Object upgrade willingness prediction and strategy generation method, device, equipment and medium

Abstract

The present application relates to the technical field of intelligent decision-making, and discloses an object upgrade willingness prediction and strategy generation method, device, equipment and medium, comprising: obtaining historical object sample data and generating an analysis model and a prediction model; associating object business data and authorized attribute behavior data of a target object to generate a target object data set; generating equity sensitivity, object value vector, object clustering result and object level gap based on the target object data set; generating upgrade willingness information through the prediction model to determine a target upgrade level and an object reach priority; generating a strategy result and sending it to a terminal device for presentation. The present application can be applied to financial technology and medical health business scenarios, and through object data fusion, value clustering, gap analysis and upgrade prediction, the strategy result has continuous data basis, and the accuracy of target upgrade level determination and strategy result output can be improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent decision-making technology, and in particular to a method, apparatus, device, and medium for predicting object upgrade intentions and generating strategies. Background Technology

[0002] As customer management and membership services shift from one-off transactions to continuous, tiered operations, the data relationships between customer levels, behaviors, rights usage, value performance, and authorized attribute behaviors are increasingly impacting the quality of subsequent service outreach. Existing technologies typically store and use data from different sources in a scattered manner, making it difficult to form a unified data foundation for customer-oriented upgrade judgments. This results in a lack of continuous analytical basis between the customer's current status, level achievement gap, and upgrade intentions, thus making upgrade planning and outreach content primarily reliant on human experience.

[0003] In the fintech business sector, scenarios such as life insurance, bank membership management, wealth management, and insurance services typically involve multiple types of data related to customers, including policy information, account service records, membership levels, rights usage records, service interaction records, and authorized consumption preferences. Existing marketing or customer management methods often focus on single transaction results, policy holding status, or manual customer segmentation, making it difficult to establish a unified basis for judging customer levels, rights sensitivity, current business status, and level achievement conditions. This results in a lack of stable data support to determine whether a customer is likely to upgrade, what gap exists between their current level and the target level, and how to reach them subsequently.

[0004] In the healthcare sector, data such as service records, health check-up services, chronic disease follow-up, and membership-based medical services are often used in a fragmented manner, including data on service records, health records, usage of benefits, follow-up behavior, status change events, and service preferences. Current service operation methods rely heavily on single service records or individual health information for assessment, making it difficult to link and analyze the current service level, health service behavior, level of concern for benefits, and target service level conditions. This results in a lack of quantitative basis for judging the user's willingness to upgrade service levels and for planning service pathways. Summary of the Invention

[0005] The main objective of this invention is to provide a method, apparatus, device, and storage medium for predicting object upgrade intentions and generating strategies. This invention aims to solve the technical problem that existing technologies struggle to integrate scattered object business data, authorized attribute behavior data, rights usage data, and level achievement conditions into a unified upgrade judgment basis, resulting in a lack of continuous data support for object upgrade intentions, level gaps, and outreach strategies.

[0006] To achieve the above objectives, the present invention provides a method for predicting object upgrade intentions and generating strategies, comprising: Acquire historical object sample data, and generate analysis and prediction models based on the historical object sample data; Obtain the object identifier, object business data, and authorized attribute behavior data of the target object, and generate a target object data set based on the object identifier, associating the object business data and the authorized attribute behavior data. Extract the current object level and object value data from the target object dataset, generate equity sensitivity and object value vectors based on the object value data, and process the object value vectors based on the analysis model to generate object clustering results; Based on the current object level, candidate upgrade levels and level achievement conditions are determined, and object level gaps are generated based on the target object data set and the level achievement conditions. Upgrade-related features are extracted from the target object dataset. Based on the prediction model, the current object level, the rights and interests sensitivity, the object level gap, and the upgrade-related features are processed to generate upgrade intention information. Based on the upgrade intention information, the target upgrade level and object reach priority are determined from the candidate upgrade levels. Based on the target object data set, the object grouping results, the rights and interests sensitivity, the current object level, the target upgrade level, and the object level gap, a strategy result is generated, and the object reach priority and the strategy result are sent to the terminal device for presentation.

[0007] Furthermore, to achieve the above objectives, the present invention provides an apparatus for predicting object upgrade intentions and generating strategies, comprising: The historical sample modeling module is used to acquire historical object sample data and generate analysis and prediction models based on the historical object sample data. The object data fusion module is used to obtain the object identifier, object business data, and authorized attribute behavior data of the target object, and generate a target object data set based on the object identifier, associating the object business data and the authorized attribute behavior data. The object value clustering module is used to extract the current object level and object value data from the target object data set, generate equity sensitivity and object value vectors based on the object value data, and process the object value vectors based on the analysis model to generate object clustering results; The upgrade gap analysis module is used to determine candidate upgrade levels and level achievement conditions based on the current object level, and to generate object level gaps based on the target object data set and the level achievement conditions. The upgrade intention prediction module is used to extract upgrade-related features from the target object data set, process the current object level, the rights and interests sensitivity, the object level gap and the upgrade-related features based on the prediction model, generate upgrade intention information, and determine the target upgrade level and object reach priority from the candidate upgrade levels based on the upgrade intention information; The strategy presentation module is used to generate strategy results based on the target object data set, the object grouping results, the rights and interests sensitivity, the current object level, the target upgrade level, and the object level gap, and send the object reach priority and the strategy results to the terminal device for presentation.

[0008] Furthermore, to achieve the above objectives, the present invention also provides a computer device, the computer device including a memory, a processor, and an object upgrade intention prediction and strategy generation program stored in the memory and executable on the processor, wherein when the object upgrade intention prediction and strategy generation program is executed by the processor, it implements the steps of the object upgrade intention prediction and strategy generation method as described above.

[0009] Furthermore, to achieve the above objectives, the present invention also provides a computer-readable storage medium storing an object upgrade intention prediction and strategy generation program, wherein when the object upgrade intention prediction and strategy generation program is executed by a processor, it implements the steps of the object upgrade intention prediction and strategy generation method described above.

[0010] Beneficial Effects: This invention relates to the field of intelligent decision-making technology, and discloses a method, apparatus, device, and medium for predicting object upgrade intentions and generating strategies. The method includes: acquiring historical object sample data and generating an analysis model and a prediction model; associating the target object's business data and authorized attribute behavior data to generate a target object data set; generating equity sensitivity, object value vector, object grouping results, and object level gap based on the target object data set; generating upgrade intention information through the prediction model to determine the target upgrade level and object reach priority; and generating strategy results and sending them to a terminal device for presentation. This invention can be applied to business scenarios such as fintech and healthcare. By integrating object business data, authorized attribute behavior data, equity sensitivity, object level gap, and upgrade-related features, it provides continuous data support for object grouping, upgrade prediction, and strategy generation. Since the object level gap reflects the difference between the current state and the conditions for achieving the level, and the upgrade intention information reflects the response probability of candidate upgrade levels, it can improve the accuracy of determining the target upgrade level and outputting strategy results. Attached Figure Description

[0011] The present invention will be further described below with reference to the accompanying drawings and embodiments. In the accompanying drawings: Figure 1 This is a schematic diagram of an application environment for the object upgrade intention prediction and strategy generation method in one embodiment of the present invention; Figure 2 This is a flowchart illustrating an embodiment of the object upgrade intention prediction and strategy generation method of the present invention. Figure 3 A schematic diagram of functional modules of a preferred embodiment of the object upgrade intention prediction and strategy generation device of the present invention; Figure 4 This is a schematic diagram of the structure of a computer device according to an embodiment of the present invention; Figure 5 This is another structural schematic diagram of a computer device according to one embodiment of the present invention. Detailed Implementation

[0012] It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the invention.

[0013] The object upgrade intention prediction and strategy generation method provided in this invention can be applied to, for example... Figure 1 In this application environment, the client communicates with the server via a network. The server can obtain historical object sample data from the client and generate analysis and prediction models; associate the target object's business data and authorized attribute behavior data to generate a target object data set; generate equity sensitivity, object value vector, object segmentation results, and object level gap based on the target object data set; generate upgrade intention information through the prediction model to determine the target upgrade level and object reach priority; generate strategy results and send them to the terminal device for presentation. This invention can be applied to business scenarios such as fintech and healthcare. By integrating object business data, authorized attribute behavior data, equity sensitivity, object level gap, and upgrade association features, it provides continuous data support for object segmentation, upgrade prediction, and strategy generation. Since the object level gap reflects the difference between the current state and the conditions for achieving the level, and the upgrade intention information reflects the response probability of the candidate upgrade level, it can improve the accuracy of target upgrade level determination and strategy result output. The client can be, but is not limited to, various personal computers, laptops, smartphones, tablets, and portable wearable devices. The server can be implemented using a standalone server or a server cluster consisting of multiple servers. The invention will be described in detail below through specific embodiments.

[0014] Please see Figure 2 , Figure 2 This is a flowchart illustrating an embodiment of the object upgrade intention prediction and strategy generation method provided by the present invention. It should be noted that although the logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than that shown here.

[0015] like Figure 2 As shown, the object upgrade intention prediction and strategy generation method proposed in this invention includes the following steps: S10, acquire historical object sample data, and generate analysis and prediction models based on the historical object sample data; In this embodiment, historical object sample data is read from object business records, level change records, rights interaction records, and authorized attribute behavior records, and grouped into a training data set according to object identifiers. The read data undergoes field name unification, time stamp unification, level encoding unification, and missing field completion to obtain a data format suitable for input into the model. The input data for the analysis model includes sample object level, sample value performance, and sample rights sensitivity; the output data are cluster centers and cluster labels. The analysis model includes an input layer, a standardization layer, a vector encoding layer, a clustering decision layer, and a cluster output layer. The input layer connects to the standardization layer, the standardization layer connects to the vector encoding layer, the vector encoding layer connects to the clustering decision layer, and the clustering decision layer connects to the cluster output layer.

[0016] The prediction model's input data includes sample object level, sample interest sensitivity, sample level gap, sample upgrade association features, and object upgrade markers. The output data is the upgrade intention result. The prediction model consists of a categorical feature embedding layer, a continuous feature transformation layer, a feature fusion layer, and an intention output layer. The categorical feature embedding layer and the continuous feature transformation layer are connected to the feature fusion layer, which in turn is connected to the intention output layer. During training, the analysis model uses sample object value vectors for cluster training. Training parameters include the number of clusters, sample weights, and iteration rounds. The prediction model is trained using object upgrade markers as a supervisory signal. Training parameters include a 30-day sample time window, a learning rate of 0.001, a batch size of 128, the positive and negative sample ratio, and probability calibration parameters.

