Group insurance product recommendation method and device, computer device, and storage medium

Abstract

The application relates to the technical field of artificial intelligence, and discloses a group insurance product recommendation method and device, computer equipment and a storage medium, which comprises the following steps: acquiring multi-source group information of a target group; acquiring historical service information of at least one historical salesperson having an association relationship with the target group based on the multi-source group information; performing graph construction based on the multi-source group information and the historical service information to obtain an original relationship graph; performing link query on the original relationship graph to obtain a service relationship link, which is used for representing a relationship link between the target group and each historical salesperson; performing score calculation on the service relationship link to obtain an intimacy score of the service relationship link; screening the historical salesperson based on the intimacy score to obtain a target salesperson, and recommending an insurance product to the target group through the target salesperson. The application can be applied to the group insurance product recommendation scene of financial technology, and improves the accuracy and business conversion rate of group insurance recommendation.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and is applicable to the financial technology field, particularly to a method, apparatus, computer equipment, and storage medium for recommending group insurance products. Background Technology

[0002] In group insurance business, customer allocation typically employs a strategy of prioritizing historical affiliations and randomly assigning new customers. For existing customers, they are automatically assigned to their original sales representatives based on their past transaction records to improve conversion efficiency. For new customers, the system randomly assigns a sales representative. However, in real-world scenarios, new customers may have potential connections with existing historical customers. Sales representatives often rely solely on personal experience when reaching out to new customers, lacking a systematic understanding of these connections. This makes it difficult to accurately grasp customer needs, easily leading to resource misallocation and resulting in low conversion rates for group insurance recommendations, impacting the accuracy of group insurance recommendations and overall business conversion rates. Summary of the Invention

[0003] This application provides a method, apparatus, computer equipment, and storage medium for recommending group insurance products, in order to solve the technical problem that the lack of identification of relationships in the allocation of new customers makes it difficult for salespersons to accurately grasp customer needs, thereby affecting the accuracy and conversion efficiency of group insurance recommendations.

[0004] Firstly, a method for recommending group insurance products is provided, including: Obtain multi-source group information for the target group; Based on the multi-source group information, obtain the historical service information of at least one historical salesperson who is associated with the target group; Based on the multi-source community information and the historical service information, a graph is constructed to obtain the original relationship graph; A link query is performed on the original relationship graph to obtain service relationship links, wherein the service relationship links are used to characterize the relationship links between the target group and each of the historical salespersons; A score is calculated on the service relationship links to obtain the intimacy score of the service relationship links; Based on the intimacy score, the historical sales agents are screened to obtain target sales agents, and insurance products are recommended to the target group through the target sales agents.

[0005] Secondly, a group insurance product recommendation device is provided, comprising: The group information acquisition module is used to acquire multi-source group information of the target group; The salesperson information acquisition module is used to acquire historical service information of at least one historical salesperson who is associated with the target group based on the multi-source group information. The graph construction module is used to construct a graph based on the multi-source community information and the historical service information to obtain an original relationship graph. The link query module is used to perform link query on the original relationship graph to obtain service relationship links, wherein the service relationship links are used to characterize the relationship links between the target group and each of the historical salespersons; The intimacy score calculation module is used to calculate the intimacy score of the service relationship link. The product recommendation module is used to filter the historical salespersons based on the intimacy score, obtain target salespersons, and recommend insurance products to the target group through the target salespersons.

[0006] Thirdly, 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 implement the steps of the aforementioned group insurance product recommendation method.

[0007] Fourthly, a computer-readable storage medium is provided, which stores a computer program that, when executed by a processor, implements the steps of the aforementioned group insurance product recommendation method.

[0008] The aforementioned group insurance product recommendation method, device, computer equipment, and storage medium can be applied to group insurance product recommendation scenarios in fintech. First, it acquires multi-source group information about the target group, providing a rich data foundation for subsequent relationship mining and avoiding relationship omissions caused by a single data source. Based on the multi-source group information, it obtains historical service information of at least one historical agent associated with the target group, ensuring that candidate agent selection has clear data basis rather than random allocation. Next, it constructs a graph based on the multi-source group information and historical service information, obtaining an original relationship graph. This provides a computable graph structure for link queries, overcoming the limitation of traditional tabular data in expressing relationships. Link queries are then performed on the original relationship graph to obtain service relationship links. These service relationship links characterize the relationship links between the target group and each historical agent, making the connection path between agents and groups clear and traceable, providing a clear chain of evidence for subsequent scoring. Furthermore, a score is calculated on the service relationship links to obtain a closeness score, transforming the relationship into a quantifiable numerical indicator and solving the problem of lack of standards in subjective judgment. Finally, based on the intimacy score, historical agents are screened to obtain target agents, and insurance products are recommended to target groups through these target agents. This allows agents to use specific information in the relationship chain (such as co-ownership, historical insurance policies, etc.) as a link to build connections, enhance customer trust, and solve the problem that agents cannot accurately grasp customer needs due to the lack of identification of related relationships in the allocation of new customers. This helps to improve the accuracy and conversion efficiency of group insurance recommendations. Attached Figure Description

