously collecting member real-time behavior data, updating the consumption portrait at a preset period, combining a preset operation strategy rule to recommend personalized marketing data for members; and thus through standardized processing and feature quantization of multi-source data, the constructed member portrait has dynamicity, predictability and business adaptability, can accurately match member features to push personalized marketing, and greatly improves member operation efficiency and marketing accuracy.

Description

Technical Field

[0001] This application relates to the field of member profiling technology, and more specifically, to methods and systems for constructing and applying member profiles based on multi-source data. Background Technology

[0002] For offline entertainment businesses like arcades, refined membership management is crucial for improving operational efficiency and revenue growth. Member profiles, as the core carrier of membership management, directly determine the effectiveness of membership operations based on their completeness, dynamism, and practicality. Currently, offline entertainment businesses like arcades still suffer from numerous technical deficiencies in member profile construction and application, failing to meet the demands of intelligent and refined operations. Specific problems include: First, data collection dimensions are limited and lack standardization. Existing technologies primarily collect basic static attribute data such as member numbers and contact information, as well as simple transaction data like recharge and consumption. They fail to cover core dynamic behavioral data unique to arcades, such as game device preferences and marketing responses. Furthermore, the lack of differentiated data collection frequency leads to repetitive static data collection and delayed dynamic data collection. The raw data also lacks unified statistical indicators, hindering subsequent quantitative analysis. Second, core feature processing lacks scientific methods. Existing technologies have not established a reasonable feature quantification system, fail to compare and analyze member behavior data with benchmark data to uncover potential features such as churn risk, and fail to address different dimensions of behavioral data. Standardization processes prevent raw data from being directly input into algorithm models, making it difficult to uncover underlying patterns in member behavior. Third, member profiling lacks hierarchy and predictability. Existing technologies often generate single descriptive tags through simple statistics, failing to integrate machine learning algorithms to build a tag system with trend prediction capabilities. Profiling only reflects past objective behavior and cannot predict potential characteristics such as member churn risk or device preferences. Furthermore, the tag system is not integrated into a structured consumer profile based on operational dimensions, resulting in poor readability and adaptability. Fourth, profiling applications are disconnected from operational actions. Existing technologies often create static data sets that cannot be dynamically updated based on real-time member behavior. Moreover, there is no intelligent matching mechanism between profiling and marketing strategies; instead, a generalized marketing push approach is used, leading to low marketing accuracy, poor member experience, and the inability to effectively realize the commercial value of profiling.

[0003] In response to the shortcomings of the existing technologies, there is an urgent need for a method and system for building and applying member profiles that can achieve differentiated collection of multi-source data, scientifically quantify core features, construct hierarchical and predictable member consumption profiles, and deeply integrate profiles with personalized marketing. This would solve the problems of non-standard data collection, lack of feature processing methods, lack of predictability in profiles, and disconnect between application and operation in the existing technologies, thereby improving the level of refined member operation in offline business formats. Summary of the Invention

[0004] The purpose of this application is to provide a method and system for constructing and applying member profiles based on multi-source data. Through standardized processing and feature quantification of multi-source data, the constructed member profiles are dynamic, predictive, and business-adaptable, and can accurately match member characteristics to push personalized marketing, thereby significantly improving member operation efficiency and marketing accuracy.

[0005] This application also provides a method for constructing and applying member profiles based on multi-source data, including the following steps: Static attribute data and real-time dynamic behavior data are collected according to a preset collection frequency. Obtain average RFM behavior data and combine it with real-time dynamic behavior data to obtain core feature quantification data; Based on the core feature quantitative data and static attribute data, a member tag system is obtained and a member consumption profile is generated through a preset hybrid algorithm model. Acquire member behavior data, update member consumption profiles based on the member behavior data, and recommend personalized marketing data based on the member consumption profiles and preset operational strategy rules.

[0006] Optionally, in the method for constructing and applying member profiles based on multi-source data described in this application, the step of collecting static attribute data and real-time dynamic behavior data according to a preset collection frequency specifically includes: The preset acquisition frequency includes either acquisition upon update or acquisition at a preset period. Static attribute data, including member number, gender, birthday, and contact information, is obtained by updating the collection frequency. Real-time dynamic data is obtained through pre-set periodic collection, including RFM behavior data, gaming device preference data, and marketing response data.

[0007] Optionally, in the method for constructing and applying member profiles based on multi-source data described in this application, the RFM behavioral data, gaming device preference data, and marketing response data specifically include: RFM behavioral data includes the number of days between purchases, the number of visits to the store within a preset first time period, and the total amount spent in the store. Gaming device preference data includes preferred device type data and coin insertion data within a preset time period; Marketing response data includes the types of offers users received during the first preset time period, as well as the corresponding offer redemption conversion rate and consumption increase rate.

[0008] Optionally, in the method for constructing and applying member profiles based on multi-source data described in this application, the step of obtaining RFM behavior average data and processing it in combination with real-time dynamic behavior data to obtain core feature quantification data specifically includes: Obtain average RFM behavior data within a preset second time period, including average consumption interval in days, average number of visits to the store, and average total spending per visit; Churn risk characteristic data were obtained by processing RFM behavioral data and RFM behavioral average data, including the rate of change of consumption interval days, the rate of change of number of visits, and the rate of change of total in-store consumption. Based on the preferred device type data and coin insertion data, the coin insertion ratio data corresponding to each preferred device type is calculated. Based on the coin insertion ratio data, the device entropy value corresponding to the preferred device is calculated using the entropy method. The marketing sensitivity data for this type of offer is calculated based on the discount redemption conversion rate and the consumption increase rate. After normalizing the churn risk characteristic data, device entropy value, and marketing sensitivity data, we obtain the core characteristic quantitative data.

[0009] Optionally, in the method for constructing and applying member profiles based on multi-source data described in this application, the step of obtaining a member tag system and generating a member consumption profile by processing core feature quantified data and static attribute data through a preset hybrid algorithm model specifically includes: By using a hybrid algorithm model that combines pre-set statistical analysis and machine learning, and by processing core feature quantified data and static attribute data, a member tag system is obtained, including basic fact tags, model prediction tags, and business decision tags. The model prediction tags include a member value stratification model, a churn risk prediction model, a device preference prediction model, and a marketing sensitivity prediction model.