[0017] In offline batch training scenarios, training data is generated using a periodic extraction method. In fintech business, policy service records, account service records, membership level change records, benefit claim records, and account manager contact records are extracted. Sample benefit sensitivity is generated based on the number of benefit claims and benefit browsing behavior. The sample object value vector is input into the analysis model to obtain cluster centers. Sample level differences, sample upgrade correlation features, and object upgrade markers are input into the prediction model to obtain upgrade intention results.

[0018] In the context of continuous updates, an incremental sample caching approach is used to update the model. New business records, benefit interaction records, and level change records are written to the incremental sample cache, and data with missing fields or abnormal timestamps are filtered out. The analysis model retains the cluster centers from the previous period and performs local updates using the value vectors of newly added sample objects. The prediction model retains the model parameters from the previous period and performs incremental training using upgrade markers for newly added objects and upgrade association features of newly added samples.

[0019] In the fintech business, life insurance service platforms read membership level change records, policy service records, rights and benefits claim records, account manager contact records, and authorized consumption preference records to generate sample object value vectors and object upgrade tags. The analysis model outputs customer segmentation results based on the sample object value vectors, while the prediction model outputs customer upgrade intention results based on sample level differences, sample upgrade correlation characteristics, and object upgrade tags.

[0020] In the healthcare business, the health management platform reads member service level records, physical examination appointment records, health consultation records, follow-up response records, and authorized health behavior records to generate sample object value vectors and object upgrade tags. The analysis model outputs service object clustering results, and the prediction model outputs service level upgrade intention results.

[0021] This embodiment converts historical object sample data into sample object value vectors and training data with object upgrade labels. The analysis model can form object clustering capabilities, and the prediction model can learn the relationship between upgrade intention judgment. Therefore, the model input is consistent, the interference of scattered data on training results is reduced, and the stability of object clustering and upgrade prediction is improved.

[0022] S20, obtain the object identifier, object business data, and authorized attribute behavior data of the target object, and generate a target object data set based on the object identifier, associating the object business data and the authorized attribute behavior data; In this embodiment, the object identifier is used to establish a matching basis for the same object across multiple data sources. It can be a member ID, customer ID, service ID, encrypted account tag, or a de-identified object primary key. After the object identifier is read from data requests, business records, or authorization records, it needs to undergo format unification and consistency verification to ensure that the same object from different data sources can be identified as the same data subject. Object business data represents the existing state of the target object in the business system and can include level records, service records, contract records, interaction records, rights usage entry records, and service response records. Authorized attribute behavior data represents the attribute changes and behavior changes of the target object within the authorized scope and can include preference tags, service requirements, behavior categories, state change events, and authorization channel records. Before association, the object business data and authorized attribute behavior data undergo field name unification, time stamp unification, source tag writing, and missing field completion. They are then merged using the object identifier as the matching key. The target object data set consists of multiple types of data after association, retaining source tags, time stamps, and integrity tags for subsequent reading of object status according to a unified data standard.

[0023] When object data is stored separately on multiple business platforms, an object identifier mapping table is used to associate object business data and authorized attribute behavior data. In fintech businesses, the member ID is used to read policy service records, account service records, and rights usage records, while the authorization credential is used to read preference tags and consumption behavior categories. In healthcare businesses, the service member ID is used to read health management service records, physical examination rights appointment records, and follow-up response records, while the authorization credential is used to read health behavior categories and status change events. Before associating, fields from different platforms are uniformly named, and the collection time is converted to a uniform timestamp. After associating, a target object data set is generated.

[0024] When object identifiers have multiple encodings, an identifier verification table and encrypted mapping values ​​are used for association. The business number in the object's business data is first converted to an encrypted mapping value, and the authorization number in the authorization attribute behavior data is also converted to an encrypted mapping value. Consistency verification is then performed using these encrypted mapping values. After successful verification, the object's business data and authorization attribute behavior data are merged, and any mismatched data is written to the pending review data area to prevent erroneous associations from affecting the target object data set.

[0025] In the fintech business, after receiving data requests from target customers, the life insurance service platform reads policy service records, rights claim records, and account manager contact records based on the member number, and reads consumption preference tags and service demand change records based on authorization credentials. After all types of records are standardized and time-stamped, a target object data set is generated according to the member number.

[0026] In the healthcare business, the health management platform reads health consultation records, physical examination appointment records, and follow-up response records based on service member IDs, and reads health behavior categories and status change events based on authorization credentials. The system converts records from different sources into a unified field format and completes the association according to the service member ID to generate a target object data set.

[0027] This embodiment uses object identifiers to uniformly associate object business data and authorized attribute behavior data. Before associating, field names, timestamps, and source tags are organized, allowing the target object dataset to simultaneously retain changes in business status and authorized behavior. Because data from different sources is converted to a unified data standard, subsequent readings of object status can reduce deviations caused by data fragmentation and inconsistent identifiers, improving the integrity and usability of the target object dataset.

[0028] S30, extract the current object level and object value data from the target object data set, generate equity sensitivity and object value vector based on the object value data, and process the object value vector based on the analysis model to generate object clustering results; In this embodiment, the level field in the target object dataset is mapped to generate the current object level through level encoding, and the interaction field, rights field, and value field are extracted as object value data. Object value data includes interaction status data, interaction frequency data, value contribution data, rights usage records, and rights content attention records. Rights usage records are encoded using usage intensity, and rights content attention records are encoded using attention intensity. The two types of encoding results are normalized and fused to generate rights sensitivity.

[0029] The object value vector is generated from the current object level, object value data, and rights sensitivity. The current object level is converted into a level ordinal code, interaction status data is converted into a status code, interaction frequency data is converted into a period frequency code, value contribution data is converted into a standardized value, and rights sensitivity is written into the rights response dimension. The analysis model includes an input layer, a standardization layer, a vector alignment layer, a clustering determination layer, and a cluster output layer. After standardization and vector alignment, the object value vector is matched with the cluster centers to generate object clustering results.

[0030] In the context of customer segmentation in fintech businesses, an object value vector is generated using membership level, policy service records, account service records, benefit redemption records, and customer outreach records. Membership level generates the current object level, customer outreach response count generates interaction frequency data, policy service activity and account service status generate value contribution data, and benefit redemption and benefit browsing behavior generates benefit sensitivity. The analysis model outputs customer segmentation results.

[0031] In the context of service recipient segmentation in healthcare services, a recipient value vector is generated using service level, health management service records, physical examination appointment records, health consultation records, and follow-up response records. Service level generates the current recipient level; health consultation frequency and follow-up response status generate recipient value data; physical examination appointments and health content browsing generate benefit sensitivity; and the analysis model outputs service recipient segmentation results.

[0032] This embodiment converts the current object level, object value data, and equity sensitivity into an object value vector with the same structure. The analysis model can generate object clustering results according to a unified input, reducing the impact of differences in fields from multiple sources on object status identification and improving the stability of object clustering results.

[0033] S40, determine candidate upgrade levels and level achievement conditions based on the current object level, and generate object level gap based on the target object data set and the level achievement conditions; In this embodiment, the current object level is represented by a level code, level name, or level ordinal number, and is used to determine the current position of the target object in the level system. Candidate upgrade levels are obtained by comparing the current object level with the level system configuration data. The level system configuration data records the ordinal relationship, level name, level activation status, and level achievement conditions for each level. When filtering candidate upgrade levels, the current level ordinal number corresponding to the current object level is read, and levels with higher ordinal numbers and in an activated state are selected to avoid including unavailable levels in the candidate range.

[0034] The level achievement conditions represent the conditions and baseline data that must be met to enter a candidate upgrade level. Conditions may include service usage frequency, benefit usage frequency, business activity status, service response status, health service participation, and financial service activity. Baseline data represents the target state that each condition must achieve. The current state data corresponding to the level achievement conditions in the target object dataset is extracted and compared item by item with the baseline data to generate the object level gap. The object level gap can be recorded using a structure of gap items, gap direction, gap degree, and compensation priority, allowing the difference between the current object and the candidate upgrade level to be quantified.

[0035] When the hierarchical system is relatively stable, a hierarchical configuration table is used to determine candidate upgrade levels and the conditions for achieving those levels. The hierarchical configuration table records the level sequence, level name, activation status, condition items, and condition baseline data. After reading the current object level, the system filters for higher levels according to the level sequence, extracts the corresponding condition items from the hierarchical configuration table, and then reads the current status data from the target object dataset to generate the object level gap.

[0036] In situations where grade-level conditions need to be adjusted according to business cycles, versioned grade configuration data is used to determine the grade achievement conditions. Versioned grade configuration data includes the effective date, expiration date, and business scenario marker. In fintech businesses, the system can retrieve financial service activity, benefit usage status, and customer outreach response status based on the member service cycle; in healthcare businesses, it can retrieve the number of health consultations, physical examination appointment status, and follow-up response status based on the service cycle. The system selects the corresponding grade achievement conditions based on the effective date and compares them with the current status data in the target object's dataset.

[0037] In the fintech business, life insurance service platforms read the current membership level of target customers and filter for higher membership levels from the membership level configuration data. The system reads the number of times benefits are used, service activity status, and customer outreach response requirements corresponding to higher membership levels, and then compares them with policy service records, benefit usage records, and outreach response records in the target customer's database to generate the target customer's membership level gap.

[0038] In the healthcare business, the health management platform reads the service level of target members and filters for higher service levels. The system reads the number of health consultations, physical examination appointment status, and follow-up response requirements corresponding to higher service levels, and then compares them with the health service records, appointment records, and follow-up records in the target object's database to generate object level differences.

[0039] This embodiment determines candidate upgrade levels based on the current object level and compares the level achievement conditions with the current status data in the target object dataset. This transforms the level upgrade judgment from level name recognition to condition item difference recognition. Since the object level difference includes the difference items, difference direction, and difference degree, the difference between the current object and the candidate upgrade level can be quantified, improving the accuracy and traceability of level difference judgment.

[0040] S50, extract upgrade-related features from the target object data set, process the current object level, the rights and interests sensitivity, the object level gap and the upgrade-related features based on the prediction model, generate upgrade intention information, and determine the target upgrade level and object reach priority from the candidate upgrade levels based on the upgrade intention information; In this embodiment, data related to upgrade judgment in the target object dataset is extracted as upgrade-related features. These features can consist of object historical behavior features, object attribute features, status change event markers, and upgrade acceptance capacity indicators. Object historical behavior features can originate from service access, rights response, consultation records, appointment records, and outreach responses, reflecting changes in the target object's behavior within a set period. Object attribute features can originate from object category, service preferences, stable level status, and authorized attribute behavior. Status change event markers are used to indicate changes in family structure, health status, service needs, or business stage. Upgrade acceptance capacity indicators represent the target object's degree of acceptance of higher-level conditions, rights content, and service resource investment.