[0009] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments of this application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 This is a schematic diagram of an application environment for a group insurance product recommendation method in one embodiment of this application; Figure 2 This is a flowchart illustrating a group insurance product recommendation method in one embodiment of this application; Figure 3 yes Figure 2 A schematic diagram of a specific implementation method for step S30; Figure 4 yes Figure 2 A schematic diagram of a specific implementation of step S40; Figure 5 yes Figure 2A schematic diagram of a specific implementation method for step S50; Figure 6 yes Figure 5 A schematic diagram of a specific implementation method for step S51; Figure 7 yes Figure 5 A flowchart illustrating a specific implementation of step S53; Figure 8 yes Figure 2 A schematic diagram of a specific implementation method for step S60; Figure 9 This is a schematic diagram of a group insurance product recommendation device in one embodiment of this application; Figure 10 This is a schematic diagram of the structure of a computer device according to one embodiment of this application. Detailed Implementation

[0011] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0012] First, let's analyze some of the terms used in this application: Artificial intelligence (AI) is a new branch of computer science that studies, develops, and applies theories, methods, technologies, and systems to simulate, extend, and expand human intelligence. It aims to understand the essence of intelligence and produce intelligent machines that can react in a way similar to human intelligence. Research in this field includes robotics, speech recognition, image recognition, natural language processing, and expert systems. AI can simulate the information processes of human consciousness and thought. Furthermore, AI utilizes digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceiving the environment, acquiring knowledge, and using that knowledge to achieve optimal results.

[0013] Natural Language Processing (NLP): NLP uses computers to process, understand, and utilize human language (such as Chinese and English). It is a branch of artificial intelligence and an interdisciplinary field of computer science and linguistics, often referred to as computational linguistics. NLP includes syntactic analysis, semantic analysis, and discourse understanding. It is commonly used in machine translation, handwritten and printed character recognition, speech recognition and text-to-speech conversion, intent recognition, information extraction and filtering, text classification and clustering, sentiment analysis, and opinion mining. It involves data mining, machine learning, knowledge acquisition, knowledge engineering, artificial intelligence research, and linguistic research related to language computation.

[0014] Information extraction is a text processing technique that extracts factual information such as entities, relationships, and events from natural language text and outputs it as structured data. Information extraction is a technique for extracting specific information from text data. Text data is composed of specific units, such as sentences, paragraphs, and chapters. Text information is composed of smaller, specific units, such as characters, words, phrases, sentences, paragraphs, or combinations of these units. Extracting noun phrases, names of people, and place names from text data is an example of text information extraction. Of course, text information extraction techniques can extract information of various types.

[0015] Group insurance is a corporate insurance business offered by insurance companies to legal entities or non-legal entities (such as various enterprises, government agencies, social organizations, etc.). Essentially, group insurance is an underwriting method where a specific group acts as the policyholder, purchasing insurance from an insurance company for its members (such as employees). The insurance company then provides coverage for all members under a single master insurance contract.

[0016] In group insurance business, customer allocation typically employs a strategy of prioritizing historical affiliations and randomly assigning new customers. For existing customers, they are automatically assigned to their original sales representatives based on their past transaction records to improve conversion efficiency. For new customers, the system randomly assigns a sales representative. However, in real-world scenarios, new customers may have potential connections with existing historical customers. Sales representatives often rely solely on personal experience when reaching out to new customers, lacking a systematic understanding of these connections. This makes it difficult to accurately grasp customer needs, easily leading to resource misallocation and resulting in low conversion rates for group insurance recommendations, impacting the accuracy of group insurance recommendations and overall business conversion rates.

[0017] Based on this, embodiments of this application provide a method, apparatus, computer equipment, and storage medium for recommending group insurance products, in order to solve the technical problem that the lack of identification of relationships in the allocation of new customers makes it difficult for salespersons to accurately grasp customer needs, thereby affecting the accuracy and conversion efficiency of group insurance recommendations.

[0018] The group insurance product recommendation method, apparatus, computer equipment, and storage medium provided in this application are specifically described through the following embodiments. First, the group insurance product recommendation method in this application embodiment is described.

[0019] The group insurance product recommendation method provided in this application embodiment 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 multi-source group information of the target group from the client; based on the multi-source group information, it can obtain historical service information of at least one historical agent who is related to the target group; based on the multi-source group information and historical service information, it can construct a graph to obtain an original relationship graph; it can perform link queries on the original relationship graph to obtain service relationship links, where service relationship links are used to represent the relationship links between the target group and each historical agent; it can calculate a score for the service relationship links to obtain a closeness score; based on the closeness score, it can filter historical agents to obtain target agents, feed back the target agents to the client, and assign the group insurance recommendation task to the target agents so that the target agents can recommend insurance products to the target group.

[0020] This application, applicable to group insurance product recommendation scenarios in fintech, firstly acquires multi-source group information of the target group, providing a rich data foundation for subsequent relationship mining and avoiding relationship omissions caused by a single data source. Based on the multi-source group information, it obtains historical service information of at least one historical agent associated with the target group, ensuring that candidate agent selection has clear data basis rather than random allocation. Next, a graph is constructed based on the multi-source group information and historical service information to obtain an original relationship graph, providing a computable graph structure for link queries and overcoming the limitation of traditional tabular data in expressing relationships. Link queries are then performed on the original relationship graph to obtain service relationship links. These service relationship links characterize the relationship links between the target group and each historical agent, making the connection path between agents and groups clear and traceable, providing a clear chain of evidence for subsequent scoring. Furthermore, a score is calculated on the service relationship links to obtain a closeness score, transforming the relationship into a quantifiable numerical indicator and solving the problem of lack of standards in subjective judgment. Finally, based on the intimacy score, historical agents are screened to obtain target agents, and insurance products are recommended to target groups through these target agents. This allows agents to use specific information in the relationship chain (such as co-ownership, historical insurance policies, etc.) as a link to build connections, enhance customer trust, and solve the problem that agents cannot accurately grasp customer needs due to the lack of identification of related relationships in the allocation of new customers. This helps to improve the accuracy and conversion efficiency of group insurance recommendations.