[0010] Optionally, in the method for constructing and applying member profiles based on multi-source data described in this application, the member value stratification model, churn risk prediction model, device preference prediction model, and marketing sensitivity prediction model specifically include: The membership value tiering model uses the K-Means clustering algorithm to cluster RFM behavioral data to obtain membership value levels. The churn risk prediction model uses a logistic regression model, takes churn risk characteristic data as input, and outputs a churn risk score. The device preference prediction model uses a collaborative filtering algorithm to obtain the co-occurrence patterns of device play among members and generate a personalized list of recommended gaming devices. The marketing sensitivity prediction model uses the random forest algorithm to predict the probability of members' responses to different marketing methods and generate marketing sensitivity labels.

[0011] Optionally, in the method for constructing and applying member profiles based on multi-source data described in this application, the steps of obtaining member behavior data, updating member consumption profiles based on member behavior data, and recommending personalized marketing data based on member consumption profiles and preset operational strategy rules specifically include: Acquire member behavior data, including device interaction data, consumption data, and marketing interaction data; Update member consumption profiles according to a preset third time period based on device interaction data, consumption data, or marketing interaction data. Personalized marketing data is pushed to members' contact information based on their consumption profiles and pre-set operational strategies and rules.

[0012] Secondly, this application provides a member profile construction and application system based on multi-source data. The system includes a memory and a processor. The memory stores a program for a member profile construction and application method based on multi-source data. When the program for a member profile construction and application method based on multi-source data is executed by the processor, it implements the following steps: Static attribute data and real-time dynamic behavior data are collected according to a preset collection frequency. Obtain average RFM behavior data and combine it with real-time dynamic behavior data to obtain core feature quantification data; Based on the core feature quantitative data and static attribute data, a member tag system is obtained and a member consumption profile is generated through a preset hybrid algorithm model. Acquire member behavior data, update member consumption profiles based on the member behavior data, and recommend personalized marketing data based on the member consumption profiles and preset operational strategy rules.

[0013] Optionally, in the member profile construction and application system based on multi-source data described in this application, the step of collecting static attribute data and real-time dynamic behavior data according to a preset collection frequency specifically includes: The preset acquisition frequency includes either acquisition upon update or acquisition at a preset period. Static attribute data, including member number, gender, birthday, and contact information, is obtained by updating the collection frequency. Real-time dynamic data is obtained through pre-set periodic collection, including RFM behavior data, gaming device preference data, and marketing response data.

[0014] Optionally, in the member profile construction and application system based on multi-source data described in this application, the RFM behavior data, game device preference data, and marketing response data specifically include: RFM behavioral data includes the number of days between purchases, the number of visits to the store within a preset first time period, and the total amount spent in the store. Gaming device preference data includes preferred device type data and coin insertion data within a preset time period; Marketing response data includes the types of offers users received during the first preset time period, as well as the corresponding offer redemption conversion rate and consumption increase rate.

[0015] As can be seen from the above, the member profile construction and application method and system based on multi-source data provided in this application collects static attribute data and real-time dynamic behavior data of members at a preset frequency, obtains the average RFM behavior data of a preset period, and then processes it with the real-time dynamic behavior data to obtain core feature quantification data; integrates the core feature quantification data and static attribute data, generates a member tag system through a hybrid algorithm model of statistical analysis and machine learning, and constructs a structured member consumption profile based on the tag system; continuously collects real-time member behavior data and updates the consumption profile at a preset period, and recommends personalized marketing data to members in combination with preset operation strategy rules; thus, through the standardized processing and feature quantification of multi-source data, the constructed member profile has dynamism, predictability and business adaptability, can accurately match member characteristics to push personalized marketing, and significantly improve member operation efficiency and marketing accuracy.

[0016] Other features and advantages of this application will be set forth in the following description and will be apparent in part from the description or may be learned by practicing the application. The objectives and other advantages of this application may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A flowchart illustrating the method for constructing and applying member profiles based on multi-source data provided in this application embodiment; Figure 2 A flowchart illustrating the method for constructing and applying member profiles based on multi-source data, as provided in this application embodiment, for obtaining static attribute data and real-time dynamic behavior data; Figure 3 A schematic diagram of the interface for obtaining average RFM behavior data in the method for constructing and applying member profiles based on multi-source data provided in the embodiments of this application; Figure 4 A schematic diagram of the structure of a member profile construction and application system based on multi-source data provided in this application embodiment; Figure 5 This is a schematic diagram illustrating the logical flow of the method for constructing and applying member profiles based on multi-source data, as provided in the embodiments of this application. Detailed Implementation

[0019] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0020] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0021] Please refer to Figure 1 , Figure 1 This is a flowchart illustrating a method for constructing and applying member profiles based on multi-source data, as described in some embodiments of this application. This method is used in terminal devices such as computers and mobile phones. The method includes the following steps: S11. Collect static attribute data and real-time dynamic behavior data according to the preset collection frequency; S12. Obtain the average data of RFM behavior, and process it in combination with real-time dynamic behavior data to obtain the core feature quantification data. S13. Based on the core feature quantification data and static attribute data, a member tag system is obtained and a member consumption profile is generated through a preset hybrid algorithm model. S14. Obtain member behavior data, update member consumption profiles based on member behavior data, and recommend personalized marketing data based on member consumption profiles and preset operation strategy rules.

[0022] Understandably, the process involves first pre-setting differentiated collection frequencies based on data characteristics and frequency of change, collecting relatively fixed static attribute data and real-time dynamic behavior data, then obtaining pre-set long-term RFM behavior average data as an analysis benchmark, combining it with real-time dynamic behavior data, and obtaining standardized core feature quantification data through comparison and algorithm calculation. Subsequently, the core feature quantification data is integrated with static attribute data, and a multi-level member tag system is generated based on a pre-built hybrid algorithm model combining statistical analysis and machine learning. A structured member consumption profile is then constructed based on this tag system. Finally, real-time behavior data such as member device interaction, consumption, and marketing interaction in the arcade operation scenario is continuously acquired, and the member consumption profile is dynamically updated according to a pre-set cycle. The updated profile is then intelligently matched with the pre-set operation strategy rules in the system to accurately recommend personalized marketing data to members, realizing a complete technical closed loop from multi-source data collection to the realization of the commercial value of the profile.

[0023] Please refer to Figure 2 , Figure 2 This is a flowchart illustrating the method for constructing and applying member profiles based on multi-source data, as provided in this application, for obtaining static attribute data and real-time dynamic behavior data. According to an embodiment of the present invention, the step of collecting and obtaining static attribute data and real-time dynamic behavior data according to a preset collection frequency specifically includes: S21. The preset acquisition frequency includes either update acquisition or preset periodic acquisition. S22. Obtain static attribute data, including member number, gender, birthday, and contact information, by updating the collection frequency. S23. Obtain real-time dynamic data through a preset periodic collection method, including RFM behavior data, gaming device preference data, and marketing response data.