[0041] The predictive model receives current object level, rights sensitivity, object level gap, and upgrade-related features. The current object level is converted into a level code, rights sensitivity is converted into a rights response dimension, the object level gap is converted into a gap item vector, and the upgrade-related features are converted into a behavior and attribute fusion vector. The predictive model may include an input layer, a categorical feature embedding layer, a continuous feature transformation layer, a feature fusion layer, and a willingness output layer. The input layer receives multiple types of input; the categorical feature embedding layer processes level, event, and label data; the continuous feature transformation layer processes rights sensitivity, object level gap, and behavior intensity data; the feature fusion layer merges multiple class vectors; and the willingness output layer generates upgrade willingness information corresponding to candidate upgrade levels.

[0042] Upgrade intention information indicates the likelihood of a target object responding to a candidate upgrade level. This information may include the candidate upgrade level identifier, intention strength, confidence level, and ranking position. After candidate upgrade levels are sorted according to upgrade intention information, the level that meets the ranking requirements and has a reachable object level difference is selected as the target upgrade level. Object reach priority is jointly generated by the upgrade intention information corresponding to the target upgrade level and the reachable object level difference. Objects with higher intention strength and smaller level differences receive a higher reach order.

[0043] In batch prediction scenarios, an offline feature table is used to generate upgrade-related features. In fintech businesses, the offline feature table may include customer service response counts, benefit claim behavior, policy service status, account activity status, and status change events; in healthcare businesses, the offline feature table may include health consultation counts, physical examination benefit appointments, follow-up responses, health content browsing, and health status changes. After reading the current object level, benefit sensitivity, object level gap, and upgrade-related features, the prediction model outputs upgrade intention information corresponding to candidate upgrade levels and generates the target upgrade level and object outreach priority.

[0044] In real-time outreach scenarios, a recent behavior window is used to update associated features. The window length can be set to 30 or 90 days. After behavioral data enters the window, it updates rights response, service access, consultation appointment, and status change events. The prediction model retains the trained model parameters and only updates the window for the input features, ensuring that recent behavioral changes are included in the upgrade intention information. The target reach priority is updated synchronously according to the upgrade intention information and the target level difference, so that the terminal device can present the reachable targets in sequence.

[0045] In the fintech business, life insurance service platforms extract customer service response, rights claim behavior, policy service status, and account activity status from target user datasets as upgrade-related features. Predictive models combine current user level, rights sensitivity, and user level gaps to output upgrade intention information for each candidate member level, and determine target member levels and customer outreach priorities.

[0046] In the healthcare business, health management platforms extract health consultation frequency, physical examination appointments, follow-up responses, and changes in health status from the target population's data set as upgrade-related features. The predictive model combines service level, benefit sensitivity, and the gap between the target population's service level and their own, outputting upgrade intention information for each candidate service level and determining the target service level and service delivery priority.

[0047] This embodiment extracts upgrade-related features from the target object dataset and inputs the current object level, rights sensitivity, object level gap, and upgrade-related features into the prediction model. The upgrade intention information can simultaneously reflect the level status, rights response, gap size, and behavioral changes. Since the target upgrade level is determined by the upgrade intention information among the candidate upgrade levels, and the object reach priority is generated by combining the upgrade intention information and the object level gap, the selection and ranking of upgrade objects have a unified data basis, which can improve the stability of upgrade prediction results and reach order.

[0048] S60, based on the target object data set, the object grouping result, the rights sensitivity, the current object level, the target upgrade level, and the object level gap, a strategy result is generated, and the object reach priority and the strategy result are sent to the terminal device for presentation.

[0049] In this embodiment, the target object dataset provides the data range required for generating the strategy results. The object clustering results represent the group categories of target objects in terms of value and behavior. The rights sensitivity represents the responsiveness of target objects to rights resources. The current object level and the target upgrade level determine the direction of level change. The object level gap represents the conditions that the target object still needs to meet to reach the target upgrade level. When generating the strategy results, the rights records, service records, interaction records, and status data in the target object dataset are filtered, and the strategy category is determined in conjunction with the object clustering results. Then, the display order of rights content is determined based on the rights sensitivity, and the upgrade description content is determined based on the current object level, the target upgrade level, and the object level gap. Finally, data results that can be displayed by the terminal device are generated.

[0050] The strategy results can include benefit recommendation content, product recommendation content, gap analysis content, upgrade path content, and communication content. Benefit recommendation content is generated based on benefit sensitivity and the benefit resources corresponding to the target upgrade level; product recommendation content is generated based on target grouping results and the business status in the target target data set; gap analysis content is generated based on target level differences; upgrade path content is generated based on the gap items and compensation conditions in the target level differences; and communication content is generated based on the current target level, the target upgrade level, benefit recommendation content, and upgrade path. The target reach priority and strategy results are sent to the terminal device together. The terminal device displays targets according to their reach priority and shows the content in the strategy results that matches the target targets.

[0051] In the context of customer management within fintech businesses, a strategy template engine is used to generate strategy results. The engine reads policy service status, account service status, benefit redemption records, and customer outreach records from the target customer's dataset. It then combines this with customer segmentation results to determine the customer management type, filters benefit content that customers are more likely to respond to based on benefit sensitivity, and generates a level upgrade description based on the current customer level, target upgrade level, and the gap between customer levels. After the strategy results are sent to the account manager's terminal, the terminal interface arranges the customer list according to outreach priority and displays customer segments, upgrade goals, and strategy content.

[0052] In the service operation of healthcare services, a service strategy generator is used to generate strategy results. The service strategy generator reads health service records, physical examination appointment records, health consultation records, and follow-up response records from the target object's dataset. It then determines the service object's status based on object segmentation results, filters physical examination benefits, health consultation benefits, or health management benefits based on benefit sensitivity, and generates service level improvement suggestions based on object level differences. After the strategy results are sent to the service terminal, the terminal interface displays service objects and corresponding service content according to object reach priority.

[0053] This embodiment generates a strategy result by combining the target object dataset, object grouping results, rights sensitivity, current object level, target upgrade level, and object level gap. The strategy result can simultaneously reflect the object status, rights response level, level change target, and gap situation. After the object reach priority and strategy result are sent to the terminal device, the content displayed on the terminal can be arranged according to object value and upgrade probability, reducing the deviation caused by manual screening and sorting, and improving the accuracy and operability of strategy presentation.

[0054] In one embodiment, step S10 includes: S101, extract sample object identifier, sample object business data, sample authorization attribute behavior data, historical rights interaction records and historical level change records from historical object sample data, and generate a historical sample data set based on the sample object identifier associated with the sample object business data, the sample authorization attribute behavior data, the historical rights interaction records and the historical level change records; S102, extract sample object level, sample value characteristics, sample rights usage records and sample rights content attention records from the historical sample data set, generate sample rights sensitivity based on the sample rights usage records and the sample rights content attention records, and generate sample object value vector based on the sample object level, the sample value characteristics and the sample rights sensitivity; S103, perform cluster training based on the value vectors of the sample objects to generate an analysis model; S104, Based on the historical level change records, generate object upgrade markers, extract sample upgrade association features from the historical sample data set, and determine sample candidate upgrade levels and corresponding level achievement conditions from the preset level system based on the sample object level. S105, extract the current state data of the sample from the historical sample data set, and generate the sample object level difference corresponding to the sample candidate upgrade level based on the level achievement conditions corresponding to the sample candidate upgrade level and the current state data of the sample. S106, a prediction model is generated by performing prediction training based on the sample object level, the sample interest sensitivity, the sample object level gap corresponding to the sample candidate upgrade level, the sample upgrade association features, and the object upgrade marker.

[0055] In this embodiment, historical object sample data is merged according to sample object identifiers before entering training. The sample object identifier is used to distinguish different training objects and serves as the association key between sample object business data, sample authorized attribute behavior data, historical rights interaction records, and historical level change records. Sample object business data may include level records, service records, interaction records, and business status records; sample authorized attribute behavior data may include attribute tags, behavior categories, and status change information within the authorized scope; historical rights interaction records are used to extract rights usage behavior and rights content attention behavior; historical level change records are used to determine whether a training object has experienced a level upgrade during the observation period. After unifying field names, timestamps, and level codes for all types of data, they are grouped into the historical sample data set according to the sample object identifiers.

[0056] The sample object level, sample value features, sample rights usage records, and sample rights content attention records extracted from the historical sample dataset are used to generate the sample object value vector. The sample object level is converted into a level ordinal code, the sample value features are converted into numerical features, the sample rights usage records are converted into rights usage intensity, and the sample rights content attention records are converted into rights attention intensity. Rights usage intensity and rights attention intensity are fused to generate the sample rights sensitivity. The sample object level, sample value features, and sample rights sensitivity are then concatenated, standardized, and dimensionally aligned to obtain the sample object value vector.

[0057] The analysis model is trained using the value vectors of sample objects. The model includes an input layer, a normalization layer, a vector encoding layer, a clustering decision layer, and a grouping output layer. The input layer receives the value vectors of sample objects; the normalization layer unifies the value ranges of different dimensions; the vector encoding layer converts the rank, value, and equity dimensions into the same vector structure; the clustering decision layer generates cluster centers based on vector distance or similarity; and the grouping output layer stores the cluster centers and cluster labels. Training parameters may include the number of clusters, sample weights, distance metric, and iteration rounds.

[0058] Object upgrade markers are generated from historical level change records. These records include the starting level of the training period, the ending level of the training period, the time of the level change, and the status of the level change. Object upgrade markers are generated by comparing the starting and ending levels. Sample upgrade-related features are extracted from the historical sample dataset and may include historical behavioral features, attribute features, state change events, and upgrade-capacity indicators. Candidate upgrade levels are generated from the sample object level and a preset level system. The level achievement conditions corresponding to the candidate upgrade level are read from the preset level system.

[0059] The current state data of the sample is extracted from the historical sample dataset and compared item by item with the achievement conditions corresponding to the candidate upgrade levels of the sample, generating the sample object level gap corresponding to the candidate upgrade levels. The training input of the prediction model includes the sample object level, sample interest sensitivity, sample object level gap corresponding to the candidate upgrade levels, sample upgrade-related features, and object upgrade labels. The prediction model includes a category feature embedding layer, a continuous feature transformation layer, a feature fusion layer, and a willingness output layer. The category feature embedding layer receives level, event, and label data; the continuous feature transformation layer receives interest sensitivity, level gap, and behavior intensity data; the feature fusion layer merges feature vectors from multiple classes; and the willingness output layer generates the upgrade willingness prediction result.