[0021] 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 following detailed description uses specific embodiments to illustrate this application.

[0022] Please see Figure 2 As shown, Figure 2 A flowchart illustrating the group insurance product recommendation method provided in this application embodiment includes the following steps: S10. Obtain multi-source group information of the target group; It is important to understand that the target group refers to group clients who require insurance product recommendations, such as enterprises, institutions, or other legal entities. The multi-source group information for the target group includes, but is not limited to: the target group's business registration information (such as company name, unified social credit code, registered capital, establishment date, business scope, etc.), key natural person information of the target group (such as legal representative, directors, supervisors, senior management, shareholders, etc.), key contact information of the target group (such as financial officer, administrative officer, etc.), and the target group's historical policy information, inquiry and quotation records, etc. The multi-source group information can be obtained by the client from various data sources, including enterprise registration databases, insurance companies' internal core business systems (such as policy management systems and customer relationship management systems), external credit data platforms, and internal policy systems. It is important to emphasize that the multi-source group information for the target group is obtained through legal and compliant channels, adhering to relevant laws, regulations, and standards.

[0023] Understandably, by acquiring multi-source group information of the target group, the comprehensiveness of information acquisition can be improved, avoiding the problems of incomplete information or missing relationships caused by a single data source, and providing a data foundation for subsequent relationship mining and accurate matching.

[0024] S20. Obtain historical service information of at least one historical salesperson who is associated with the target group based on multi-source group information; Specifically, based on multi-source group information, the client queries the insurance company's internal business systems (such as policy systems, CRM systems, and sales opportunity management systems) for all historical agents with various relationships to the target group, and extracts these agents' historical service records and profile data. Here, "relationship" refers to past business contact, and relationships may include, but are not limited to: The salesperson has directly provided services to the target group (such as underwriting, quotation inquiry, business visits, etc.). The salesperson has provided services directly to the target group's holding company / parent and subsidiary companies, as well as to key natural persons of the target group (such as legal representatives, directors, supervisors, senior managers, shareholders, etc.); The salesperson has provided services to key natural persons of the target group's holding company / parent and subsidiary companies, as well as to the family members of key natural persons of the target group.

[0025] Among them, the salespersons with the aforementioned relationship are the historical salespersons corresponding to the target group.

[0026] After identifying the historical salesperson, obtain the historical service information of that salesperson and the target group / group with which they are related. The historical service information includes, but is not limited to, the salesperson's basic information (name, employee number, affiliated organization), historical performance data, and service records (order records, inquiry processing records, business opportunity follow-up records, visit records, etc.).

[0027] Understandably, the process involves filtering out at least one historical salesperson who is associated with the target group based on multi-source group information and obtaining the historical service information of the historical salesperson. This provides a data basis for subsequent selection of salespersons to connect with the target group, rather than adopting the traditional random assignment strategy. This avoids mismatch between the needs of the salesperson and the group customers, which would affect the efficiency of the connection and the insurance referral conversion rate.

[0028] In one embodiment, if a historical agent with a relationship cannot be found in the insurance's internal business system based on multi-source group information, it indicates that the target group is a completely new customer. In this case, the agent allocation strategy can be adopted by random or workload-based assignment according to geographical area (e.g., by administrative division), industry attribute (e.g., manufacturing specialty), or polling mechanism. The agent allocation mechanism based on affinity score adopted in the embodiments of this application is not applicable.

[0029] S30. Construct a graph based on multi-source community information and historical service information to obtain the original relationship graph; It is important to understand that the graph construction here refers to building a heterogeneous relationship graph based on node information and relationships between nodes from multiple sources of group information and historical service information. This transforms scattered and isolated data into structured graph data, enabling the relationship between the target group and historical salespersons to be clearly expressed, thus providing a data foundation for subsequent salesperson screening.

[0030] Among them, such as Figure 3 As shown, step S30, which involves constructing a graph based on multi-source community information and historical service information to obtain the original relationship graph, includes the following steps: S31. Perform information cleaning on multi-source group information to obtain target group information; S32. Clean the historical service information to obtain the target service information; First, the multi-source group information is cleaned (including inspection, verification, filtering, and standardization) to remove noisy data that is irrelevant to the business or has low credibility, correct format errors, fill in missing key fields, and unify data coding standards to obtain the target group information. This process eliminates noisy and low-confidence data that is not directly related to insurance business at the source, effectively avoiding misjudgments and resource misallocations caused by incorrect associations in subsequent graph construction. It significantly improves the purity and credibility of the data, laying a reliable data foundation for building an accurate relationship graph.

[0031] Specifically, information cleaning of multi-source group information may include, but is not limited to: removing contact information from financing-related insurance policies (such as insurance attached to loan business), filtering out obviously non-compliant company names or unified social credit codes (such as incorrect length or check digit errors), standardizing and deduplicating the format of company contact phone numbers, and marking suspected non-salesperson or non-directly connected numbers (such as customer service hotlines or switchboard transfer numbers) based on the internal number database, and disambiguating shareholder and director / supervisor information to distinguish between different people with the same name.