[0024] Understandably, the preset collection frequency is first divided into two categories: "update-based collection" and "preset-period collection" to achieve efficient collection of data with different characteristics. For static attribute data such as member numbers and contact information, which are relatively fixed and have extremely low update frequencies, the "update-based collection" method is used, collecting data only when it is first entered or modified, reducing collection costs. For real-time dynamic behavioral data reflecting core member behaviors, the "preset-period collection" method is used, specifically collecting three categories: RFM behavioral data, game device preference data, and marketing response data, covering all behavioral scenarios of member consumption transactions, device play, and marketing participation, ensuring the timeliness and business relevance of the data. In this embodiment, data collection can be achieved through multiple channels such as member CRM systems, POS / cash register systems, IoT devices, and mini-program platforms. After data collection, quality verification is performed, automatically cleaning the collected raw data, and using interpolation to fill in missing values. The 3σ principle is used to eliminate outliers and ensure data accuracy.

[0025] According to embodiments of the present invention, the RFM behavioral data, gaming device preference data, and marketing response data specifically include: RFM behavioral data includes the number of days between purchases, the number of visits to the store within a preset first time period, and the total amount spent in the store. Gaming device preference data includes preferred device type data and coin insertion data within a preset time period; Marketing response data includes the types of offers users received during the first preset time period, as well as the corresponding offer redemption conversion rate and consumption increase rate.

[0026] Understandably, RFM behavioral data focuses on the core dimensions of member consumption value, statistically analyzing the number of days between purchases, number of visits, and total in-store spending within a preset first time period, serving as the core basis for member value segmentation. Gaming device preference data aligns with the characteristics of the arcade industry, statistically analyzing member preferred device types and coin-operated data within a preset time period, directly reflecting member device play preferences. Marketing response data focuses on member marketing feedback behavior, statistically analyzing the types of offers within a preset first time period, as well as the corresponding offer redemption conversion rate and consumption increase rate, serving as a key indicator for measuring member marketing sensitivity. All indicators support flexible adjustment of the statistical period according to the arcade's operational needs. In this embodiment... In the RFM behavioral data, R represents the consumption interval in days, i.e., the number of days since the last visit; F represents the number of visits, i.e., the number of visits within a preset time period; M represents the total consumption amount within the preset first time period; in this embodiment, the preset first time period is 30 days; the discount redemption conversion rate refers to the ratio of the number of discounts actually redeemed by members after receiving a certain type of discount to the number of discounts reached, reflecting the member's "acceptance" of the discount; the consumption increase rate refers to the ratio of the consumption amount after a member redeems a certain type of discount to the average consumption amount before redemption, reflecting the member's "consumption driving ability" of the discount; discount types include discount coupons, bonus coins, and doubled points.

[0027] According to an embodiment of the present invention, the step of obtaining RFM behavior average data and processing it in combination with real-time dynamic behavior data to obtain core feature quantification data specifically includes: Obtain average RFM behavior data within a preset second time period, including average consumption interval in days, average number of visits to the store, and average total spending per visit; Churn risk characteristic data were obtained by processing RFM behavioral data and RFM behavioral average data, including the rate of change of consumption interval days, the rate of change of number of visits, and the rate of change of total in-store consumption. Based on the preferred device type data and coin insertion data, the coin insertion ratio data corresponding to each preferred device type is calculated. Based on the coin insertion ratio data, the device entropy value corresponding to the preferred device is calculated using the entropy method. The marketing sensitivity data for this type of offer is calculated based on the discount redemption conversion rate and the consumption increase rate. After normalizing the churn risk characteristic data, device entropy value, and marketing sensitivity data, we obtain the core characteristic quantitative data.

[0028] Understandably, the process begins by acquiring average RFM behavior data within a pre-defined second time period, including average purchase interval days, average number of visits, and average total in-store spending, as a benchmark for judging changes in member behavior. Real-time RFM behavior data is then compared to this benchmark to calculate the rate of change in purchase interval days, the rate of change in number of visits, and the rate of change in total in-store spending, forming churn risk characteristic data. Simultaneously, based on member preferred device types and coin-operated data, the coin-operated percentage for each device is calculated, and the entropy value reflecting the concentration of game preferences is calculated using the entropy method. Finally, based on the discount redemption conversion rate and the consumption increase rate, the corresponding... Marketing sensitivity data for discount types; finally, the churn risk characteristic data, device entropy value, and marketing sensitivity data are normalized to eliminate dimensional differences, mapping all data to the same numerical range, ultimately obtaining core feature quantification data that can be directly input into the algorithm model; in this embodiment, the total number of game device types participating in the statistics can be obtained based on the preferred device type data; the second time period is preset to 90 days; the purpose of normalization is to eliminate the influence of dimensional differences. In this embodiment, the min-max normalization formula is used for normalization calculation. The normalized data is in the [0,1] interval, and the normalization calculation formula is: ; in, Here, X represents the normalized value, and X represents the current member's data to be normalized. The minimum value among members. This is the maximum value among the members; The formula for calculating the entropy of a device is: ;in, The device entropy value calculated for member i for game device type j, k is a fixed coefficient 1 / ln, and n is the total number of game device types participating in the statistics. The percentage of coins inserted by member i on the j-type gaming device; The formula for calculating the sensitivity coefficient is: ;in, For marketing sensitivity coefficient, To improve the conversion rate of discount redemption, The consumption increase rate; coin data refers to the number of coins a member uses to consume on the preferred device type within a preset time period. The coin percentage data is calculated by dividing the coin data corresponding to the preferred device by the total number of coins a member uses on all preferred devices within the preset time period.

[0029] According to an embodiment of the present invention, the step of obtaining a member tag system and generating a member consumption profile by processing core feature quantification data and static attribute data through a preset hybrid algorithm model specifically includes: By using a hybrid algorithm model that combines pre-set statistical analysis and machine learning, and by processing core feature quantified data and static attribute data, a member tag system is obtained, including basic fact tags, model prediction tags, and business decision tags. The model prediction tags include a member value stratification model, a churn risk prediction model, a device preference prediction model, and a marketing sensitivity prediction model.

