Insurance industry exclusive questionnaire generation and data collection and analysis system based on applet

Abstract

The application discloses an insurance industry exclusive questionnaire generation and data collection and analysis system based on applets, relates to the technical field of data analysis, and is used for solving the problems of single questionnaire generation mode, insufficient data analysis dimension and low data processing resource utilization efficiency in the prior art.The application is characterized in that the small programs of a plurality of users are managed, the data information cached in the small programs is uploaded to a search engine platform to aggregate fragmented information, the fragmented information is filtered to obtain a public fragmented information set and a private fragmented information cluster, a regular questionnaire is constructed based on the public fragmented information set, the regular questionnaire is processed into a portrait questionnaire based on the private fragmented information cluster, corresponding questionnaire feature data is collected for each portrait questionnaire, data analysis is performed based on the questionnaire feature data, a plurality of questionnaire circles are constructed based on the results of the data analysis, and information focus areas of the plurality of questionnaire circles are located to perform questionnaire directional pushing.

Description

Technical Field

[0001] This invention relates to the field of data analysis technology, specifically to a system for generating and collecting data for the insurance industry based on a mini-program. Background Technology

[0002] In existing technologies, market research and customer needs assessment in the insurance industry mainly rely on standardized questionnaires. These questionnaires typically adopt a "one-size-fits-all" generation model, with fixed and universal question settings that cannot adapt to the personalized backgrounds and potential needs of different users, resulting in a low degree of relevance between the questionnaire content and users' actual insurance needs.

[0003] At the data collection and analysis level, existing systems typically only focus on the user's final answer, while ignoring the user's interactive behavior sequence and questionnaire metadata during the questionnaire completion process. This makes it difficult to effectively identify the deep characteristics and preference differences of different user groups. In addition, the rigid backend data processing resource allocation mechanism of the system leads to prominent bottlenecks in data processing efficiency.

[0004] Ultimately, the aforementioned problems collectively lead to three core defects in existing insurance questionnaire systems: low personalization of questionnaire generation, superficial and limited data analysis dimensions, and inefficient utilization of system resources. These defects restrict the system's ability to accurately understand user needs and deliver targeted services. Summary of the Invention

[0005] The purpose of this invention is to provide a questionnaire generation and data collection and analysis system for the insurance industry based on a mini-program, in order to solve the problems in the background technology.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a questionnaire generation and data collection and analysis system for the insurance industry based on a mini-program, including a search engine platform, wherein the search engine platform is communicatively connected to a mini-program module, a questionnaire intelligent generation module, a questionnaire data collection module, and a questionnaire data analysis module;

[0007] The mini-program module is used to manage the mini-programs of several users and to upload the cached data information within the mini-programs to the search engine platform, where fragmented information is aggregated based on the data information.

[0008] The questionnaire intelligent generation module is used to filter fragmented information to obtain a public fragmented information set and a private fragmented information cluster. It constructs a regular questionnaire based on the public fragmented information set and processes the regular questionnaire into a profile questionnaire based on the private fragmented information cluster.

[0009] The questionnaire data acquisition module is used to collect corresponding questionnaire feature data for each profile questionnaire;

[0010] The questionnaire data analysis module constructs several questionnaire circles based on questionnaire feature data, performs data analysis on each questionnaire circle, and matches and connects several questionnaire circles based on the analysis results to obtain information focus areas for targeted questionnaire push.

[0011] Furthermore, the process involves managing individual mini-programs for several users, uploading cached data from these mini-programs to a search engine platform, and then aggregating fragmented information based on this data on the search engine platform.

[0012] Each user's identity information is obtained to build their respective data account, a mini-program platform is created, a mini-program is created for each user on the mini-program platform, a remote management space for each mini-program is built on the mini-program platform, and a near-end cache space is built at the data account.

[0013] The remote management space is used to manage access operations on the mini-program platform, cache the content of the access operations as data information, establish a remote address for the remote management space, divide the data information into several information fragments based on the preset field length, and build a fragment relationship library based on historical insurance policies to define several fragment matching relationships.

[0014] Several fragment matching relationships are used as the search benchmark data of the search engine platform. All information fragments and fragment matching relationships are traversed, and the information fragments are aggregated and processed into several categories of fragmented information. The near-end cache space creates a near-end address based on the remote address of the mini-program, associates the near-end address with the remote address to construct a mapping address set, and obtains the fragmented information corresponding to each remote address based on the mapping address set and caches it in the corresponding near-end cache space.