[0032] Next, the historical service information is cleaned to remove noise data that is irrelevant to the business or has low credibility, so as to obtain the target service information. Each service record is real, valid, timely and can be accurately attributed to a specific salesperson. This eliminates information bias caused by personnel turnover, system redundancy and incomplete records, and provides high-quality and traceable historical interaction data for subsequent relationship extraction, thereby ensuring the authenticity and reliability of the relationship between salesperson and customer.

[0033] Specifically, clarifying historical service information may include, but is not limited to: removing invalid policy records that have been cancelled or generated in the testing environment; marking the succession relationship for historical service records of departing agents, finding and supplementing the successor agent information through system records; verifying the timeliness of inquiry and quotation records and business opportunity records, filtering out outdated records that have exceeded a set time window (such as more than two years) and have no subsequent conversion; deduplicating and merging duplicate service records of the same agent for the same customer, retaining the latest or most complete record version; and supplementing records with missing agent numbers through system logs or work order logs.

[0034] By performing information cleaning on multi-source group information and historical service information through the above steps S31 and S32, target group information and target service information are obtained, which can improve the purity and credibility of the data and lay a reliable data foundation for constructing an accurate relationship graph.

[0035] S33. Extract entities from target group information and target service information to obtain graph node data; Specifically, by extracting entities from the target group information and target service information after information cleaning, key objects with independent business meanings that can serve as graph nodes are identified and extracted, resulting in graph node data. The graph node data includes, but is not limited to: company names, key natural person information (such as legal representatives, directors, supervisors, senior managers, shareholders, etc.), family members / friends of key natural persons, and names of historical salespersons. This transforms information scattered in heterogeneous data sources into unified and standardized graph nodes, which are used as graph nodes for constructing relationship graphs.

[0036] S34. Extract the relationship between the target group information and the target service information to obtain the graph node relationship data; Specifically, based on the already defined graph node data, business connections between nodes are identified and extracted from target group information and target service information, and the relationship type is labeled for each connection to obtain graph node relationship data. Among them, graph node relationship data includes, but is not limited to: holding, parent and subsidiary companies, shareholding, employment, insurance services, etc., which provides a data foundation for subsequent construction of relationship graph.

[0037] S35. Construct a graph based on graph node data and graph node relationship data to obtain the original relationship graph.

[0038] Specifically, after extracting the graph node data and graph node relationship data, the data is structured and organized to clearly present the multidimensional connections between entities in a graph structure, forming a traversable and queryable heterogeneous relationship network, which is the original relationship graph. This original relationship graph serves as the foundational data for subsequent relationship chain retrieval, path search, and affinity calculation.

[0039] It should be noted that the original relationship graph can be either a directed graph or an undirected graph, depending on the actual business needs. However, this does not affect subsequent link queries and affinity score calculations based on the original relationship graph.

[0040] S40. Perform a link query on the original relationship graph to obtain the service relationship links, where the service relationship links are used to represent the relationship links between the target group and each historical salesperson. It's important to understand that link querying here refers to using a preset search algorithm (such as Breadth-First Search (BFS) or Depth-First Search (DFS)) to traverse the original relationship graph starting from the target group, extracting all paths whose endpoints are historical salesperson nodes, and using these as service relationship links. For example, Figure 4 As shown, step S40, which involves performing a link query on the original relationship graph to obtain the service relationship links, includes the following steps: S41. Using the target group as the starting point of the link and each historical salesperson as the ending point of the link, perform a path search on the original relationship graph to obtain multiple original relationship links; S42. Based on a preset link length threshold, perform link filtering on multiple original relationship links to obtain service relationship links.

[0041] Specifically, using a pre-defined search algorithm (such as Breadth-First Search (BFS) or Depth-First Search (DFS), the algorithm traverses the original relationship graph, starting with the target group and ending with each historical salesperson. All reachable paths obtained are considered original relationship links. There must be at least one original relationship link from the target group to a specific historical salesperson, and no path length limit is set during the path search process, aiming to discover as many potential connections as possible.

[0042] Understandably, the number of nodes in each original relationship link varies. It might have only two nodes (target group, historical salesperson, link length 1); three nodes (target group, intermediate node, historical salesperson, link length 2); four nodes (target group, intermediate node, intermediate node, historical salesperson, link length 3); or five nodes (target group, intermediate node, intermediate node, intermediate node, historical salesperson, link length 4), and so on. However, if the original relationship link is too long, it indicates a weak connection between the target group and the corresponding historical salesperson, and this original relationship path can be disregarded.

[0043] Therefore, a link length threshold can be preset. This threshold is used to limit the effective range of relationship links. If the length of the original relationship link exceeds this range, it will be rejected. Specifically, the link length threshold can be set to 2, 3, etc., depending on the actual business scenario. No specific limit is set here.

[0044] Furthermore, based on a preset link length threshold, multiple original relationship links are filtered out. Links whose link lengths do not meet the requirements are removed from all original relationship links, and the remaining links are service relationship links. This avoids including relationship links with little effect in the subsequent intimacy score calculation and salesperson screening process, thereby improving the interpretability and feasibility of salesperson screening.

[0045] S50. Calculate the intimacy score of the service relationship link. It is important to understand that the score calculation here refers to calculating a quantitative score for each service relationship link according to the preset weight mapping rules, and obtaining an affinity score, thereby quantifying the strength of the connection between the target group and a certain historical salesperson. The higher the score, the closer and more direct the connection between the two.