[0030] Understandably, the system first integrates core feature quantification data and static attribute data. Through a pre-defined statistical analysis + machine learning hybrid algorithm model, a multi-layered member tag system is generated. Statistical analysis is used to uncover objective behavioral facts, while machine learning is used to achieve trend prediction; the two complement each other to enhance the practicality of the tag system. This tag system is clearly divided into three layers: basic fact tags, model prediction tags, and business decision tags. From factual description to trend prediction to business decision-making, each layer progresses to meet operational needs. Simultaneously, the model prediction tags clearly include four core models: a member value stratification model, a churn risk prediction model, a device preference prediction model, and a marketing sensitivity prediction model, corresponding to four core operational scenarios: member value management, retention management, experience management, and marketing management, respectively. Finally, using this multi-layered tag system as the core component, the tags are integrated according to member consumption analysis dimensions to generate a structured and interpretable member consumption profile. The basic fact tags, based on statistical analysis algorithms, are directly generated by the system after objectively processing member static attribute data, core feature quantification data, and raw real-time dynamic behavioral data; they have no predictive power. The system first extracts and organizes static attribute data such as member number and gender to generate basic factual tags for identity. Then, according to preset statistical periods and business thresholds, it performs statistical calculations on data such as member RFM behavior, gaming device preferences, and marketing responses, including frequency, proportion, and extreme values. The results are matched with preset business judgment rules, and those that meet the rules are tagged and refined. For example, spending over 5,000 yuan in the past 90 days generates a "high-spending member," and spending over 60% of the lottery machine's coin-operated portion generates a "lottery machine enthusiast." Finally, all tags are bound to the member's unique identifier, forming the basic layer of the member tag system. Business decision tags are generated by combining the arcade's operational rules and business objectives. Based on basic factual tags and model-predicted tags, the rules engine completes multi-tag association matching and condition judgment, and these tags are the core connecting profiles and operational actions. The system first integrates basic fact tags and model prediction tags to form a comprehensive feature tag pool for members. Then, it calls a preset operation rule engine to intelligently match and logically verify the tags in the tag pool with the engine's "multi-tag combination + numerical threshold" judgment rules. Targeted decision tags are extracted for members who meet specific operation rules, and the final generated tags are directly connected to the operation execution end, providing a clear decision basis for automated operation actions. In this embodiment, the three layers of member tags—basic facts, model prediction, and business decision—are first classified and organized according to core dimensions of the consumer profile, such as identity attributes, consumption behavior, device preferences, value level, risk trends, and marketing suitability. Then, differentiated weights are assigned to each dimension tag in conjunction with the arcade's operation goals, and quantitative integration is performed to form comprehensive feature values ​​for each dimension. Finally, all dimension features are integrated and bound to the member's unique identifier to generate a structured member consumption profile that can be directly connected to operations.

[0031] According to embodiments of the present invention, the member value stratification model, churn risk prediction model, device preference prediction model, and marketing sensitivity prediction model specifically include: The membership value tiering model uses the K-Means clustering algorithm to cluster RFM behavioral data to obtain membership value levels. The churn risk prediction model uses a logistic regression model, takes churn risk characteristic data as input, and outputs a churn risk score. The device preference prediction model uses a collaborative filtering algorithm to obtain the co-occurrence patterns of device play among members and generate a personalized list of recommended gaming devices. The marketing sensitivity prediction model uses the random forest algorithm to predict the probability of members' responses to different marketing methods and generate marketing sensitivity labels.

[0032] Understandably, the member value stratification model uses the K-Means clustering algorithm, taking member RFM behavior data as input, and divides members into different value levels through unsupervised clustering, outputting member value level labels; the churn risk prediction model uses the logistic regression model, taking churn risk feature data as input, and outputs a churn risk score between 0 and 1 through regression analysis, quantifying the probability of member churn; the device preference prediction model uses the collaborative filtering algorithm, taking member gaming device preference data as input, mining the device co-occurrence patterns of member groups, and outputting a personalized gaming device recommendation list; the marketing sensitivity prediction model uses the random forest algorithm, taking member marketing sensitivity data as input, predicting the probability of member response to different marketing methods, and outputting targeted marketing sensitivity labels. The output results of the four models directly serve subsequent operational decisions; co-occurrence patterns refer to the association patterns between different gaming devices / game types, where the same member plays them simultaneously / frequently. Simply put, "members who like to play on device A are likely to also like to play on device B," and A and B form a set of device co-occurrence patterns.

[0033] According to an embodiment of the present invention, the steps of acquiring member behavior data, updating member consumption profiles based on member behavior data, and recommending personalized marketing data based on member consumption profiles and preset operational strategy rules specifically include: Acquire member behavior data, including device interaction data, consumption data, and marketing interaction data; Update member consumption profiles according to a preset third time period based on device interaction data, consumption data, or marketing interaction data. Personalized marketing data is pushed to members' contact information based on their consumption profiles and pre-set operational strategies and rules.

[0034] Understandably, the core trigger data for member device interaction data, consumption data, and marketing interaction data are clearly defined as the core data for updating the member profile. This covers all core real-time behaviors of members in the arcade, ensuring the comprehensiveness of the profile update. Device interaction data refers to the data generated by members interacting with devices; consumption data includes recharge amount and consumption amount; and marketing interaction includes the number of discount redemptions and the number of discount notifications delivered. Based on the above behavioral data, the member consumption profile is dynamically updated according to a preset third time period. This period adopts a layered mechanism to adapt to the real-time requirements of different data, ensuring that the profile can reflect changes in member behavior in real time. Among these, it is set that... The system updates interaction data and consumption data at the second / minute level, and marketing interaction data at the hour level. Finally, based on the dynamically updated member consumption profile and combined with preset operation strategy rules, personalized marketing data is accurately pushed to the contact information provided by members. Unlike traditional generalized marketing, this achieves precise matching between marketing actions and member characteristics. In this embodiment, the system pre-builds an operation strategy rule engine and enters preset IF-THEN type operation strategy rules that match the arcade's operation goals, such as "IF member tag is racing machine enthusiast + this consumption > 100 yuan, THEN push 20% off coupon for simulation driving equipment".