[0060] This embodiment merges historical object sample data by identifying sample objects and converts sample object level, sample value characteristics, and sample rights sensitivity into sample object value vectors. The analysis model can form a clustering capability based on a unified vector structure. Object upgrade markers are generated by recording historical level changes, and the prediction model is trained by combining sample object level, sample rights sensitivity, sample object level gap, and sample upgrade correlation features. The prediction model can learn the correspondence between level changes and object behavior, rights response, and level gap, thereby improving the consistency of training inputs between the analysis model and the prediction model and reducing the interference of differences in multi-source sample data on the training results.

[0061] In one embodiment, step S20 above includes: S201, Obtain the data acquisition request of the target object, parse the object identifier and authorization credential from the data acquisition request, and generate an authorization verification result based on the authorization credential; S202, when the authorization verification result indicates that the authorization is passed, read the object business data based on the object identifier, and read the authorization attribute behavior data based on the object identifier and the authorization credential; S203, unify the field names, unify the time stamps, and write the source stamps to the object business data and the authorized attribute behavior data to generate object business data to be associated and authorized attribute behavior data to be associated; S204, extract the object identifier field from the business data of the object to be associated and the authorization attribute behavior data to be associated, perform consistency verification between the object identifier field and the object identifier, and generate the object identifier verification result; S205, when the object identifier verification result is passed, generate associated object data based on the object identifier to associate the business data of the object to be associated and the authorized attribute behavior data to be associated; S206, Generate a data integrity tag based on the associated object data, and generate a target object data set based on the associated object data, the source tag in the business data of the object to be associated, the source tag in the authorized attribute behavior data to be associated, the time tag, and the data integrity tag.

[0062] In this embodiment, the data retrieval request for the target object may carry an object identifier, authorization credential, request time, request source, and data scope marker. The object identifier is used to identify the same object across different data sources and can be a member ID, customer ID, service ID, de-identified object primary key, or encrypted mapping value. The authorization credential is used to limit the scope of authorized attribute behavior data reading and can be an authorization token, authorization record number, authorization validity period marker, and authorization scope marker. After parsing the data retrieval request, the authorization credential is verified for validity period, authorization scope, and source consistency to obtain the authorization verification result. The authorization verification result is used to control subsequent data reading actions and prevent unauthorized data from being mixed into the target object's data set.

[0063] When the authorization verification result indicates that the authorization has passed, the object's business data is read based on the object identifier. The object's business data may include level records, service records, contract status, business processing records, interaction records, and rights entry records. Authorized attribute and behavior data is read based on the object identifier and authorization credentials, used to obtain information on attribute tags, behavior categories, preference changes, requirement changes, and status changes within the authorized scope. The object identifier is used to limit the objects read, and the authorization credentials are used to limit read permissions and field ranges; together, they reduce object mismatches and out-of-bounds reads.

[0064] When object business data and authorized attribute behavior data come from different data sources, field names, time formats, and source tags may differ. Standardized field names are used to convert data items such as level names, service statuses, behavior types, and time fields into unified field names; standardized time tags are used to convert different time formats to the same time granularity; and source tags are used to record whether the data originates from business records, authorization records, benefit records, or interaction records. After completing the above steps, object business data generates object business data to be associated, and authorized attribute behavior data generates authorized attribute behavior data to be associated.

[0065] The object identifier field is extracted from both the business data of the object to be associated and the authorization attribute behavior data to be associated. The object identifier field is then compared with the object identifier parsed from the data retrieval request for consistency verification. Consistency verification can include field value matching, encrypted mapping value matching, masked primary key matching, and multi-field combination matching. The object identifier verification result is used to determine whether the two types of data belong to the same target object. Data that fails the verification can be written to the abnormal data area and will not participate in the generation of associated object data.

[0066] When the object identifier verification result is successful, associated object data is generated by associating the business data of the object to be associated and the authorized attribute behavior data of the object to be associated based on the object identifier. The association process can use primary key merging, time window merging, or field mapping merging. Primary key merging is used for data with completely identical object identifiers; time window merging is used for time-sharing records of the same object generated in different systems; and field mapping merging is used for records where different field names express the same data meaning. The associated object data retains source and time stamps for easy tracing of data source and time status later.

[0067] Data integrity markers are generated from associated object data and can be based on key field coverage status, object business data existence status, authorized attribute behavior data existence status, timestamp continuity status, and source marker integrity status. The target object data set consists of associated object data, source markers from the business data of the object to be associated, source markers from the authorized attribute behavior data to be associated, timestamps, and data integrity markers. Once the target object data set is formed, the business status, authorized attribute behavior, and data source status of the same target object are integrated into the same data structure.

[0068] This embodiment reduces the entry of unauthorized data and non-target object data into the target object data set by parsing the object identifier and authorization credentials from the data acquisition request and using the authorization verification results to control the reading scope of object business data and authorized attribute behavior data. By unifying field names, timestamps, and source tags, it reduces field format differences between different data sources. By verifying the consistency between the object identifier field and the object identifier, it reduces object mismatches. By generating the target object data set by associating object data and data integrity tags, it ensures that the target object's business data, authorized behavior data, and data quality status are simultaneously preserved, improving the integrity and reliability of subsequent data reading.

[0069] In one embodiment, step S30 above includes: S301, Read the level field from the target object data set, generate the current object level, obtain the object value field mapping table, and extract interaction status data, interaction frequency data, value contribution data, rights usage records and rights content attention records from the target object data set based on the object value field mapping table to generate object value data; S302, Based on the number of times rights were used, the type of rights used, and the time of rights used in the rights use record of the object value data, generate rights use intensity features; S303, Based on the content browsing behavior, content response behavior and content collection behavior in the rights and interests content attention record of the object value data, generate rights and interests content attention intensity characteristics; S304, Generate rights sensitivity based on the rights usage intensity characteristics and the rights content attention intensity characteristics; S305, Generate an object value vector based on the current object level, the interaction status data, the interaction frequency data, the value contribution data, and the rights sensitivity; S306, Input the object value vector into the analysis model, perform clustering determination on the object value vector based on the cluster centers and cluster labels in the analysis model, and generate object clustering results.

[0070] In this embodiment, the level field in the target object dataset generates the current object level through level coding mapping. Level coding mapping converts the level name, level number, and level sequence into a unified level expression, avoiding inconsistencies in level field formats from different data sources. The object value field mapping table is used to determine the data items to be read. According to the field mapping relationship, it extracts interaction status data, interaction frequency data, value contribution data, rights usage records, and rights content attention records from the target object dataset and combines them into object value data. Interaction status data reflects the target object's response status to service outreach, business reminders, or rights notifications; interaction frequency data reflects the number of times the target object accesses, inquires, makes appointments, or responds within a set period; and value contribution data reflects the depth of the target object's use of business resources, service resources, or rights resources.

[0071] The frequency, type, and duration of use of rights in the rights usage record are used to generate rights usage intensity characteristics. Frequency reflects the degree of use, type reflects the preferred rights category, and duration reflects the relevance of the behavior. Content browsing, response, and collection behaviors in the rights content attention record are used to generate rights content attention intensity characteristics. Content browsing reflects the scope of attention, response reflects the willingness to interact, and collection reflects the tendency to maintain continued attention. After normalization and fusion, the rights usage intensity characteristics and rights content attention intensity characteristics generate rights sensitivity, enabling rights usage and attention behaviors to be converted into the same evaluation dimension.

[0072] The object value vector is generated from the current object level, interaction status data, interaction frequency data, value contribution data, and rights sensitivity. The current object level is converted to a level ordinal code, interaction status data is converted to a status code, interaction frequency data is converted to a periodic frequency code, value contribution data is converted to a standardized value, and rights sensitivity is written into the rights response dimension. After missing value imputation, outlier truncation, and vector length alignment across different dimensions, an object value vector that can be recognized by the analysis model is formed.

[0073] The analysis model comprises a standardization layer, a vector alignment layer, a clustering determination layer, and a clustering output layer. The standardization layer unifies the value ranges of different dimensions in the object's value vector. The vector alignment layer fixes the positions of the level, interaction, value, and rights dimensions. The clustering determination layer matches the object's value vector with the cluster centers based on distance or similarity. The clustering output layer reads the cluster labels based on the matching results and generates the object clustering results. The object clustering results reflect the comprehensive category of the target object in terms of level status, behavioral activity, value performance, and rights response.

[0074] This embodiment extracts object value data from the target object dataset through an object value field mapping table, and converts rights usage records and rights content attention records into rights sensitivity. Level status, interaction activity, value performance, and rights response level can all be included in the same object value vector. After the analysis model performs clustering based on the object value vector, the format and dimensional differences between fields from different sources are reduced, improving the stability and interpretability of the object clustering results.

[0075] In one embodiment, step S40 above includes: S401, Read the level system configuration data, and determine the current level sequence corresponding to the current object level from the level system configuration data; S402, Based on the current level position, filter levels with higher level positions than the current level position from the level system configuration data to generate candidate upgrade levels; S403, extract the level achievement conditions corresponding to the candidate upgrade level from the level system configuration data, and split the level achievement conditions into a set of condition items corresponding to the candidate upgrade level and condition benchmark data corresponding to the candidate upgrade level; S404, extract the current status data from the target object data set based on the set of condition items corresponding to the candidate upgrade level; S405, compare the current status data with the condition benchmark data corresponding to the candidate upgrade level item by item to generate the item gap data corresponding to the candidate upgrade level; S406, Summarize the project gap data corresponding to the candidate upgrade levels according to the candidate upgrade levels, and generate the object level gap.

[0076] In this embodiment, the grading system configuration data stores the grading sequence, grading name, activation status, version flag, applicable object scope, condition item identifier, and condition baseline data. After reading the current object's grading level, the current object's grading level is converted to the current grading sequence using a grading code mapping table. The grading code mapping table is used to resolve inconsistencies in grading names, grading numbers, and grading sequences from different data sources, ensuring that subsequent candidate upgrade grading selection follows a unified grading order.

[0077] Candidate upgrade levels are selected based on the current level ranking and the level system configuration data. During selection, levels with a higher ranking than the current level are read from the level system configuration data, and levels ineligible for upgrade consideration are removed based on their activation status, version flag, and applicable scope. This results in candidate upgrade levels. There can be one or multiple candidate upgrade levels. When multiple candidate upgrade levels exist simultaneously, each candidate upgrade level retains its corresponding level ranking and level identifier for subsequent generation of difference results.