[0046] Among them, such as Figure 5 As shown, step S50, which involves calculating the score of the service relationship link to obtain the intimacy score of the service relationship link, includes the following steps: S51. Classify groups based on multi-source group information to obtain group type data of the target group; It's important to understand that group classification here refers to categorizing target groups into specific customer types according to pre-defined classification rules, providing a basis for subsequent differentiated weighting and recommendation strategies. Among them, such as Figure 6 As shown, step S51, which involves classifying groups based on multi-source group information to obtain group type data for the target group, includes the following steps: S511. Extract information from multi-source group information to obtain group insurance records and group consultation records; S512. Classify groups based on group underwriting records and group consultation records to obtain group type data for the target groups.

[0047] Specifically, the key content related to historical policy information and inquiry and quotation records is first extracted from multi-source group information to obtain the group underwriting records and group consultation records of the target group, providing a data foundation for subsequent classification.

[0048] Group underwriting records include insurance contracts between the target group as the policyholder or insured and the insurance company. In group insurance business, underwriting records typically include: policy number, type of insurance (such as property insurance, liability insurance, accident insurance, etc.), sum insured, premium amount, policy application date, policy start date, policy termination date, current policy status (insured, lapsed, invalid), historical agent number, number of renewals, etc. Underwriting records are the direct basis for determining whether a customer is a "historical customer" and for further categorizing their "insured" or "lapsed" status.

[0049] Group consultation records contain information about interactions between the target group and the insurance company before a final sale, even if the deal didn't go through. In group insurance scenarios, consultation records typically include: inquiry and quotation records (inquiry time, insurance type requirements, budget range), lead records (lead source, follow-up stage, estimated probability of closing the deal), and salesperson visit records (visit time, visitor, visit minutes), etc. Consultation records are a key basis for determining whether a customer is a "historical potential customer."

[0050] After obtaining group underwriting records and group consultation records, groups are classified based on these records to obtain group type data for the target groups. This group type data includes: If there are past underwriting records in the group insurance records of the target group, then the target group is determined to be a historical customer. Historical potential customers: If the target group has no past insurance records in its group insurance records, but has past consultation records in its group consultation records (quotation inquiry records, business opportunity lead records, salesperson visit records), then the target group is determined to be a historical potential customer. If a target group is considered a new customer, and there are no previous underwriting records in the group's group underwriting records and no previous consultation records in the group's consultation records, then the target group is considered a new customer.

[0051] It should be noted that step S50 is executed after step S20. In step S20, target groups with historical salespersons whose relationships cannot be found in the insurance company's internal business system have been filtered out. Therefore, "new customers" in step S512 refer to target groups that have not purchased or consulted about insurance themselves, but whose key natural persons, family members, holding companies / parent and subsidiary companies, and key natural persons of holding companies / parent and subsidiary companies have underwriting / quotation records with insurance companies.

[0052] Through the above steps S511 to S512, accurate classification of target group types is achieved, providing a fine-grained type index for the precise selection of subsequent weight mapping tables.

[0053] S52. Obtain a preset relationship weight mapping table based on group type data; It should be noted that the relationship weight mapping table is set up in advance according to the actual business scenario, and a corresponding relationship weight mapping table is set up for each group type of data.

[0054] For example: Suppose the preset link length threshold is 3, For each group type of data, a relation weight mapping table is set up. This relation weight mapping table is a major category, which contains the following subcategories: Subclass 1: The link length is 1. The target group has been the policyholder or insured, and there are historical policies and historical agents. The corresponding weight value is assumed to be A. Subcategory 2: The link length is 1, and the target group has a history of having a salesperson who has acted as the inquiry processor, business opportunity follow-up person, or first visit salesperson. The corresponding weight value is assumed to be B. Subcategory 3: The link length is 2, and it is related through corporate equity. The target group's holding company, parent and subsidiary companies, and joint controlling shareholders (legal persons / shareholders) have historical salespersons. The corresponding weight value is assumed to be C. Subcategory 4: The link length is 2, and the association is through contact sharing. The target group and the associated company share the same contact number, or the target group has taken out insurance for the associated company as the policyholder. There are historical salespersons. The corresponding weight value is assumed to be D. Subcategory 5: The link length is 2, which is associated through the personal insurance policies of key natural persons. The legal representative, directors, supervisors, senior managers and shareholders of the target group have purchased personal insurance policies and there are historical agents. The corresponding weight value is assumed to be E. Subcategory 6: The link length is 2. It is associated through departure and takes over. After the historical agent of the target group leaves, the historical agent takes over the historical insurance policy. The corresponding weight value is assumed to be F. Subcategory 7: The link length is 3. It is associated through hierarchical extension. The legal representatives, directors, supervisors, senior managers, shareholders and their families of the target group's holding company, parent and subsidiary companies have purchased personal insurance policies and there are historical salespersons. The corresponding weight value is assumed to be G.

[0055] The weight values ​​for major categories can be obtained by averaging the weight values ​​A, B, C, D, E, F, and G.

[0056] It should be noted that the weight values ​​A, B, C, D, E, F, and G are different in the relational weight mapping table for the same group type of data, and are set by the insurance company according to the actual business scenario; and the weight values ​​A, B, C, D, E, F, and G are also different in the relational weight mapping tables for different group types of data, which is not limited here.

[0057] For step S52 above, the corresponding relationship weight mapping table is obtained according to the group type data, which enables on-demand loading and dynamic adaptation of weight configuration to meet the needs of different types of groups.

[0058] S53. Evaluate the service relationship links based on the relationship weight mapping table to obtain the intimacy score of the service relationship links.