[0035] It is worth mentioning that, after obtaining the device play co-occurrence patterns of the member group through the collaborative filtering algorithm, the device preference prediction model also includes: The co-occurrence mode is the associated combination after the game devices are grouped in pairs; Extract the total number of members for a single device and the number of members co-occurring on the device in the co-occurrence mode, and divide the number of members co-occurring on the device by the total number of members for a single device to obtain the device co-occurrence rate, including the first device co-occurrence rate and the second device co-occurrence rate; The combined co-occurrence rate is calculated based on the co-occurrence rate of the first device and the co-occurrence rate of the second device, combined with a preset weighting coefficient. The co-occurrence rate of each equipment combination is calculated one by one, and the normalized co-occurrence rate is obtained by normalizing the co-occurrence rate. The normalized co-occurrence rate is then compared with the preset co-occurrence mode level evaluation threshold. The preset co-occurrence mode level evaluation thresholds include a first threshold and a second threshold, with the first threshold being greater than the second threshold; If the co-occurrence rate of the normalized combination is greater than or equal to the first threshold, then the co-occurrence pattern level is the core co-occurrence combination; If the co-occurrence rate of the normalized combination is less than the first threshold and greater than or equal to the second threshold, then the co-occurrence pattern level is a potential co-occurrence combination. If the co-occurrence rate of the normalized combination is less than the second threshold, the co-occurrence pattern level is a weak co-occurrence combination; Based on the co-occurrence pattern level, a corresponding equipment combination marketing strategy is obtained.

[0036] It is understandable that targeted marketing strategies based on device co-occurrence are more conducive to improving success rates and revenue. Therefore, the existing game device types are grouped in pairs. For each group, the co-occurrence rate of the two devices in the group is calculated first, and then the combined co-occurrence rate is calculated by weighted summation. The combined co-occurrence rate can better reflect the co-occurrence correlation of the devices in the group. In this embodiment, the first threshold is set to 0.75 and the second threshold is set to 0.6. The device combination marketing strategy includes "for core co-occurrence combinations, give coins after playing one game on each of the two types of devices in the combination, and double the points for playing the combination" for core co-occurrence combinations; for potential co-occurrence combinations, generate device trial guidance strategies, such as giving a free trial coupon or discount coupon for the other type of device after playing the core device in the combination.

[0037] This invention also discloses a member profile construction and application system based on multi-source data, including a memory 41 and a processor 42. The memory stores a member profile construction and application method program based on multi-source data. When the processor executes the member profile construction and application method program based on multi-source data, it performs the following steps: Static attribute data and real-time dynamic behavior data are collected according to a preset collection frequency. Obtain average RFM behavior data and combine it with real-time dynamic behavior data to obtain core feature quantification data; Based on the core feature quantitative data and static attribute data, a member tag system is obtained and a member consumption profile is generated through a preset hybrid algorithm model. Acquire member behavior data, update member consumption profiles based on the member behavior data, and recommend personalized marketing data based on the member consumption profiles and preset operational strategy rules.

[0038] Understandably, the process involves first pre-setting differentiated collection frequencies based on data characteristics and frequency of change, collecting relatively fixed static attribute data and real-time dynamic behavior data, then obtaining pre-set long-term RFM behavior average data as an analysis benchmark, combining it with real-time dynamic behavior data, and obtaining standardized core feature quantification data through comparison and algorithm calculation. Subsequently, the core feature quantification data is integrated with static attribute data, and a multi-level member tag system is generated based on a pre-built hybrid algorithm model combining statistical analysis and machine learning. A structured member consumption profile is then constructed based on this tag system. Finally, real-time behavior data such as member device interaction, consumption, and marketing interaction in the arcade operation scenario is continuously acquired, and the member consumption profile is dynamically updated according to a pre-set cycle. The updated profile is then intelligently matched with the pre-set operation strategy rules in the system to accurately recommend personalized marketing data to members, realizing a complete technical closed loop from multi-source data collection to the realization of the commercial value of the profile.

[0039] According to an embodiment of the present invention, the step of acquiring static attribute data and real-time dynamic behavior data according to a preset acquisition frequency specifically includes: The preset acquisition frequency includes either acquisition upon update or acquisition at a preset period. Static attribute data, including member number, gender, birthday, and contact information, is obtained by updating the collection frequency. Real-time dynamic data is obtained through pre-set periodic collection, including RFM behavior data, gaming device preference data, and marketing response data.

[0040] Understandably, the preset collection frequency is first divided into two categories: "update-based collection" and "preset-period collection" to achieve efficient collection of data with different characteristics. For static attribute data such as member numbers and contact information, which are relatively fixed and have extremely low update frequencies, the "update-based collection" method is used, collecting data only when it is first entered or modified, reducing collection costs. For real-time dynamic behavioral data reflecting core member behaviors, the "preset-period collection" method is used, specifically collecting three categories: RFM behavioral data, game device preference data, and marketing response data, covering all behavioral scenarios of member consumption transactions, device play, and marketing participation, ensuring the timeliness and business relevance of the data. In this embodiment, data collection can be achieved through multiple channels such as member CRM systems, POS / cash register systems, IoT devices, and mini-program platforms. After data collection, quality verification is performed, automatically cleaning the collected raw data, and using interpolation to fill in missing values. The 3σ principle is used to eliminate outliers and ensure data accuracy.

[0041] According to embodiments of the present invention, the RFM behavioral data, gaming device preference data, and marketing response data specifically include: RFM behavioral data includes the number of days between purchases, the number of visits to the store within a preset first time period, and the total amount spent in the store. Gaming device preference data includes preferred device type data and coin insertion data within a preset time period; Marketing response data includes the types of offers users received during the first preset time period, as well as the corresponding offer redemption conversion rate and consumption increase rate.

[0042] Understandably, RFM behavioral data focuses on the core dimensions of member consumption value, statistically analyzing the number of days between purchases, number of visits, and total in-store spending within a preset first time period, serving as the core basis for member value segmentation. Gaming device preference data aligns with the characteristics of the arcade industry, statistically analyzing member preferred device types and coin-operated data within a preset time period, directly reflecting member device play preferences. Marketing response data focuses on member marketing feedback behavior, statistically analyzing the types of offers within a preset first time period, as well as the corresponding offer redemption conversion rate and consumption increase rate, serving as a key indicator for measuring member marketing sensitivity. All indicators support flexible adjustment of the statistical period according to the arcade's operational needs. In this embodiment... In the RFM behavioral data, R represents the consumption interval in days, i.e., the number of days since the last visit; F represents the number of visits, i.e., the number of visits within a preset time period; M represents the total consumption amount within the preset first time period; in this embodiment, the preset first time period is 30 days; the discount redemption conversion rate refers to the ratio of the number of discounts actually redeemed by members after receiving a certain type of discount to the number of discounts reached, reflecting the member's "acceptance" of the discount; the consumption increase rate refers to the ratio of the consumption amount after a member redeems a certain type of discount to the average consumption amount before redemption, reflecting the member's "consumption driving ability" of the discount; discount types include discount coupons, bonus coins, and doubled points.