[0078] The conditions for achieving a promotion level are extracted from the promotion level configuration data and mapped one-to-one with the candidate promotion levels. These conditions are broken down into a set of condition items and baseline condition data. The set of condition items identifies the status fields that need to be read from the target object data set, such as service usage status, benefit usage status, interaction response status, business activity status, and level maintenance status. The baseline condition data records the target value, target range, status requirement, or completion marker for each condition item. The split set of condition items and the baseline condition data maintain the correspondence between candidate promotion levels, preventing the mixing of achievement conditions across multiple candidate promotion levels.

[0079] The current status data is extracted from the target object dataset based on the set of conditional items. During extraction, the corresponding fields are read according to their identifiers in the conditional item set, and issues such as missing fields, differences in field format, and differences in time range are addressed. For time-related fields, data is extracted according to the period range corresponding to the level achievement conditions; for status-related fields, data is converted to a unified status code according to the status dictionary; for frequency and intensity-related fields, current values ​​are generated according to the statistical period. The current status data is then compared item by item with the baseline data corresponding to the candidate upgrade levels to generate the item gap data corresponding to the candidate upgrade levels. The item gap data may include the gap item identifier, current value, target value, gap direction, gap degree, and compensation status.

[0080] The object level gap is obtained by aggregating the gap data of items corresponding to the candidate upgrade levels. During aggregation, the gap data of each item is collected according to the candidate upgrade level, and items that have not been achieved, have been achieved, are close to being achieved, and cannot be determined are retained under each candidate upgrade level. The object level gap includes both the fine-grained gap of each condition item and the overall gap result at the candidate upgrade level, so that subsequent upgrade judgments can directly read the conditional differences between the target object and the candidate upgrade level.

[0081] This embodiment determines the current level order and candidate upgrade levels through the level system configuration data, and breaks down the level achievement conditions corresponding to the candidate upgrade levels into a set of condition items and condition benchmark data. The current status data in the target object data set can be accurately extracted according to the condition items. By comparing the current status data with the condition benchmark data item by item and summarizing them into the object level difference, the level upgrade judgment is transformed from a simple level identifier judgment to a condition difference judgment, reducing the deviation caused by the mixed use of level conditions and inconsistency of status fields, and improving the accuracy and traceability of object level difference generation.

[0082] In one embodiment, step S50 above includes: S501, extract object historical behavior features, object attribute features, state change event markers and upgrade acceptance capability indicators from the target object data set, and generate upgrade association features based on the object historical behavior features, the object attribute features, the state change event markers and the upgrade acceptance capability indicators; S502, based on the candidate upgrade level, the current object level, the rights and interests sensitivity, the object level difference corresponding to the candidate upgrade level and the upgrade association feature are combined into a candidate level prediction sample; S503, Input the candidate level prediction sample into the prediction model to generate upgrade intention information corresponding to the candidate upgrade level; S504, Sort the candidate upgrade levels based on the upgrade intention information corresponding to the candidate upgrade levels to generate a candidate upgrade level sequence; S505, determine the target upgrade level from the candidate upgrade level sequence, and extract the upgrade intention information corresponding to the target upgrade level; S506, generate object reach priority based on the upgrade intention information corresponding to the target upgrade level and the object level difference corresponding to the target upgrade level.

[0083] In this embodiment, the behavior records, attribute records, status change records, and upgrade acceptance capability records stored in the target object dataset are extracted to form upgrade association features. The object's historical behavior features are generated from service access counts, rights response records, consultation records, appointment records, contact feedback records, and behavior time intervals, representing the target object's activity level and response changes within a set period. The object attribute features are generated from object category, level stability, preference category, service stage, and authorization attribute tags, representing the target object's static state and long-term preferences. Status change event markers are generated from status change time, event category, event duration, and event impact scope, indicating whether the target object has recently experienced any changes affecting upgrade judgment. Upgrade acceptance capability indicators are generated from level condition completion status, service resource utilization capacity, rights content acceptance degree, and interactive response capability, representing the target object's acceptance degree of higher level conditions and rights content. The above data is processed through field encoding, time window filtering, missing value imputation, and vector concatenation to generate upgrade association features.

[0084] Candidate upgrade level prediction samples are generated on a per-candidate upgrade-level basis. The current object level is converted into a level ordinal code, the rights sensitivity is converted into a rights response value, the object level gap corresponding to the candidate upgrade level is converted into a gap item vector, and the upgrade-related features are converted into a behavioral attribute fusion vector. Candidate upgrade levels are written into the same prediction sample as candidate target dimensions, ensuring that each candidate upgrade level has independent input data. Before entering the prediction model, candidate upgrade level prediction samples undergo dimension alignment and value normalization to avoid the impact of dimensional differences between the level field, gap field, and behavioral field on the model output.

[0085] After receiving candidate level prediction samples, the prediction model generates upgrade intention information corresponding to the candidate upgrade levels. The prediction model includes a categorical feature embedding layer, a continuous feature transformation layer, a feature fusion layer, and an intention output layer. The categorical feature embedding layer processes the current object level, candidate upgrade levels, state change event markers, and object attribute features. The continuous feature transformation layer processes rights sensitivity, object level gap, object historical behavior features, and upgrade acceptance capacity indicators. The feature fusion layer merges categorical vectors and continuous vectors. The intention output layer outputs the upgrade intention information corresponding to the candidate upgrade levels. Upgrade intention information may include intention strength, confidence level, candidate level identifier, and ranking score.

[0086] Candidate upgrade levels are sorted based on their corresponding upgrade intention information, generating a candidate upgrade level sequence. During sorting, intention strength, confidence level, and target level difference all contribute to the ranking; candidate upgrade levels with higher intention strength, higher confidence level, and smaller target level difference are ranked higher. Once the target upgrade level is determined from the candidate upgrade level sequence, the upgrade intention information and the target level difference corresponding to the target upgrade level are extracted. Target reach priority is generated based on the upgrade intention information and the target level difference corresponding to the target upgrade level; higher intention strength and smaller difference result in higher target reach priority.

[0087] This embodiment extracts historical behavioral features, attribute features, state change event markers, and upgrade acceptance capacity indicators from the target object dataset. The upgrade-related features can simultaneously reflect the target object's behavioral changes, attribute status, recent events, and acceptance capacity. By combining the current object level, rights sensitivity, the object level gap corresponding to the candidate upgrade level, and upgrade-related features into a candidate level prediction sample, the prediction model can generate upgrade intention information for different candidate upgrade levels. By generating a candidate upgrade level sequence based on the upgrade intention information and combining it with the object level gap corresponding to the target upgrade level to generate object reach priority, the upgrade level determination and reach ranking can simultaneously refer to the degree of intention and the gap status, improving the accuracy of upgrade prediction results and reach order.

[0088] In one embodiment, step S60 above includes: S601, Generate an object profile based on the target object data set, the object grouping results and the rights and interests sensitivity, and read the rights and interests tags, product tags and level achievement conditions corresponding to the target upgrade level from the level system configuration data; S602, Generate a benefit matching result based on the degree of matching between the object profile and the benefit tag, and generate a product matching result based on the degree of matching between the object profile and the product tag; S603, generate a level enhancement attractiveness based on the current object level and the target upgrade level, and generate a rights and benefits product matching degree based on the rights and benefits matching result, the product matching result and the level enhancement attractiveness; S604, Generate equity recommendation results and product recommendation results based on the equity product matching degree; S605, Generate equity gap analysis results based on the object level gap and the level achievement conditions, and generate upgrade path planning results based on the equity gap analysis results; S606, Generate communication content based on the object profile, the current object level, the target upgrade level, the benefit recommendation result, the product recommendation result, the benefit gap analysis result, and the upgrade path planning result; S607, based on the benefit recommendation results, the product recommendation results, the benefit gap analysis results, the upgrade path planning results, and the communication content generation strategy results; S608, generate terminal display data based on the object reach priority and the strategy result, and send the terminal display data to the terminal device for presentation.

[0089] In this embodiment, the target object dataset, object grouping results, and rights sensitivity are used together to generate object profiles. The target object dataset provides business status, service records, rights records, interaction records, and status data; the object grouping results provide group category labels to distinguish different objects in terms of value status and behavioral status; and the rights sensitivity provides the degree of rights response to identify the intensity of an object's attention to rights resources. When generating object profiles, the business status field, group category label, and rights response field are normalized, labels are merged, and weights are assigned to obtain profile data that includes level status, value status, rights preference, and interaction status.

[0090] The system reads the benefit tags, product tags, and level achievement conditions for the target upgrade level from the level system configuration data. Benefit tags describe the types of benefits offered by the target upgrade level, product tags describe the service or product categories of the recommended objects, and level achievement conditions describe the conditions that must be met for the target upgrade level. The object profile is matched against the benefit tags and product tags respectively. The benefit preferences, interaction states, and value states in the profile are matched and similarity is determined with the tag fields to generate benefit matching results and product matching results. The level difference between the current object level and the target upgrade level is converted into a level upgrade attractiveness level, which represents the strength of the target upgrade level's attractiveness relative to the current object level in terms of benefit scope, service level, and accessibility.

[0091] The combination of benefit matching results, product matching results, and the attractiveness of level upgrades constitutes the benefit-product matching score. This score is used to determine the recommendation order among multiple candidate benefits and products; benefits or products with higher matching scores are included in the benefit recommendation results and product recommendation results. The target level gap and level achievement conditions are used to generate benefit gap analysis results, which record unmet conditions, met conditions, the degree of gap, and directions for improvement. The upgrade path planning results are generated based on the benefit gap analysis results and may include conditions to be improved, suggested completion order, available benefit hints, and a description of the target upgrade level.

[0092] The communication content is generated from the target profile, current target level, target upgrade level, benefit recommendation results, product recommendation results, benefit gap analysis results, and upgrade path planning results. The communication content can be output using a combination of structured fields and natural language text. The structured fields are used for display on terminal devices, while the text content is used by service personnel or service interfaces. The strategy results are encapsulated from the benefit recommendation results, product recommendation results, benefit gap analysis results, upgrade path planning results, and communication content, retaining the target identifier, generation time, strategy category, and data source markers. The target reach priority is combined with the strategy results to form the terminal display data. After the terminal display data is sent to the terminal device, the terminal device arranges the targets according to their reach priority and displays the matching strategy results.