[0059] It is important to understand that the score evaluation here refers to quantifying and scoring each service relationship link based on the corresponding weight value in the relationship weight mapping table, thereby obtaining the intimacy score corresponding to each service relationship link. This transforms the abstract relationship type into a comparable numerical score, achieving standardization and calculability of intimacy evaluation.

[0060] Among them, such as Figure 7 As shown, step S53, which involves evaluating the service relationship links based on the relationship weight mapping table to obtain the intimacy score of the service relationship links, includes the following steps: S531. Obtain the type weight data corresponding to the group type data; S532. Obtain subclass weight data by querying the relationship weight mapping table based on the service relationship link; S533. Aggregate the type weight data and subclass weight data to obtain the intimacy score of the service relationship link.

[0061] It should be noted that the type weight data here refers to the category weight of the relationship weight mapping corresponding to the group type data. The type weight data is used to characterize the absolute importance of the relationship link in the overall strategy.

[0062] Next, the corresponding subclass is matched from the relationship weight mapping table according to the service relationship link, and the weight value of the subclass is obtained as the subclass weight data. The subclass weight data is used to characterize the relative importance of the relationship link within the larger class.

[0063] After obtaining the type weight data and subclass weight data corresponding to the service relationship link, the type weight data and subclass weight data are multiplied to obtain the intimacy score of the service relationship link. The intimacy score is used to characterize the global contribution weight of the relationship link in the entire intimacy evaluation system, reflecting the strength of the connection between the salesperson and the target group embodied in the service relationship link. The higher the score, the more recommendable the link is.

[0064] S60. Screen historical sales agents based on intimacy scores to obtain target sales agents, and recommend insurance products to target groups through target sales agents.

[0065] It is important to understand that the screening here refers to selecting one or more of the best salespeople from the candidate historical salespeople based on the ranking results of the intimacy score, as the target salesperson.

[0066] Among them, such as Figure 8 As shown, step S60, which involves filtering historical salespeople based on affinity scores to obtain target salespeople, includes the following steps: S61. For each historical salesperson, obtain all service relationship links related to the historical salesperson, and sum the intimacy scores of all service relationship links to obtain the total intimacy score of the historical salesperson. It should be noted that there must be at least one original relationship link between the target group and a certain historical salesperson. If a certain historical salesperson has multiple service relationship links, the intimacy scores of all service relationship links are summed to obtain the total intimacy score of the historical salesperson. If a certain historical salesperson has only one service relationship link, the intimacy score of that service relationship link is used as the total intimacy score of the historical salesperson.

[0067] Understandably, since a single service relationship link can only reflect a single-dimensional connection, if there are multiple service relationship links between a historical salesperson and the target group, it indicates that the connection between the historical salesperson and the target group is stronger. By summing the intimacy scores of multiple service relationship links, multi-source relationship clues can be fully integrated, and multi-dimensional and multi-path connection information can be condensed into a total score, overcoming the limitations of single-path evaluation and helping to improve the comprehensiveness and robustness of intimacy evaluation.

[0068] S62. Sort the historical salespersons based on the total affinity score to obtain the salesperson sequence; S63. Filter the salesperson sequence to obtain the target salesperson.

[0069] After calculating the total affinity score for each historical salesperson, all historical salespersons are organized into an ordered list according to their total affinity scores in descending order. This salesperson sequence directly reflects the affinity between historical salespersons and the target group, which helps improve the efficiency of subsequent screening decisions.

[0070] It should be noted that the number of target salespersons is usually one, but multiple target salespersons can also be set according to the actual business scenario to serve the target group together. Based on the required number of target salespersons, the one or several historical salespersons with the highest total affinity score are selected from the salesperson sequence and designated as the target salesperson. This target salesperson will be identified as the salesperson serving the target customer.

[0071] After identifying the target agents, the list of target agents is fed back to the client so that the client can assign the service tasks for the target group to the target agents. The target agents then recommend insurance products and provide services to the target group. On the one hand, the agents with the highest affinity scores have the strongest connection with the target group, and their recommendation conversion efficiency and success rate are higher than those of random assignment. On the other hand, agents can use specific information in the service relationship chain (such as co-ownership, historical policies, personal policies, etc.) as a link to effectively reduce the difficulty of cold calling, enhance customer trust, and thus improve the overall operational efficiency and scalability of group customer business.

[0072] As can be seen, the above solution can be applied to group insurance product recommendation scenarios in fintech. First, it acquires multi-source group information about the target group, providing a rich data foundation for subsequent relationship mining and avoiding relationship omissions caused by a single data source. Based on the multi-source group information, it obtains historical service information of at least one historical agent associated with the target group, ensuring that candidate agent selection has clear data basis rather than random allocation. Next, it constructs a graph based on the multi-source group information and historical service information, obtaining an original relationship graph. This provides a computable graph structure for link queries, overcoming the limitation of traditional tabular data in expressing relationships. Link queries are then performed on the original relationship graph to obtain service relationship links. These service relationship links characterize the relationship links between the target group and each historical agent, making the connection path between agents and groups clear and traceable, providing a clear chain of evidence for subsequent scoring. Furthermore, a score is calculated on the service relationship links to obtain a closeness score, transforming the relationship into a quantifiable numerical indicator and solving the problem of lack of standards in subjective judgment. Finally, based on the intimacy score, historical agents are screened to obtain target agents, and insurance products are recommended to target groups through these target agents. This allows agents to use specific information in the relationship chain (such as co-ownership, historical insurance policies, etc.) as a link to build connections, enhance customer trust, and solve the problem that agents cannot accurately grasp customer needs due to the lack of identification of related relationships in the allocation of new customers. This helps to improve the accuracy and conversion efficiency of group insurance recommendations.