[0015] Furthermore, the process of filtering fragmented information to obtain public fragmented information sets and private fragmented information clusters includes:

[0016] Based on preset public fields, fragmented information is traversed and matched. Fragmented information that fully matches the public fields is filtered into a public paragraph set, while fragmented information that does not fully match is filtered into a private element set.

[0017] The public paragraph set is used to store several identical fragmented information paragraphs. Each fragmented information paragraph consists of several public fields. The private element set is used to store several private fields that cannot be matched with the public fields.

[0018] Based on fragmented information segments, a public fragmented information set is constructed. Private fields with the same content are encapsulated into a private queue. The call address of the private queue is set, and the call address is filled into the corresponding position of the private field encapsulated in the original private element set. The call addresses of all private queues corresponding to each near-end cache space, as well as other private fields that have not been encapsulated, are clustered into their own private fragmented information clusters.

[0019] Furthermore, the process of constructing a routine questionnaire based on fragmented public information sets includes:

[0020] Create a blank questionnaire, fill it with public fragmented information, and process the blank questionnaire into a regular questionnaire. At the end of the blank questionnaire, set a corresponding number of jump links based on the number of private fragmented information clusters, create questionnaire expansion templates for each jump link, and process the private fragmented information cluster of each jump link into several questionnaire items. Each questionnaire item corresponds to a type of insurance business-related data.

[0021] Furthermore, the process of transforming a regular questionnaire into a profile questionnaire based on private, fragmented information clusters includes:

[0022] The questionnaire expansion template has several data fill points. Each data fill point is used to process data related to a certain type of insurance business. Several questionnaire items are entered into the corresponding data fill points on the questionnaire expansion template according to their respective types. The insurance business related data corresponding to all data fill points are used as their respective profile data. Each questionnaire expansion template with all profile data is merged with the original regular questionnaire to form its own profile questionnaire.

[0023] Furthermore, the process of collecting corresponding questionnaire feature data for each profile questionnaire includes:

[0024] Set up a front-end data tracking point for each profile questionnaire, set up a back-end processing program, and the back-end processing program consists of several processing threads. Each processing thread is used to process a front-end task, and data sharing is established between the front-end data tracking point and the processing threads.

[0025] The front-end data tracking system monitors the profile questionnaires and all interactive behaviors on the questionnaires in real time. All relevant data obtained in real time is packaged as a front-end task and transmitted to the back-end processing program. The back-end processing program allocates idle processing threads and obtains available resources on the cloud server to process the data of the front-end task, thereby obtaining the questionnaire feature data of each profile questionnaire. The questionnaire feature data includes questionnaire metadata and behavioral sequence data.

[0026] Furthermore, the process by which the backend processor obtains available resources to process data from the frontend task includes:

[0027] When a front-end task accesses a back-end processing program, the back-end processing program simultaneously processes data from several front-end tasks and determines whether the available resources on the cloud server can sustain the data processing for the latest front-end task.

[0028] If so, no action will be taken;

[0029] If not, then the task messages of several front-end tasks that are already in data processing are communicated with each other. Based on the task message communication, the amount of resources that can be released is obtained, and the releaseable resources are returned to the cloud server. The back-end processing program then calls the available resources that meet the needs of the latest front-end task, or directly schedules the releaseable resources to the latest front-end task.

[0030] Furthermore, the process of task message exchange includes:

[0031] Based on the time sequence, several real-time message points are set, and a number of mirror message points are set for each real-time message point corresponding to the time sequence. A message channel is built by connecting the real-time message points and mirror message points under each time sequence for message communication, and the amount of resources that can be released for all front-end tasks that are processing data under the corresponding time sequence is obtained.

[0032] If the amount of releasable resources in a given time sequence directly meets the amount of available resources required by the latest front-end task, then the releasable resources in the corresponding time sequence will be directly scheduled to the latest front-end task. The data processing program will then perform local parsing of the releasable resources at the location of the front-end task and execute data processing for the latest front-end task.

[0033] If the amount of releaseable resources in a given time series is insufficient to meet the available resources required by the latest front-end task, then the amount of releaseable resources in the next time series is obtained by adding the amount of releaseable resources to the current time series, until the total amount of resources meets the requirements. The releaseable resources of several time series are then packaged into a resource package and returned to the cloud server for cloud parsing. The parsed data is then sent to the latest front-end task based on its own cloud address. The data processing program directly calls the cloud address of the parsed data to complete the data processing of the front-end task.