[0093] This embodiment generates object profiles by using a target object dataset, object grouping results, and rights sensitivity. The strategy results can simultaneously utilize object status, group category, and rights responsiveness. It generates rights recommendation results and product recommendation results by matching rights tags, product tags, and object profiles, ensuring that the recommended content aligns with object preferences and service status. It generates rights gap analysis results and upgrade path planning results by analyzing object level differences and level achievement conditions, converting the difference between the target object and the target upgrade level into displayable content. Finally, it generates terminal display data by using object reach priority and strategy results, enabling terminal devices to present object and strategy content in priority order, improving the accuracy, readability, and ease of operation of the strategy results output.

[0094] In one embodiment, in a life insurance member management scenario, the insurance service platform provides agents with intelligent customer acquisition services based on member benefits. The platform first reads historical sample data from multiple historical member periods. This historical sample data includes member ID, member level records, policy service records, benefit redemption records, benefit browsing records, account manager contact records, authorized consumption preference records, and member level change records. The platform extracts sample object identifiers, sample object business data, sample authorized attribute behavior data, historical benefit interaction records, and historical level change records from the historical sample data. Based on the sample object identifiers, it associates the data of the same historical member across the policy service platform, member benefits platform, and customer service platform to generate a historical sample data set. The platform extracts sample object level, sample value characteristics, sample benefit usage records, and sample benefit content attention records from the historical sample data set. Based on benefit redemption, benefit reservation, and benefit redemption behaviors in the sample benefit usage records, it generates sample benefit sensitivity and converts the sample object level, sample value characteristics, and sample benefit sensitivity into a sample object value vector. The platform uses clustering training based on the value vectors of sample objects to generate an analytical model. This model enables the platform to segment members into groups such as high-level, highly active members, low-level, high-potential members, and stagnant members. The platform generates upgrade markers based on historical level change records, extracts upgrade-related features from historical sample data, and determines candidate upgrade levels and corresponding achievement conditions from a pre-defined level system based on the sample object's level. The platform extracts current state data from historical sample data, compares the achievement conditions of candidate upgrade levels with the current state data, generates the level gap between candidate upgrade levels, and then uses the sample object level, sensitivity to rights and benefits, level gap between candidate upgrade levels, upgrade-related features, and upgrade markers for predictive training to generate a predictive model. This predictive model outputs upgrade intention information based on changes in member's current level, sensitivity to rights and benefits, level gap, and behavioral attributes.

[0095] When an agent prepares to reach a target customer, the platform receives the customer's data acquisition request, parses the object identifier and authorization certificate from the request, and generates an authorization verification result based on the authorization certificate. If the authorization verification result is successful, the platform reads the object's business data based on the object identifier, and then reads the authorization attribute behavior data based on the object identifier and authorization certificate. The object's business data includes current membership level, policy service status, account service status, service interaction records, benefits entry records, and account manager outreach records. The authorization attribute behavior data includes data such as authorized consumption preferences, changes in service needs, interest preferences, changes in family structure, and changes in health status. The platform standardizes field names, timestamps, and source tags for the object's business data and authorization attribute behavior data, generating object business data to be associated and authorization attribute behavior data to be associated. Then, the platform extracts the object identifier field from both, performs consistency verification between the object identifier field and the object identifier, and generates an object identifier verification result. When the object identifier verification result is successful, the platform associates the business data of the object to be associated and the authorized attribute behavior data to be associated based on the object identifier, generates associated object data, and generates a data integrity tag based on the associated object data. The platform generates a target object data set based on the associated object data, the source tag in the business data of the object to be associated, the source tag in the authorized attribute behavior data to be associated, the time tag, and the data integrity tag, so that the target customer's data on policies, rights, service interactions, and authorization behaviors are integrated into the same data structure.

[0096] The platform reads the level field from the target object dataset to generate the current object level and obtains an object value field mapping table. Based on the object value field mapping table, the platform extracts interaction status data, interaction frequency data, value contribution data, benefit usage records, and benefit content attention records from the target object dataset to generate object value data. The number of benefit usages, benefit usage types, and benefit usage time in the benefit usage records are used to generate benefit usage intensity features, while content browsing behavior, content response behavior, and content collection behavior in the benefit content attention records are used to generate benefit content attention intensity features. The platform generates benefit sensitivity based on the benefit usage intensity features and benefit content attention intensity features, and generates an object value vector based on the current object level, interaction status data, interaction frequency data, value contribution data, and benefit sensitivity. After the object value vector is input into the analysis model, the analysis model performs clustering based on cluster centers and cluster labels to generate object clustering results. For example, a target customer is currently at the Gold Card member level and has recently been continuously browsing high-end health checkup benefits, family protection benefits, and value-added service benefits, while also showing a high response to agent outreach. The analysis model can classify the target customer as a high-potential member with high benefit attention.

[0097] The platform reads the tier system configuration data and determines the current tier ranking corresponding to the current target's tier. Based on this ranking, it then filters tiers with higher rankings from the tier system configuration data to generate candidate upgrade tiers. If the target customer is currently a Gold Card member, candidate upgrade tiers can include Platinum and Diamond memberships. The platform extracts the tier achievement conditions corresponding to the candidate upgrade tiers from the tier system configuration data and breaks these conditions down into a set of condition items and baseline condition data. The condition item set can include service activity status, benefit usage status, policy service completeness, account service activity, and outreach response status. Based on the condition item set corresponding to the candidate upgrade tiers, the platform extracts current status data from the target customer's data set and compares this data item by item with the baseline condition data, generating item gap data for each candidate upgrade tier. This item gap data is then aggregated according to the candidate upgrade tiers to generate the target tier gap. The target tier gap indicates which conditions the target customer still lacks to reach Platinum membership, such as insufficient benefit usage, incomplete value-added services, or insufficient service activity period.

[0098] The platform extracts historical behavioral characteristics, attribute characteristics, status change event markers, and upgrade acceptance capability indicators from the target object dataset. Based on these characteristics, it generates upgrade-related features. Historical behavioral characteristics may include customer manager outreach responses, benefit claim behavior, policy service access, protection consultation records, and service appointment records. Attribute characteristics may include stable membership level, benefit preference categories, and authorization preference tags. Status change event markers may include changes in family members, abnormal health checkup alerts, changes in protection needs, and changes in asset allocation stages. Upgrade acceptance capability indicators represent the target customer's willingness to accept higher-level service conditions, benefits, and service resources. Based on candidate upgrade levels, the platform combines the current object level, benefit sensitivity, the object level gap corresponding to the candidate upgrade level, and upgrade-related features into a candidate level prediction sample. This sample is then input into the prediction model to generate upgrade intention information corresponding to each candidate upgrade level. The platform sorts the candidate upgrade levels based on this information, generating a candidate upgrade level sequence. From this sequence, the platform determines the target upgrade level and extracts the corresponding upgrade intention information. If the upgrade intention information corresponding to a Platinum member is higher than that of a Diamond member, and the gap in target level corresponding to a Platinum member is easier to bridge, then the platform will identify the Platinum member as the target upgrade level and generate the target reach priority based on the upgrade intention information corresponding to the target upgrade level and the gap in target level corresponding to the target upgrade level.

[0099] The platform generates target profiles based on the target audience's data set, audience segmentation results, and sensitivity to benefits. It also retrieves the benefit tags, product tags, and level achievement conditions corresponding to the target upgrade level from the tiered system configuration data. The target profile includes membership level status, benefit preferences, service activity status, policy service status, authorization attribute behavior, and outreach response status. The platform generates benefit matching results based on the degree of matching between the target profile and benefit tags, and product matching results based on the degree of matching between the target profile and product tags. The platform generates level upgrade attractiveness based on the current target level and the target upgrade level, and generates benefit-product matching degree based on the benefit matching results, product matching results, and level upgrade attractiveness. If the target customer shows high interest in family health benefits and high-end medical check-up benefits, and the target upgrade level can unlock the corresponding benefits, the platform will include the corresponding benefits and matching products in the benefit recommendation results and product recommendation results. The platform generates benefit gap analysis results based on the target level gap and level achievement conditions, and generates upgrade path planning results based on the benefit gap analysis results. Upgrade path planning results can include conditions to be supplemented, suggested completion order, available benefit content, and level upgrade instructions. The platform then generates communication content based on the target person's profile, current target level, target upgrade level, benefit recommendation results, product recommendation results, benefit gap analysis results, and upgrade path planning results. This communication content can include key points from the agent's explanation to the customer regarding the benefits of the current membership level, the benefits of the target membership level, the level gap, and the completion path. It can also include scripts suitable for use in telephone, WeChat Work, or face-to-face meetings.

[0100] The platform generates strategy results based on benefit recommendation results, product recommendation results, benefit gap analysis results, upgrade path planning results, and communication content. It then generates terminal display data based on target audience reach priority and strategy results. This terminal display data is sent to the agent's terminal device for presentation. The agent's terminal device displays the target customer's target segmentation results, current target level, target upgrade level, target audience reach priority, recommended benefits, recommended products, benefit gap analysis, upgrade path planning, and communication content. After opening the terminal interface, the agent can see that the target customer is a high-potential member focused on benefits, currently at the Gold Card level, with a target upgrade level of Platinum. The main differences lie in several service activity conditions and benefit usage conditions. Recommended outreach content includes high-end health checkup benefits, family health benefits, and corresponding insurance product combinations. Based on the terminal display data, the agent explains the perceptible benefits after the membership level upgrade to the customer and guides the customer to complete the corresponding conditions according to the upgrade path, thereby transforming membership benefit resources into actionable strategies for customer outreach and membership upgrades.

[0101] In one embodiment, an object upgrade intention prediction and strategy generation apparatus is provided, which corresponds one-to-one with the object upgrade intention prediction and strategy generation method in the above embodiments. (Refer to...) Figure 3 , Figure 3 This is a schematic diagram of the functional modules of a preferred embodiment of the object upgrade intention prediction and strategy generation device of the present invention. The object upgrade intention prediction and strategy generation device includes: a historical sample modeling module 10, an object data fusion module 20, an object value clustering module 30, an upgrade gap analysis module 40, an upgrade intention prediction module 50, and a strategy presentation module 60. Detailed descriptions of each functional module are as follows: Historical sample modeling module 10 is used to acquire historical object sample data and generate analysis and prediction models based on the historical object sample data. The object data fusion module 20 is used to obtain the object identifier, object business data and authorized attribute behavior data of the target object, and generate a target object data set based on the object identifier, associating the object business data and the authorized attribute behavior data. The object value clustering module 30 is used to extract the current object level and object value data from the target object data set, generate equity sensitivity and object value vector based on the object value data, and process the object value vector based on the analysis model to generate object clustering results; The upgrade gap analysis module 40 is used to determine candidate upgrade levels and level achievement conditions based on the current object level, and to generate object level gaps based on the target object data set and the level achievement conditions. The upgrade intention prediction module 50 is used to extract upgrade-related features from the target object data set, process the current object level, the rights and interests sensitivity, the object level gap and the upgrade-related features based on the prediction model, generate upgrade intention information, and determine the target upgrade level and object reach priority from the candidate upgrade levels based on the upgrade intention information; The strategy presentation module 60 is used to generate a strategy result based on the target object data set, the object grouping result, the rights and interests sensitivity, the current object level, the target upgrade level, and the object level difference, and send the object reach priority and the strategy result to the terminal device for presentation.