[0073] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0074] In one embodiment, a group insurance product recommendation device is provided, which corresponds one-to-one with the group insurance product recommendation method described in the above embodiments. For example... Figure 9 As shown, the group insurance product recommendation device includes a group information acquisition module 101, a salesperson information acquisition module 102, a graph construction module 103, a link query module 104, a closeness score calculation module 105, and a product recommendation module 106. Detailed descriptions of each functional module are as follows: The group information acquisition module 101 is used to acquire multi-source group information of the target group; The salesperson information acquisition module 102 is used to acquire historical service information of at least one historical salesperson who is associated with the target group based on multi-source group information. The graph construction module 103 is used to construct a graph based on multi-source community information and historical service information to obtain an original relationship graph. Link query module 104 is used to perform link query on the original relationship graph to obtain service relationship links, wherein the service relationship links are used to represent the relationship links between the target group and each historical salesperson. The intimacy score calculation module 105 is used to calculate the intimacy score of the service relationship link. The product recommendation module 106 is used to filter historical sales agents based on intimacy scores, obtain target sales agents, and recommend insurance products to target groups through target sales agents.

[0075] In one embodiment, the map construction module 103 is specifically used for: Information cleaning is performed on multi-source group information to obtain target group information; Historical service information is cleaned to obtain target service information; Entity extraction is performed on target group information and target service information to obtain graph node data; Relationships are extracted from target group information and target service information to obtain graph node relationship data; The graph is constructed based on the graph node data and graph node relationship data to obtain the original relationship graph.

[0076] In one embodiment, the link query module 104 is specifically used for: By using the target group as the starting point of the link and each historical salesperson as the ending point of the link, a path search is performed on the original relationship graph to obtain multiple original relationship links; Based on a preset link length threshold, multiple original relationship links are filtered to obtain service relationship links.

[0077] In one embodiment, the intimacy score calculation module 105 is specifically used for: Group classification is performed based on multi-source group information to obtain group type data for the target group; A preset relationship weight mapping table is obtained based on group type data; The service relationship links are evaluated based on the relationship weight mapping table to obtain the intimacy score of the service relationship links.

[0078] In one embodiment, the intimacy score calculation module 105 is specifically used for: Information is extracted from multi-source group information to obtain group insurance records and group consultation records; Groups are classified based on group underwriting records and group consultation records to obtain group type data for the target groups.

[0079] In one embodiment, the intimacy score calculation module 105 is specifically used for: Obtain the type weight data corresponding to the group type data; Subclass weight data is obtained by querying the relationship weight mapping table based on the service relationship link; The affinity score of the service relationship link is obtained by aggregating the type weight data and the subclass weight data.

[0080] In one embodiment, the product recommendation module 106 is specifically used for: For each historical salesperson, obtain all service relationship links related to the historical salesperson, and sum the intimacy scores of all service relationship links to obtain the total intimacy score of the historical salesperson. The historical sales representatives are sorted based on their total affinity score to obtain a sales representative sequence; The salesperson sequence is filtered to obtain the target salesperson.

[0081] This application provides a group insurance product recommendation device applicable to group insurance product recommendation scenarios in fintech. First, it acquires multi-source group information of the target group, providing a rich data foundation for subsequent relationship mining and avoiding relationship omissions caused by a single data source. Based on the multi-source group information, it obtains historical service information of at least one historical agent associated with the target group, ensuring that candidate agent selection has clear data basis rather than random allocation. Next, it constructs a graph based on the multi-source group information and historical service information to obtain an original relationship graph, providing a computable graph structure for link queries and overcoming the limitation of traditional tabular data in expressing relationships. Link queries are then performed on the original relationship graph to obtain service relationship links. These service relationship links characterize the relationship links between the target group and each historical agent, making the connection path between agents and groups clear and traceable, providing a clear chain of evidence for subsequent scoring. Furthermore, it calculates scores for the service relationship links to obtain a closeness score, transforming the relationship into a quantifiable numerical indicator and solving the problem of lack of standards in subjective judgment. Finally, based on the intimacy score, historical agents are screened to obtain target agents, and insurance products are recommended to target groups through these target agents. This allows agents to use specific information in the relationship chain (such as co-ownership, historical insurance policies, etc.) as a link to build connections, enhance customer trust, and solve the problem that agents cannot accurately grasp customer needs due to the lack of identification of related relationships in the allocation of new customers. This helps to improve the accuracy and conversion efficiency of group insurance recommendations.

[0082] Specific limitations regarding the group insurance product recommendation device can be found in the limitations of the group insurance product recommendation method described above, and will not be repeated here. Each module in the aforementioned group insurance product recommendation device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the corresponding operations of each module.

[0083] Please see Figure 10 , Figure 10 The hardware structure of a computer device according to another embodiment is illustrated. The computer device includes: The processor 1001 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application. The memory 1002 can be implemented as a read-only memory (ROM), static storage device, dynamic storage device, or random access memory (RAM). The memory 1002 can store the operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 1002 and is called and executed by the processor 1001 to perform the group insurance product recommendation method of the embodiments of this application, including: Obtain multi-source group information for the target group; Based on multi-source group information, obtain historical service information of at least one historical salesperson who has a relationship with the target group; A graph is constructed based on multi-source community information and historical service information to obtain the original relationship graph. Link queries are performed on the original relationship graph to obtain service relationship links, which are used to represent the relationship links between the target group and each historical salesperson. A score is calculated for each service relationship link to obtain a closeness score for that service relationship link. Based on affinity scores, historical agents are screened to identify target agents, and insurance products are then recommended to target groups through these target agents.