[0034] Furthermore, the process of conducting data analysis based on questionnaire feature data, constructing several questionnaire circles based on the results of the data analysis, and then targeting the information focus areas of these questionnaire circles for targeted questionnaire delivery includes:

[0035] Group several users with the same questionnaire metadata in the questionnaire feature data into one user cluster, and group several users with the same behavioral sequence data into another user cluster, and label them as Private Domain User Cluster 1 and Private Domain User Cluster 2 respectively. Repeat the grouping of all questionnaire feature data to obtain several Private Domain User Cluster 1 and Private Domain User Cluster 2.

[0036] Each private domain user cluster is used to represent the user group corresponding to different questionnaire metadata categories;

[0037] Each private domain user cluster is used to represent user groups corresponding to different behavioral sequence data categories;

[0038] Each user in private domain user cluster 1 and private domain user cluster 2 is assigned a separate questionnaire circle. The questionnaire circles corresponding to private domain user cluster 1 and private domain user cluster 2 are respectively denoted as the first type of questionnaire circle and the second type of questionnaire circle. Based on the questionnaire feature data of each user's corresponding profile questionnaire, the common area of ​​each user in the first type of questionnaire circle and the second type of questionnaire circle is located and used as the information focus area of ​​the corresponding user. Subsequent insurance-related questionnaires are then pushed to the information focus area.

[0039] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0040] 1. This invention aggregates fragmented information within a user's mini-program and categorizes it into public fragmented information sets and private fragmented information clusters, thus realizing a questionnaire construction model of "general template + personalized module". On the one hand, it ensures the processing efficiency of the basic part of the questionnaire. On the other hand, by introducing the user's private data characteristics, the generated "profile questionnaire" is highly consistent with the user's individual situation, solving the problem of the questionnaire content being out of touch with the user's real needs in the prior art.

[0041] 2. This invention uses both "questionnaire metadata" and "behavioral sequence data" as the basis for analysis. By constructing a "type one questionnaire circle" representing the questionnaire content and a "type two questionnaire circle" representing user interaction behavior, it locates the user's information focus area from the intersection of the two dimensions. This can reveal the common behavioral patterns of user groups and solve the problems of single data analysis dimensions and coarse user profiles in existing technologies. It provides a reliable data basis for subsequent targeted push.

[0042] 3. This invention sets up a multi-threaded mechanism for front-end data tracking and back-end processing programs, and designs a releaseable resource scheduling mechanism based on task message communication. This enables the selective allocation of computing resources between front-end tasks and cloud servers according to real-time front-end task requirements, ensuring the stable execution of data processing tasks under high load conditions. This effectively solves the problems of rigid resource allocation and data processing efficiency bottlenecks in existing systems. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0044] Figure 1 This is a system block diagram of the present invention.

[0045] Figure 2 This is a schematic diagram of the process for constructing a profile questionnaire in this invention. Detailed Implementation

[0046] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0047] Please see Figure 1 As shown, the insurance industry-specific questionnaire generation and data collection and analysis system based on mini-programs includes a search engine platform, which is communicatively connected to a mini-program module, a questionnaire intelligent generation module, a questionnaire data collection module, and a questionnaire data analysis module.

[0048] The mini-program module is used to manage the mini-programs of several users and to upload the cached data information within the mini-programs to the search engine platform, where fragmented information is aggregated based on the data information.

[0049] The questionnaire intelligent generation module is used to filter fragmented information to obtain a public fragmented information set and a private fragmented information cluster. It constructs a regular questionnaire based on the public fragmented information set and processes the regular questionnaire into a profile questionnaire based on the private fragmented information cluster.

[0050] The questionnaire data acquisition module is used to collect corresponding questionnaire feature data for each profile questionnaire;

[0051] The questionnaire data analysis module performs data analysis based on questionnaire feature data, constructs several questionnaire circles based on the results of the data analysis, and locates the key information areas of several questionnaire circles for targeted questionnaire push.

[0052] It should be further explained that, in the specific implementation process, managing the individual mini-programs of several users and uploading the cached data information within the mini-programs to the search engine platform, and then aggregating fragmented information based on this data on the search engine platform, includes:

[0053] Obtain the user identity information corresponding to each user, build the corresponding data account based on the user identity information, create a mini-program platform, set up several program ports on the mini-program platform, and assign a program port on the mini-program platform to each user.

[0054] Connect the data account to the assigned program port to create a mini program within the mini program platform. Build a remote management space for each mini program on the mini program platform and a local cache space for each mini program at their respective data account.