[0102] In one embodiment, the historical sample modeling module 10 is specifically used for: Extract sample object identifiers, sample object business data, sample authorization attribute behavior data, historical rights interaction records, and historical level change records from historical object sample data. Based on the sample object identifiers, associate the sample object business data, the sample authorization attribute behavior data, the historical rights interaction records, and the historical level change records to generate a historical sample data set. Extract sample object level, sample value characteristics, sample rights usage records, and sample rights content attention records from the historical sample data set; generate sample rights sensitivity based on the sample rights usage records and the sample rights content attention records; and generate sample object value vector based on the sample object level, the sample value characteristics, and the sample rights sensitivity. Clustering training is performed based on the value vectors of the sample objects to generate an analytical model; Based on the historical level change records, an object upgrade marker is generated. Sample upgrade association features are extracted from the historical sample data set. Based on the sample object level, the sample candidate upgrade level and the level achievement conditions corresponding to the sample candidate upgrade level are determined from the preset level system. Extract current state data of samples from the historical sample data set, and generate sample object level gaps corresponding to the candidate upgrade levels of samples based on the level achievement conditions corresponding to the candidate upgrade levels of samples and the current state data of samples. A prediction model is generated by training based on the sample object level, the sample interest sensitivity, the sample object level gap corresponding to the sample candidate upgrade level, the sample upgrade association features, and the object upgrade label.

[0103] In one embodiment, the object data fusion module 20 is specifically used for: Obtain the data acquisition request of the target object, parse the object identifier and authorization credential from the data acquisition request, and generate an authorization verification result based on the authorization credential; When the authorization verification result is passed, the object business data is read based on the object identifier, and the authorization attribute behavior data is read based on the object identifier and the authorization credential. The object business data and the authorized attribute behavior data are written with unified field names, unified time stamps, and source stamps to generate object business data to be associated and authorized attribute behavior data to be associated. Extract the object identifier field from the business data of the object to be associated and the authorized attribute behavior data to be associated, perform consistency verification between the object identifier field and the object identifier, and generate the object identifier verification result; When the object identifier verification result is passed, associated object data is generated based on the object identifier, which is associated with the business data of the object to be associated and the authorized attribute behavior data to be associated. A data integrity tag is generated based on the associated object data, and a target object data set is generated based on the associated object data, the source tag in the business data of the object to be associated, the source tag in the authorized attribute behavior data to be associated, the time tag, and the data integrity tag.

[0104] In one embodiment, the object value clustering module 30 is specifically used for: Read the level field from the target object data set, generate the current object level, obtain the object value field mapping table, and extract interaction status data, interaction frequency data, value contribution data, rights usage records and rights content attention records from the target object data set based on the object value field mapping table to generate object value data; Based on the number of times, type, and duration of rights usage in the rights usage records of the object's value data, a rights usage intensity feature is generated. Based on the content browsing behavior, content response behavior, and content collection behavior in the rights and interests content attention records of the object value data, a rights and interests content attention intensity feature is generated; Based on the intensity characteristics of the use of rights and the intensity characteristics of attention to the content of rights, a rights sensitivity is generated; An object value vector is generated based on the current object level, the interaction status data, the interaction frequency data, the value contribution data, and the rights sensitivity. The object value vector is input into the analysis model, and the object value vector is clustered based on the cluster centers and cluster labels in the analysis model to generate object clustering results.

[0105] In one embodiment, the upgrade gap analysis module 40 is specifically used for: Read the level system configuration data and determine the current level sequence corresponding to the current object level from the level system configuration data; Based on the current level position, levels with higher level positions are selected from the level system configuration data to generate candidate upgrade levels; Extract the level achievement conditions corresponding to the candidate upgrade level from the level system configuration data, and break down the level achievement conditions into a set of condition items corresponding to the candidate upgrade level and condition benchmark data corresponding to the candidate upgrade level; Extract the current status data from the target object data set based on the set of conditional items corresponding to the candidate upgrade level; The current status data is compared item by item with the condition benchmark data corresponding to the candidate upgrade level to generate the item gap data corresponding to the candidate upgrade level. Based on the candidate upgrade levels, the project gap data corresponding to the candidate upgrade levels are summarized to generate the object level gap.

[0106] In one embodiment, the upgrade intention prediction module 50 is specifically used for: Extract historical behavior features, attribute features, state change event markers, and upgrade capability indicators from the target object dataset, and generate upgrade-related features based on the historical behavior features, attribute features, state change event markers, and upgrade capability indicators; Based on the candidate upgrade levels, the current object level, the rights and interests sensitivity, the object level gap corresponding to the candidate upgrade level, and the upgrade association features are combined into a candidate level prediction sample; The candidate level prediction samples are input into the prediction model to generate upgrade intention information corresponding to the candidate upgrade level; The candidate upgrade levels are sorted based on the upgrade intention information corresponding to the candidate upgrade levels to generate a candidate upgrade level sequence; Determine the target upgrade level from the candidate upgrade level sequence, and extract the upgrade intention information corresponding to the target upgrade level; Based on the upgrade intention information corresponding to the target upgrade level and the object level gap corresponding to the target upgrade level, an object reach priority is generated.

[0107] In one embodiment, the strategy presentation module 60 is specifically used for: Based on the target object data set, the object grouping results, and the rights and interests sensitivity, an object profile is generated, and the rights and interests tags, product tags, and level achievement conditions corresponding to the target upgrade level are read from the level system configuration data; Based on the degree of matching between the object profile and the benefit tag, a benefit matching result is generated, and based on the degree of matching between the object profile and the product tag, a product matching result is generated; The attractiveness of the level upgrade is generated based on the current object level and the target upgrade level, and the interest of the benefits and products is generated based on the benefit matching result, the product matching result and the attractiveness of the level upgrade. Based on the matching degree of the rights and benefits products, generate rights and benefits recommendation results and product recommendation results; Based on the object level gap and the level achievement conditions, an equity gap analysis result is generated, and based on the equity gap analysis result, an upgrade path planning result is generated; Communication content is generated based on the object profile, the current object level, the target upgrade level, the benefit recommendation results, the product recommendation results, the benefit gap analysis results, and the upgrade path planning results; Based on the benefit recommendation results, the product recommendation results, the benefit gap analysis results, the upgrade path planning results, and the communication content generation strategy results; Based on the object reach priority and the strategy result, terminal display data is generated and sent to the terminal device for presentation.

[0108] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 4 As shown. The computer device includes a processor, memory, network interface, and database connected via a system bus. The processor provides determination and control capabilities. The memory includes non-volatile and / or volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface is used to communicate with external clients via a network connection. When executed by the processor, the computer program implements the functions or steps of a server-side method for predicting object upgrade intentions and generating strategies.

[0109] In one embodiment, a computer device is provided, which may be a client, and its internal structure diagram may be as follows: Figure 5 As shown, the computer device includes a processor, memory, network interface, display screen, and input devices connected via a system bus. The processor provides determination and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface is used to communicate with an external server via a network connection. When executed by the processor, the computer program implements client-side functions or steps of an object upgrade intention prediction and strategy generation method.

[0110] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to perform the following steps: Acquire historical object sample data, and generate analysis and prediction models based on the historical object sample data; Obtain the object identifier, object business data, and authorized attribute behavior data of the target object, and generate a target object data set based on the object identifier, associating the object business data and the authorized attribute behavior data. Extract the current object level and object value data from the target object dataset, generate equity sensitivity and object value vectors based on the object value data, and process the object value vectors based on the analysis model to generate object clustering results; Based on the current object level, candidate upgrade levels and level achievement conditions are determined, and object level gaps are generated based on the target object data set and the level achievement conditions. Upgrade-related features are extracted from the target object dataset. Based on the prediction model, the current object level, the rights and interests sensitivity, the object level gap, and the upgrade-related features are processed to generate upgrade intention information. Based on the upgrade intention information, the target upgrade level and object reach priority are determined from the candidate upgrade levels. Based on the target object data set, the object grouping results, the rights and interests sensitivity, the current object level, the target upgrade level, and the object level gap, a strategy result is generated, and the object reach priority and the strategy result are sent to the terminal device for presentation.

[0111] In one embodiment, a computer-readable storage medium is provided, which may be non-volatile or volatile, and a computer program is stored thereon, which, when executed by a processor, performs the following steps: Acquire historical object sample data, and generate analysis and prediction models based on the historical object sample data; Obtain the object identifier, object business data, and authorized attribute behavior data of the target object, and generate a target object data set based on the object identifier, associating the object business data and the authorized attribute behavior data. Extract the current object level and object value data from the target object dataset, generate equity sensitivity and object value vectors based on the object value data, and process the object value vectors based on the analysis model to generate object clustering results; Based on the current object level, candidate upgrade levels and level achievement conditions are determined, and object level gaps are generated based on the target object data set and the level achievement conditions. Upgrade-related features are extracted from the target object dataset. Based on the prediction model, the current object level, the rights and interests sensitivity, the object level gap, and the upgrade-related features are processed to generate upgrade intention information. Based on the upgrade intention information, the target upgrade level and object reach priority are determined from the candidate upgrade levels. Based on the target object data set, the object grouping results, the rights and interests sensitivity, the current object level, the target upgrade level, and the object level gap, a strategy result is generated, and the object reach priority and the strategy result are sent to the terminal device for presentation.

[0112] It should be noted that the functions or steps that can be implemented by the computer-readable storage medium or computer device described above can be referred to the relevant descriptions on the server side and client side in the foregoing method embodiments. To avoid repetition, they will not be described one by one here.

[0113] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0114] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0115] It should be noted that any AI models, software tools, or components not belonging to this company appearing in the embodiments of this application are merely illustrative examples and do not represent actual use. All user personal information involved in the embodiments of this application has been authorized (with the knowledge and consent) by the relevant parties or has been fully authorized by all parties, and the executing entity may obtain it through various legal and compliant means. The collection, storage, use, processing, transmission, provision, and disclosure of the information, data, and signals involved all comply with relevant laws and regulations and do not violate public order and good morals.