[0084] Input / output interface 1003 is used to implement information input and output; The communication interface 1004 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.). Bus 1005 transmits information between various components of the device (e.g., processor 1001, memory 1002, input / output interface 1003, and communication interface 1004); The processor 1001, memory 1002, input / output interface 1003 and communication interface 1004 are connected to each other within the device via bus 1005.

[0085] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor: Obtain multi-source group information for the target group; Based on multi-source group information, obtain historical service information of at least one historical salesperson who has a relationship with the target group; A graph is constructed based on multi-source community information and historical service information to obtain the original relationship graph. Link queries are performed on the original relationship graph to obtain service relationship links, which are used to represent the relationship links between the target group and each historical salesperson. A score is calculated for each service relationship link to obtain a closeness score for that service relationship link. Based on affinity scores, historical agents are screened to identify target agents, and insurance products are then recommended to target groups through these target agents.

[0086] 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.

[0087] 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. When executed, the computer program 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 may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of 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.

[0088] 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.

[0089] 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.

[0090] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application 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 this application, and should all be included within the protection scope of this application.

Claims

1. A method for recommending group insurance products, characterized in that, The method includes: Obtain multi-source group information for the target group; Based on the multi-source group information, obtain the historical service information of at least one historical salesperson who is associated with the target group; Based on the multi-source community information and the historical service information, a graph is constructed to obtain the original relationship graph; A link query is performed on the original relationship graph to obtain service relationship links, wherein the service relationship links are used to characterize the relationship links between the target group and each of the historical salespersons; A score is calculated on the service relationship links to obtain the intimacy score of the service relationship links; Based on the intimacy score, the historical sales agents are screened to obtain target sales agents, and insurance products are recommended to the target group through the target sales agents.

2. The method for recommending group insurance products according to claim 1, characterized in that, The step of calculating the intimacy score of the service relationship link includes: Based on the multi-source group information, group classification is performed to obtain group type data of the target group; A preset relationship weight mapping table is obtained based on the group type data; The service relationship links are evaluated based on the relationship weight mapping table to obtain the intimacy score of the service relationship links.

3. The method for recommending group insurance products according to claim 2, characterized in that, The step of evaluating the service relationship links based on the relationship weight mapping table to obtain the intimacy score of the service relationship links includes: Obtain the type weight data corresponding to the group type data; Based on the service relationship link, subclass weight data is obtained by querying the relationship weight mapping table; The intimacy score of the service relationship link is obtained by aggregating and calculating the type weight data and the subclass weight data.

4. The method for recommending group insurance products according to claim 2, characterized in that, The process of classifying groups based on the multi-source group information to obtain group type data for the target group includes: Information is extracted from the multi-source group information to obtain group insurance records and group consultation records; Based on the group underwriting records and the group consultation records, the groups are classified to obtain the group type data of the target group.

5. The method for recommending group insurance products according to claim 1, characterized in that, The process of filtering historical sales representatives based on the affinity score to obtain target sales representatives includes: For each historical salesperson, obtain all service relationship links related to the historical salesperson, and sum the intimacy scores of all the service relationship links to obtain the total intimacy score of the historical salesperson. The historical salespersons are sorted based on the total intimacy score to obtain a salesperson sequence; The target salespersons are obtained by filtering the salesperson sequence.

6. The method for recommending group insurance products according to any one of claims 1 to 5, characterized in that, The step of performing a link query on the original relationship graph to obtain service relationship links includes: Using the target group as the starting point of the link and each of the historical salespersons as the ending point of the link, a path search is performed on the original relationship graph to obtain multiple original relationship links; Based on a preset link length threshold, multiple original relationship links are filtered to obtain the service relationship links.

7. The method for recommending group insurance products according to any one of claims 1 to 5, characterized in that, The process of constructing a graph based on the multi-source community information and the historical service information to obtain an original relationship graph includes: The multi-source community information is cleaned to obtain the target community information; The historical service information is cleaned to obtain the target service information; Entity extraction is performed on the target group information and the target service information to obtain graph node data; Relationships are extracted from the target group information and the target service information to obtain graph node relationship data; The original relational graph is obtained by constructing a graph based on the graph node data and the graph node relationship data.

8. A group insurance product recommendation device, characterized in that, The device includes: The group information acquisition module is used to acquire multi-source group information of the target group; The salesperson information acquisition module is used to acquire historical service information of at least one historical salesperson who is associated with the target group based on the multi-source group information. The graph construction module is used to construct a graph based on the multi-source community information and the historical service information to obtain an original relationship graph. The link query module is used to perform link query on the original relationship graph to obtain service relationship links, wherein the service relationship links are used to characterize the relationship links between the target group and each of the historical salespersons; The intimacy score calculation module is used to calculate the intimacy score of the service relationship link. The product recommendation module is used to filter the historical salespersons based on the intimacy score, obtain target salespersons, and recommend insurance products to the target group through the target salespersons.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the group insurance product recommendation method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the group insurance product recommendation method as described in any one of claims 1 to 7.

Images