[0043] According to an embodiment of the present invention, the step of obtaining RFM behavior average data and processing it in combination with real-time dynamic behavior data to obtain core feature quantification data specifically includes: Obtain average RFM behavior data within a preset second time period, including average consumption interval in days, average number of visits to the store, and average total spending per visit; Churn risk characteristic data were obtained by processing RFM behavioral data and RFM behavioral average data, including the rate of change of consumption interval days, the rate of change of number of visits, and the rate of change of total in-store consumption. Based on the preferred device type data and coin insertion data, the coin insertion ratio data corresponding to each preferred device type is calculated. Based on the coin insertion ratio data, the device entropy value corresponding to the preferred device is calculated using the entropy method. The marketing sensitivity data for this type of offer is calculated based on the discount redemption conversion rate and the consumption increase rate. After normalizing the churn risk characteristic data, device entropy value, and marketing sensitivity data, we obtain the core characteristic quantitative data.

[0044] Understandably, the process begins by acquiring average RFM behavior data within a pre-defined second time period, including average purchase interval days, average number of visits, and average total in-store spending, as a benchmark for judging changes in member behavior. Real-time RFM behavior data is then compared to this benchmark to calculate the rate of change in purchase interval days, the rate of change in number of visits, and the rate of change in total in-store spending, forming churn risk characteristic data. Simultaneously, based on member preferred device types and coin-operated data, the coin-operated percentage for each device is calculated, and the entropy value reflecting the concentration of game preferences is calculated using the entropy method. Finally, based on the discount redemption conversion rate and the consumption increase rate, the corresponding... Marketing sensitivity data for discount types; finally, the churn risk characteristic data, device entropy value, and marketing sensitivity data are normalized to eliminate dimensional differences, mapping all data to the same numerical range, ultimately obtaining core feature quantification data that can be directly input into the algorithm model; in this embodiment, the total number of game device types participating in the statistics can be obtained based on the preferred device type data; the second time period is preset to 90 days; the purpose of normalization is to eliminate the influence of dimensional differences. In this embodiment, the min-max normalization formula is used for normalization calculation. The normalized data is in the [0,1] interval, and the normalization calculation formula is: ; in, Here, X represents the normalized value, and X represents the current member's data to be normalized. The minimum value among members. This is the maximum value among the members; The formula for calculating the entropy of a device is: ;in, The device entropy value calculated for member i for game device type j, k is a fixed coefficient 1 / ln, and n is the total number of game device types participating in the statistics. The percentage of coins inserted by member i on the j-type gaming device; The formula for calculating the sensitivity coefficient is: ;in, For marketing sensitivity coefficient, To improve the conversion rate of discount redemption, The consumption increase rate; coin data refers to the number of coins a member uses to consume on the preferred device type within a preset time period. The coin percentage data is calculated by dividing the coin data corresponding to the preferred device by the total number of coins a member uses on all preferred devices within the preset time period.

[0045] According to an embodiment of the present invention, the step of obtaining a member tag system and generating a member consumption profile by processing core feature quantification data and static attribute data through a preset hybrid algorithm model specifically includes: By using a hybrid algorithm model that combines pre-set statistical analysis and machine learning, and by processing core feature quantified data and static attribute data, a member tag system is obtained, including basic fact tags, model prediction tags, and business decision tags. The model prediction tags include a member value stratification model, a churn risk prediction model, a device preference prediction model, and a marketing sensitivity prediction model.

[0046] Understandably, the system first integrates core feature quantification data and static attribute data. Through a pre-defined statistical analysis + machine learning hybrid algorithm model, a multi-layered member tag system is generated. Statistical analysis is used to uncover objective behavioral facts, while machine learning is used to achieve trend prediction; the two complement each other to enhance the practicality of the tag system. This tag system is clearly divided into three layers: basic fact tags, model prediction tags, and business decision tags. From factual description to trend prediction to business decision-making, each layer progresses to meet operational needs. Simultaneously, the model prediction tags clearly include four core models: a member value stratification model, a churn risk prediction model, a device preference prediction model, and a marketing sensitivity prediction model, corresponding to four core operational scenarios: member value management, retention management, experience management, and marketing management, respectively. Finally, using this multi-layered tag system as the core component, the tags are integrated according to member consumption analysis dimensions to generate a structured and interpretable member consumption profile. The basic fact tags, based on statistical analysis algorithms, are directly generated by the system after objectively processing member static attribute data, core feature quantification data, and raw real-time dynamic behavioral data; they have no predictive power. The system first extracts and organizes static attribute data such as member number and gender to generate basic factual tags for identity. Then, according to preset statistical periods and business thresholds, it performs statistical calculations on data such as member RFM behavior, gaming device preferences, and marketing responses, including frequency, proportion, and extreme values. The results are matched with preset business judgment rules, and those that meet the rules are tagged and refined. For example, spending over 5,000 yuan in the past 90 days generates a "high-spending member," and spending over 60% of the lottery machine's coin-operated portion generates a "lottery machine enthusiast." Finally, all tags are bound to the member's unique identifier, forming the basic layer of the member tag system. Business decision tags are generated by combining the arcade's operational rules and business objectives. Based on basic factual tags and model-predicted tags, the rules engine completes multi-tag association matching and condition judgment, and these tags are the core connecting profiles and operational actions. The system first integrates basic fact tags and model prediction tags to form a comprehensive feature tag pool for members. Then, it calls a preset operation rule engine to intelligently match and logically verify the tags in the tag pool with the engine's "multi-tag combination + numerical threshold" judgment rules. Targeted decision tags are extracted for members who meet specific operation rules, and the final generated tags are directly connected to the operation execution end, providing a clear decision basis for automated operation actions. In this embodiment, the three layers of member tags—basic facts, model prediction, and business decision—are first classified and organized according to core dimensions of the consumer profile, such as identity attributes, consumption behavior, device preferences, value level, risk trends, and marketing suitability. Then, differentiated weights are assigned to each dimension tag in conjunction with the arcade's operation goals, and quantitative integration is performed to form comprehensive feature values ​​for each dimension. Finally, all dimension features are integrated and bound to the member's unique identifier to generate a structured member consumption profile that can be directly connected to operations.