[0055] The remote management space is used to manage the access operations of data accounts on the mini-program platform, cache the content of the access operations as their respective data information, establish a remote address for the remote management space, divide the data information into several information fragments based on a preset field length, perform data semantic analysis based on historical insurance policies and build a fragment relationship library, which is used to define several fragment matching relationships.

[0056] The semantic analysis of historical insurance policies is conducted by constructing a bag-of-words model. This involves extracting textual information from historical policies, including policy terms, insured information, insurance liabilities, and exclusions. The text is then cleaned, irrelevant characters are removed, and standardized formatting is performed. Word segmentation and part-of-speech tagging are also performed. A bag-of-words model is constructed using natural language processing technology. This model then performs entity recognition and relation extraction on the textual information, identifying key entities in the historical policies, such as the insured, insurance company, insurance product, insured amount, and insurance period. Relationships between entities are extracted, such as "insured—purchase—insurance product" or "insurance product—includes—insurance liability."

[0057] The results of data semantic analysis are stored in the form of a knowledge graph to form a fragmented relational database. The fragmented relational database specifically includes an entity table, a relation table, and an index. The entity table stores all entities, including entity types and attributes. The relation table stores the relationships between entities, including relation types and relation strength. Relationship strength measures the tightness, importance, and reliability of the association between two entities, and is characterized by indicators such as co-occurrence frequency, semantic similarity, and statistical correlation.

[0058] Explanation of co-occurrence frequency: The frequency with which two entities appear together in the same context (such as an insurance policy or a clause). For example, "malignant tumor" and "critical illness insurance" frequently appear together in tens of thousands of insurance policies, so their co-occurrence frequency is very high.

[0059] Explanation of semantic similarity: The word vector model calculates the vector distance between two entities in the semantic space. The closer the vector distance, the higher the semantic similarity. For example, the vector distance between "myocardial infarction" and "acute myocardial infarction" is almost the same, and their semantic similarity is close to full. However, the semantic similarity between "fracture" and "serious disease" is relatively weak.

[0060] Explanation of statistical correlation: It measures the statistical dependence of two entities in their co-occurrence by using statistical indicators such as TF-IDF and PMI. For example, by analyzing a number of policy documents, it is found that when "designated driver" appears, the probability of "premium discount" appearing is extremely high, then the statistical correlation between them is very high.

[0061] Indexes are used to connect entity tables and relationship tables to enable quick queries of each specific fragment matching relationship;

[0062] Several fragment matching relationships are used as the search benchmark data of the search engine platform. All information fragments are traversed with several fragment matching relationships to obtain the matching results between different information fragments. All information fragments are aggregated and processed into several categories of fragmented information.

[0063] Each category of fragmented information is used to characterize a text set with a similar range of information.

[0064] The near-end cache space creates its own near-end address based on the remote address of the mini-program, associates the near-end address with the remote address to build a mapping address set, and obtains the fragmented information corresponding to each remote address based on the mapping address set for its own caching.

[0065] It should be noted that by obtaining data information within the respective mini-program platform, processing the data information into information fragments, and then aggregating different categories of fragmented information based on fragment matching relationships, fragmented information is used to cache the relevant fragmented information in the near-end cache space. This enables the specific processing to be carried out on the public end (i.e., the search engine platform) with abundant computing resources, while the processing results are directly obtained on their respective private ends (i.e., the near-end cache space), thereby improving the overall data processing efficiency.

[0066] It should be further explained that, in the specific implementation process, the process of filtering fragmented information to obtain public fragmented information sets and private fragmented information clusters includes:

[0067] Fragmented information in several near-end cache spaces is traversed and matched based on a preset common field. The fragmented information that completely matches the common field is filtered into a common paragraph set, and the fragmented information that does not completely match is filtered into a private element set.

[0068] The public paragraph set is used to store several identical fragmented information paragraphs, each fragmented information paragraph being composed of several corresponding public fields. The private element set is used to store several private fields in each near-end cache space that cannot be matched with the public fields.

[0069] A common fragmented information set is constructed based on the fragmented information segments that correspond to all fragmented information. Private fields in each private element set are matched, and private fields with the same content are encapsulated into a private queue. The call address of the private queue is set, and the call address is filled in the corresponding position of the private field encapsulated in the original private element set. The call addresses of all private queues corresponding to each near-end cache space, as well as other private fields that have not been encapsulated, are clustered into their own private fragmented information clusters.