[0116] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A method for predicting the upgrade willingness of an object and generating a strategy, characterized in that, Includes the following steps: Acquire historical object sample data, and generate analysis and prediction models based on the historical object sample data; Obtain the object identifier, object business data, and authorized attribute behavior data of the target object, and generate a target object data set based on the object identifier, associating the object business data and the authorized attribute behavior data. Extract the current object level and object value data from the target object dataset, generate equity sensitivity and object value vectors based on the object value data, and process the object value vectors based on the analysis model to generate object clustering results; Based on the current object level, candidate upgrade levels and level achievement conditions are determined, and object level gaps are generated based on the target object data set and the level achievement conditions. Upgrade-related features are extracted from the target object dataset. Based on the prediction model, the current object level, the rights and interests sensitivity, the object level gap, and the upgrade-related features are processed to generate upgrade intention information. Based on the upgrade intention information, the target upgrade level and object reach priority are determined from the candidate upgrade levels. Based on the target object data set, the object grouping results, the rights and interests sensitivity, the current object level, the target upgrade level, and the object level gap, a strategy result is generated, and the object reach priority and the strategy result are sent to the terminal device for presentation.

2. The method for predicting object upgrade intentions and generating strategies as described in claim 1, characterized in that, Acquire historical object sample data, and generate analysis and prediction models based on the historical object sample data, including: Extract sample object identifiers, sample object business data, sample authorization attribute behavior data, historical rights interaction records, and historical level change records from historical object sample data. Based on the sample object identifiers, associate the sample object business data, the sample authorization attribute behavior data, the historical rights interaction records, and the historical level change records to generate a historical sample data set. Extract sample object level, sample value characteristics, sample rights usage records, and sample rights content attention records from the historical sample data set; generate sample rights sensitivity based on the sample rights usage records and the sample rights content attention records; and generate sample object value vector based on the sample object level, the sample value characteristics, and the sample rights sensitivity. Clustering training is performed based on the value vectors of the sample objects to generate an analytical model; Based on the historical level change records, an object upgrade marker is generated. Sample upgrade association features are extracted from the historical sample data set. Based on the sample object level, the sample candidate upgrade level and the level achievement conditions corresponding to the sample candidate upgrade level are determined from the preset level system. Extract current state data of samples from the historical sample data set, and generate sample object level gaps corresponding to the candidate upgrade levels of samples based on the level achievement conditions corresponding to the candidate upgrade levels of samples and the current state data of samples. A prediction model is generated by training based on the sample object level, the sample interest sensitivity, the sample object level gap corresponding to the sample candidate upgrade level, the sample upgrade association features, and the object upgrade label.

3. The method for predicting object upgrade intentions and generating strategies as described in claim 1, characterized in that, Obtain the object identifier, object business data, and authorized attribute behavior data of the target object, and generate a target object data set based on the object identifier, associating the object business data and the authorized attribute behavior data, including: Obtain the data acquisition request of the target object, parse the object identifier and authorization credential from the data acquisition request, and generate an authorization verification result based on the authorization credential; When the authorization verification result is passed, the object business data is read based on the object identifier, and the authorization attribute behavior data is read based on the object identifier and the authorization credential. The object business data and the authorized attribute behavior data are written with unified field names, unified time stamps, and source stamps to generate object business data to be associated and authorized attribute behavior data to be associated. Extract the object identifier field from the business data of the object to be associated and the authorized attribute behavior data to be associated, perform consistency verification between the object identifier field and the object identifier, and generate the object identifier verification result; When the object identifier verification result is passed, associated object data is generated based on the object identifier, which is associated with the business data of the object to be associated and the authorized attribute behavior data to be associated. A data integrity tag is generated based on the associated object data, and a target object data set is generated based on the associated object data, the source tag in the business data of the object to be associated, the source tag in the authorized attribute behavior data to be associated, the time tag, and the data integrity tag.

4. The method for predicting object upgrade intentions and generating strategies as described in claim 1, characterized in that, Extract current object level and object value data from the target object dataset, generate equity sensitivity and object value vectors based on the object value data, and process the object value vectors based on the analysis model to generate object clustering results, including: Read the level field from the target object data set, generate the current object level, obtain the object value field mapping table, and extract interaction status data, interaction frequency data, value contribution data, rights usage records and rights content attention records from the target object data set based on the object value field mapping table to generate object value data; Based on the number of times, type, and duration of rights usage in the rights usage records of the object's value data, a rights usage intensity feature is generated. Based on the content browsing behavior, content response behavior, and content collection behavior in the rights and interests content attention records of the object value data, a rights and interests content attention intensity feature is generated; Based on the intensity characteristics of the use of rights and the intensity characteristics of attention to the content of rights, a rights sensitivity is generated; An object value vector is generated based on the current object level, the interaction status data, the interaction frequency data, the value contribution data, and the rights sensitivity. The object value vector is input into the analysis model, and the object value vector is clustered based on the cluster centers and cluster labels in the analysis model to generate object clustering results.

5. The method for predicting object upgrade intentions and generating strategies as described in claim 1, characterized in that, Based on the current object level, candidate upgrade levels and level achievement conditions are determined, and an object level gap is generated based on the target object dataset and the level achievement conditions, including: Read the level system configuration data and determine the current level sequence corresponding to the current object level from the level system configuration data; Based on the current level position, levels with higher level positions than the current level position are selected from the level system configuration data to generate candidate upgrade levels; Extract the level achievement conditions corresponding to the candidate upgrade level from the level system configuration data, and break down the level achievement conditions into a set of condition items corresponding to the candidate upgrade level and condition benchmark data corresponding to the candidate upgrade level; Extract the current status data from the target object data set based on the set of conditional items corresponding to the candidate upgrade level; The current status data is compared item by item with the condition benchmark data corresponding to the candidate upgrade level to generate the item gap data corresponding to the candidate upgrade level. Based on the candidate upgrade levels, the project gap data corresponding to the candidate upgrade levels are summarized to generate the object level gap.

6. The method for predicting object upgrade intentions and generating strategies as described in claim 1, characterized in that, Upgrade-related features are extracted from the target object dataset. Based on the prediction model, the current object level, the rights sensitivity, the object level gap, and the upgrade-related features are processed to generate upgrade intention information. Based on this upgrade intention information, the target upgrade level and object reach priority are determined from the candidate upgrade levels, including: Extract historical behavior features, attribute features, state change event markers, and upgrade capability indicators from the target object dataset, and generate upgrade-related features based on the historical behavior features, attribute features, state change event markers, and upgrade capability indicators; Based on the candidate upgrade levels, the current object level, the rights and interests sensitivity, the object level difference corresponding to the candidate upgrade level, and the upgrade association features are combined into a candidate level prediction sample; The candidate level prediction samples are input into the prediction model to generate upgrade intention information corresponding to the candidate upgrade level; The candidate upgrade levels are sorted based on the upgrade intention information corresponding to the candidate upgrade levels to generate a candidate upgrade level sequence; Determine the target upgrade level from the candidate upgrade level sequence, and extract the upgrade intention information corresponding to the target upgrade level; Based on the upgrade intention information corresponding to the target upgrade level and the object level gap corresponding to the target upgrade level, an object reach priority is generated.

7. The method for predicting object upgrade intentions and generating strategies as described in claim 1, characterized in that, Based on the target object dataset, the object segmentation results, the rights sensitivity, the current object level, the target upgrade level, and the object level gap, a strategy result is generated. The object reach priority and the strategy result are then sent to the terminal device for presentation, including: Based on the target object data set, the object grouping results, and the rights and interests sensitivity, an object profile is generated, and the rights and interests tags, product tags, and level achievement conditions corresponding to the target upgrade level are read from the level system configuration data; Based on the degree of matching between the object profile and the benefit tag, a benefit matching result is generated, and based on the degree of matching between the object profile and the product tag, a product matching result is generated; The attractiveness of the level upgrade is generated based on the current object level and the target upgrade level, and the interest of the rights and benefits is generated based on the rights and benefits matching results, the product matching results and the attractiveness of the level upgrade. Based on the matching degree of the rights and benefits products, generate rights and benefits recommendation results and product recommendation results; Based on the object level gap and the level achievement conditions, an equity gap analysis result is generated, and based on the equity gap analysis result, an upgrade path planning result is generated; Communication content is generated based on the object profile, the current object level, the target upgrade level, the benefit recommendation results, the product recommendation results, the benefit gap analysis results, and the upgrade path planning results; Based on the benefit recommendation results, the product recommendation results, the benefit gap analysis results, the upgrade path planning results, and the communication content generation strategy results; Based on the object reach priority and the strategy result, terminal display data is generated and sent to the terminal device for presentation.

8. A device for predicting object upgrade intentions and generating strategies, characterized in that, The object upgrade intention prediction and strategy generation device includes: The historical sample modeling module is used to acquire historical object sample data and generate analysis and prediction models based on the historical object sample data. The object data fusion module is used to obtain the object identifier, object business data, and authorized attribute behavior data of the target object, and generate a target object data set based on the object identifier, associating the object business data and the authorized attribute behavior data. The object value clustering module is used to extract the current object level and object value data from the target object data set, generate equity sensitivity and object value vectors based on the object value data, and process the object value vectors based on the analysis model to generate object clustering results; The upgrade gap analysis module is used to determine candidate upgrade levels and level achievement conditions based on the current object level, and to generate object level gaps based on the target object data set and the level achievement conditions. The upgrade intention prediction module is used to extract upgrade-related features from the target object data set, process the current object level, the rights and interests sensitivity, the object level gap and the upgrade-related features based on the prediction model, generate upgrade intention information, and determine the target upgrade level and object reach priority from the candidate upgrade levels based on the upgrade intention information; The strategy presentation module is used to generate strategy results based on the target object data set, the object grouping results, the rights and interests sensitivity, the current object level, the target upgrade level, and the object level gap, and send the object reach priority and the strategy results to the terminal device for presentation.

9. A computer device, characterized in that, The computer device includes a memory, a processor, and an object upgrade intention prediction and strategy generation program stored in the memory and executable on the processor. When the object upgrade intention prediction and strategy generation program is executed by the processor, it implements the steps of the object upgrade intention prediction and strategy generation method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The storage medium stores an object upgrade intention prediction and strategy generation program, which, when executed by a processor, implements the steps of the object upgrade intention prediction and strategy generation method as described in any one of claims 1-7.

Images