[0047] According to embodiments of the present invention, the member value stratification model, churn risk prediction model, device preference prediction model, and marketing sensitivity prediction model specifically include: The membership value tiering model uses the K-Means clustering algorithm to cluster RFM behavioral data to obtain membership value levels. The churn risk prediction model uses a logistic regression model, takes churn risk characteristic data as input, and outputs a churn risk score. The device preference prediction model uses a collaborative filtering algorithm to obtain the co-occurrence patterns of device play among members and generate a personalized list of recommended gaming devices. The marketing sensitivity prediction model uses the random forest algorithm to predict the probability of members' responses to different marketing methods and generate marketing sensitivity labels.

[0048] Understandably, the member value stratification model uses the K-Means clustering algorithm, taking member RFM behavior data as input, and divides members into different value levels through unsupervised clustering, outputting member value level labels; the churn risk prediction model uses the logistic regression model, taking churn risk feature data as input, and outputs a churn risk score between 0 and 1 through regression analysis, quantifying the probability of member churn; the device preference prediction model uses the collaborative filtering algorithm, taking member gaming device preference data as input, mining the device co-occurrence patterns of member groups, and outputting a personalized gaming device recommendation list; the marketing sensitivity prediction model uses the random forest algorithm, taking member marketing sensitivity data as input, predicting the probability of member response to different marketing methods, and outputting targeted marketing sensitivity labels. The output results of the four models directly serve subsequent operational decisions; co-occurrence patterns refer to the association patterns between different gaming devices / game types, where the same member plays them simultaneously / frequently. Simply put, "members who like to play on device A are likely to also like to play on device B," and A and B form a set of device co-occurrence patterns.

[0049] According to an embodiment of the present invention, the steps of acquiring member behavior data, updating member consumption profiles based on member behavior data, and recommending personalized marketing data based on member consumption profiles and preset operational strategy rules specifically include: Acquire member behavior data, including device interaction data, consumption data, and marketing interaction data; Update member consumption profiles according to a preset third time period based on device interaction data, consumption data, or marketing interaction data. Personalized marketing data is pushed to members' contact information based on their consumption profiles and pre-set operational strategies and rules.

[0050] Understandably, the core trigger data for member device interaction data, consumption data, and marketing interaction data are clearly defined as the core data for updating the member profile. This covers all core real-time behaviors of members in the arcade, ensuring the comprehensiveness of the profile update. Device interaction data refers to the data generated by members interacting with devices; consumption data includes recharge amount and consumption amount; and marketing interaction includes the number of discount redemptions and the number of discount notifications delivered. Based on the above behavioral data, the member consumption profile is dynamically updated according to a preset third time period. This period adopts a layered mechanism to adapt to the real-time requirements of different data, ensuring that the profile can reflect changes in member behavior in real time. Among these, it is set that... The system updates interaction data and consumption data at the second / minute level, and marketing interaction data at the hour level. Finally, based on the dynamically updated member consumption profile and combined with preset operation strategy rules, personalized marketing data is accurately pushed to the contact information provided by members. Unlike traditional generalized marketing, this achieves precise matching between marketing actions and member characteristics. In this embodiment, the system pre-builds an operation strategy rule engine and enters preset IF-THEN type operation strategy rules that match the arcade's operation goals, such as "IF member tag is racing machine enthusiast + this consumption > 100 yuan, THEN push 20% off coupon for simulation driving equipment".

[0051] It is worth mentioning that, after obtaining the device play co-occurrence patterns of the member group through the collaborative filtering algorithm, the device preference prediction model also includes: The co-occurrence mode is the associated combination after the game devices are grouped in pairs; Extract the total number of members for a single device and the number of members co-occurring on the device in the co-occurrence mode, and divide the number of members co-occurring on the device by the total number of members for a single device to obtain the device co-occurrence rate, including the first device co-occurrence rate and the second device co-occurrence rate; The combined co-occurrence rate is calculated based on the co-occurrence rate of the first device and the co-occurrence rate of the second device, combined with a preset weighting coefficient. The co-occurrence rate of each equipment combination is calculated one by one, and the normalized co-occurrence rate is obtained by normalizing the co-occurrence rate. The normalized co-occurrence rate is then compared with the preset co-occurrence mode level evaluation threshold. The preset co-occurrence mode level evaluation thresholds include a first threshold and a second threshold, with the first threshold being greater than the second threshold; If the co-occurrence rate of the normalized combination is greater than or equal to the first threshold, then the co-occurrence pattern level is the core co-occurrence combination; If the co-occurrence rate of the normalized combination is less than the first threshold and greater than or equal to the second threshold, then the co-occurrence pattern level is a potential co-occurrence combination. If the co-occurrence rate of the normalized combination is less than the second threshold, the co-occurrence pattern level is a weak co-occurrence combination; Based on the co-occurrence pattern level, a corresponding equipment combination marketing strategy is obtained.

[0052] It is understandable that targeted marketing strategies based on device co-occurrence are more conducive to improving success rates and revenue. Therefore, the existing game device types are grouped in pairs. For each group, the co-occurrence rate of the two devices in the group is calculated first, and then the combined co-occurrence rate is calculated by weighted summation. The combined co-occurrence rate can better reflect the co-occurrence correlation of the devices in the group. In this embodiment, the first threshold is set to 0.75 and the second threshold is set to 0.6. The device combination marketing strategy includes "for core co-occurrence combinations, give coins after playing one game on each of the two types of devices in the combination, and double the points for playing the combination" for core co-occurrence combinations; for potential co-occurrence combinations, generate device trial guidance strategies, such as giving a free trial coupon or discount coupon for the other type of device after playing the core device in the combination.

[0053] This invention discloses a method and system for constructing and applying member profiles based on multi-source data. It collects static attribute data and real-time dynamic behavior data of members at a preset frequency, obtains average RFM behavior data for a preset period, and then processes this data in conjunction with the real-time dynamic behavior data to obtain quantified core feature data. The quantified core feature data is then integrated with the static attribute data, and a member tag system is generated through a hybrid algorithm model combining statistical analysis and machine learning. A structured member consumption profile is then constructed based on this tag system. Real-time member behavior data is continuously collected and updated in a preset period to update the consumption profile. Personalized marketing data is then recommended to members based on preset operational strategy rules. Thus, through standardized processing and feature quantification of multi-source data, the constructed member profile possesses dynamism, predictability, and business adaptability, enabling accurate matching of member characteristics to push personalized marketing, significantly improving member operation efficiency and marketing accuracy.