[0070] It should be further explained that, in the specific implementation process, the process of constructing a regular questionnaire based on public fragmented information sets and processing the regular questionnaire into a profile questionnaire based on private fragmented information clusters includes:

[0071] Please refer to the flowchart illustrating the process of converting a standard questionnaire into a profile questionnaire. Figure 2 As shown;

[0072] The public fragmented information set is used to characterize the convergent features of different users and to reflect the same choices made by different users on several insurance terms offered by the insurance market. A blank questionnaire is created, and the blank questionnaire is filled in based on the public fragmented information set, and then the blank questionnaire is processed and constructed into a regular questionnaire.

[0073] At the end of the blank questionnaire, based on the number of private fragmented information clusters, a corresponding number of jump links are set, and questionnaire expansion templates are created for each jump link. The private fragmented information cluster corresponding to each jump link is processed into several questionnaire items, and each questionnaire item corresponds to a type of insurance business-related data.

[0074] The questionnaire expansion template has several data fill points. Each data fill point is used to process data related to a certain type of insurance business. Several questionnaire items are entered into their respective data fill points on the questionnaire expansion template according to their respective types. The insurance business related data corresponding to all data fill points are used as their respective profile data. Each questionnaire expansion template with all corresponding profile data is merged with the original regular questionnaire to form its respective profile questionnaire.

[0075] Each private fragmented information cluster has a unique profile, representing a distinct user's insurance-related profile. Once all data entry points on the questionnaire expansion template have been completed for the corresponding type of insurance business-related data, each user successfully generates their respective insurance questionnaire. The regular questionnaire, as a part of the public insurance content, combines the regular questionnaire with the user's personalized data (private fragmented information clusters) to form the final profile questionnaire. On the one hand, the same data corresponding to the public insurance data only needs to be processed once, reducing the workload of data processing. On the other hand, the introduction of private fragmented information clusters achieves the goal of generating personalized insurance questionnaires.

[0076] It should be further explained that, in the specific implementation process, the process of collecting corresponding questionnaire feature data for each profile questionnaire includes:

[0077] Several portrait questionnaires were numbered, and the numbers of the several portrait questionnaires were as follows: = 1, 2, 3, ..., n, where n is a natural number greater than 0, from Starting with the profile questionnaire with a value of 1, a front-end data tracking point is set for each profile questionnaire in sequence, and the number of the profile questionnaire serves as its corresponding identity identification identifier.

[0078] A backend processing program is set up, consisting of several processing threads. Each processing thread is used to process a frontend task. Data sharing is established between the frontend data tracking points and the processing threads. The frontend data tracking points monitor the profile questionnaire in real time, as well as all corresponding interactive behaviors on the profile questionnaire. All relevant data obtained in real time is packaged as a frontend task and transmitted to the backend processing program. The backend processing program allocates idle processing threads and obtains available resources on the cloud server to process the data of the frontend task, thereby obtaining the questionnaire feature data corresponding to each profile questionnaire.

[0079] The questionnaire feature data includes questionnaire metadata and behavioral sequence data;

[0080] The questionnaire metadata is used to record the static characteristics presented by users on the profile questionnaire, including the total number of questions on the questionnaire, the distribution of questions in each section and the proportion of question types, as well as the behaviors such as adding, deleting, modifying and querying in the questionnaire, and the content change markers of each questionnaire.

[0081] The behavioral sequence data refers to the relevant behavioral data of users throughout the entire process of answering the questionnaire. Specifically, it includes the total completion time of the questionnaire, the response time from the questionnaire being pushed to the corresponding user accessing it, the questionnaire dwell time (i.e., the time the user spends on each question; too short a time indicates casualness or resistance, while too long a time indicates confusion or high attention), and answer modification behavior (i.e., recording the number of times the user modifies the answer to a question and the modification trajectory (e.g., changing from option A to B and then back to A), which reflects the degree of decision-making hesitation and sensitivity to the question).

[0082] The backend processing program obtains available resources to process data for the frontend task as follows: when the frontend task accesses the backend processing program, a resource visualization node is generated for the frontend task. The resource visualization node is used to visualize the execution progress of the frontend task after obtaining available resources. The backend processing program simultaneously processes data for several frontend tasks and determines whether the available resources on the cloud server can maintain the data processing for the latest frontend task.

[0083] If so, no action will be taken;

[0084] If not, then the task messages of several front-end tasks that are already in data processing are communicated with each other. Based on the task message communication, the amount of resources that can be released is obtained, and the releaseable resources are returned to the cloud server. The back-end processing program then calls the available resources that meet the needs of the latest front-end task, or directly schedules the releaseable resources to the latest front-end task.