[0054] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0055] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units. They may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of this embodiment according to actual needs.

[0056] In addition, in the various embodiments of the present invention, each functional unit can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0057] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0058] Alternatively, if the integrated units of this invention are implemented as software functional modules and sold or used as independent products, they can also be stored in a readable storage medium. Based on this understanding, the technical solutions of the embodiments of this invention, or the parts that contribute to the prior art, can be embodied in the form of a software product. This 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 methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROM, RAM, magnetic disks, or optical disks.

Claims

1. A method for constructing and applying member profiles based on multi-source data, characterized in that, include: Static attribute data and real-time dynamic behavior data are collected according to a preset collection frequency. Obtain average RFM behavior data and combine it with real-time dynamic behavior data to obtain core feature quantification data; Based on the core feature quantitative data and static attribute data, a member tag system is obtained and a member consumption profile is generated through a preset hybrid algorithm model. Acquire member behavior data, update member consumption profiles based on the member behavior data, and recommend personalized marketing data based on the member consumption profiles and preset operational strategy rules.

2. The method for constructing and applying member profiles based on multi-source data according to claim 1, characterized in that, The process of acquiring static attribute data and real-time dynamic behavior data according to a preset acquisition frequency specifically includes: The preset acquisition frequency includes either acquisition upon update or acquisition at a preset period. Static attribute data, including member number, gender, birthday, and contact information, is obtained by updating the collection frequency. Real-time dynamic data is obtained through pre-set periodic collection, including RFM behavior data, gaming device preference data, and marketing response data.

3. The method for constructing and applying member profiles based on multi-source data according to claim 2, characterized in that, The RFM behavioral data, gaming device preference data, and marketing response data specifically include: RFM behavioral data includes the number of days between purchases, the number of visits to the store within a preset first time period, and the total amount spent in the store. Gaming device preference data includes preferred device type data and coin insertion data within a preset time period; Marketing response data includes the types of offers users received during the first preset time period, as well as the corresponding offer redemption conversion rate and consumption increase rate.

4. The method for constructing and applying member profiles based on multi-source data according to claim 1, characterized in that, The process of obtaining average RFM behavior data and combining it with real-time dynamic behavior data to obtain core feature quantification data specifically includes: Obtain average RFM behavior data within a preset second time period, including average consumption interval in days, average number of visits to the store, and average total spending per visit; Churn risk characteristic data were obtained by processing RFM behavioral data and RFM behavioral average data, including the rate of change of consumption interval days, the rate of change of number of visits, and the rate of change of total in-store consumption. Based on the preferred device type data and coin insertion data, the coin insertion ratio data corresponding to each preferred device type is calculated. Based on the coin insertion ratio data, the device entropy value corresponding to the preferred device is calculated using the entropy method. The marketing sensitivity data for this type of offer is calculated based on the discount redemption conversion rate and the consumption increase rate. After normalizing the churn risk characteristic data, device entropy value, and marketing sensitivity data, we obtain the core characteristic quantitative data.

5. The method for constructing and applying member profiles based on multi-source data according to claim 4, characterized in that, The process of obtaining a member tag system and generating a member consumption profile by combining core feature quantified data with static attribute data through a preset hybrid algorithm model specifically includes: By using a hybrid algorithm model that combines pre-set statistical analysis and machine learning, and by processing core feature quantified data and static attribute data, a member tag system is obtained, including basic fact tags, model prediction tags, and business decision tags. The model prediction tags include a member value stratification model, a churn risk prediction model, a device preference prediction model, and a marketing sensitivity prediction model.

6. The method for constructing and applying member profiles based on multi-source data according to claim 5, characterized in that, The aforementioned member value stratification model, churn risk prediction model, device preference prediction model, and marketing sensitivity prediction model specifically include: The membership value tiering model uses the K-Means clustering algorithm to cluster RFM behavioral data to obtain membership value levels. The churn risk prediction model uses a logistic regression model, takes churn risk characteristic data as input, and outputs a churn risk score. The device preference prediction model uses a collaborative filtering algorithm to obtain the co-occurrence patterns of device play among members and generate a personalized list of recommended gaming devices. The marketing sensitivity prediction model uses the random forest algorithm to predict the probability of members' responses to different marketing methods and generate marketing sensitivity labels.

7. The method for constructing and applying member profiles based on multi-source data according to claim 6, characterized in that, The process of acquiring member behavior data, updating member consumption profiles based on the member behavior data, and recommending personalized marketing data based on the member consumption profiles and preset operational strategy rules specifically includes: Acquire member behavior data, including device interaction data, consumption data, and marketing interaction data; Update member consumption profiles according to a preset third time period based on device interaction data, consumption data, or marketing interaction data. Personalized marketing data is pushed to members' contact information based on their consumption profiles and pre-set operational strategies and rules.

8. A member profile construction and application system based on multi-source data, characterized in that, The system includes a memory and a processor. The memory contains a program for constructing and applying member profiles based on multi-source data. When the processor executes the program, the program implements the following steps: Static attribute data and real-time dynamic behavior data are collected according to a preset collection frequency. Obtain average RFM behavior data and combine it with real-time dynamic behavior data to obtain core feature quantification data; Based on the core feature quantitative data and static attribute data, a member tag system is obtained and a member consumption profile is generated through a preset hybrid algorithm model. Acquire member behavior data, update member consumption profiles based on the member behavior data, and recommend personalized marketing data based on the member consumption profiles and preset operational strategy rules.

9. The member profile construction and application system based on multi-source data according to claim 8, characterized in that, The process of acquiring static attribute data and real-time dynamic behavior data according to a preset acquisition frequency specifically includes: The preset acquisition frequency includes either acquisition upon update or acquisition at a preset period. Static attribute data, including member number, gender, birthday, and contact information, is obtained by updating the collection frequency. Real-time dynamic data is obtained through pre-set periodic collection, including RFM behavior data, gaming device preference data, and marketing response data.

10. The member profile construction and application system based on multi-source data according to claim 9, characterized in that, The RFM behavioral data, gaming device preference data, and marketing response data specifically include: RFM behavioral data includes the number of days between purchases, the number of visits to the store within a preset first time period, and the total amount spent in the store. Gaming device preference data includes preferred device type data and coin insertion data within a preset time period; Marketing response data includes the types of offers users received during the first preset time period, as well as the corresponding offer redemption conversion rate and consumption increase rate.

Images