[0085] The content of the task message communication is as follows: based on the time sequence, several instant message points are set, and for each instant message point corresponding to each time sequence, a number of mirror message points are set. The total number of instant message points and mirror message points corresponding to each time sequence is numerically equal to the number of front-end tasks that are already in data processing.

[0086] A message channel is constructed by connecting the instant message points and mirror message points under each time series. The message channel is used to communicate between an instant message point and several corresponding mirror message points under the same time series. Through message communication, the amount of resources that can be released for all front-end tasks that are processing data under the corresponding time series is obtained.

[0087] If the amount of releasable resources in a given time sequence directly meets the amount of available resources required by the latest front-end task, then the releasable resources in the corresponding time sequence will be directly scheduled to the latest front-end task. The data processing program will then perform local parsing of the releasable resources at the location of the front-end task and execute data processing for the latest front-end task.

[0088] If the amount of releaseable resources for a given time series is insufficient to meet the available resources required by the latest front-end task, the system continues to acquire the amount of releaseable resources for the next time series, summed up with the current time series, until the total amount of resources is sufficient to process the data for the latest front-end task. The releaseable resources corresponding to several time series are then packaged into a resource package and returned to the cloud server. The cloud server performs cloud-based parsing on the resource package and sends the parsed data to the latest front-end task based on its own cloud address. The data processing program then directly calls the cloud address of the parsed data to complete the data processing for the front-end task.

[0089] It should be noted that when the total available resources released by several front-end tasks that have already processed data within a given time series are sufficient to meet the processing needs of the latest front-end task, the amount of releaseable resources within a single time series is relatively small. This allows for direct local parsing and processing at the front-end task level, accelerating data parsing efficiency. However, when releaseable resources span multiple time series, the complexity of data parsing increases significantly. In such cases, the resource packages are returned to the cloud server for cloud-based parsing. The parsed data from the cloud is then used by the front-end task, enhancing the reliability of data parsing and improving data processing efficiency, while avoiding data errors caused by local parsing.

[0090] It should be further explained that, in the specific implementation process, the process of conducting data analysis based on questionnaire characteristic data, constructing several questionnaire circles based on the results of the data analysis, and targeting the key information areas of several questionnaire circles for targeted questionnaire delivery includes:

[0091] Data analysis is performed on the questionnaire feature data corresponding to several users. Several users with the same questionnaire metadata in the questionnaire feature data are grouped into a user cluster and labeled as Private Domain User Cluster 1. Several users with the same behavioral sequence data in the questionnaire feature data are grouped into a user cluster and labeled as Private Domain User Cluster 2. The corresponding grouping operation is repeated for all questionnaire feature data to obtain several Private Domain User Cluster 1 and Private Domain User Cluster 2.

[0092] Each private domain user cluster is used to represent the user group corresponding to different questionnaire metadata categories;

[0093] Each private domain user cluster is used to represent user groups corresponding to different behavioral sequence data categories;

[0094] For each user belonging to Private Domain User Cluster 1, a separate questionnaire circle is constructed, and the questionnaire circle corresponding to Private Domain User Cluster 1 is denoted as Category 1 questionnaire circle. For each user belonging to Private Domain User Cluster 2, a separate questionnaire circle is constructed, and the questionnaire circle corresponding to Private Domain User Cluster 2 is denoted as Category 2 questionnaire circle. Based on the questionnaire feature data of each user's profile questionnaire, the common area of ​​each user in Category 1 and Category 2 questionnaire circles is located, and it is used as the information focus area for the corresponding user. Subsequent insurance-related questionnaires are then pushed to the information focus area.

[0095] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A questionnaire generation and data collection and analysis system for the insurance industry based on a mini-program, characterized in that: This includes a search engine platform, which is communicatively connected to a mini-program module, a questionnaire intelligent generation module, a questionnaire data collection module, and a questionnaire data analysis module. The mini-program module is used to manage the mini-programs of several users and to upload the cached data information within the mini-programs to the search engine platform, where fragmented information is aggregated based on the data information. This includes: obtaining each user's user identity information to build their respective data accounts, creating a mini-program platform, creating a mini-program for each user on the mini-program platform, building a remote management space for each mini-program on the mini-program platform, and building a near-end cache space at the data account. The questionnaire intelligent generation module is used to filter fragmented information to obtain a public fragmented information set and a private fragmented information cluster. It constructs a regular questionnaire based on the public fragmented information set and processes the regular questionnaire into a profile questionnaire based on the private fragmented information cluster. The process of filtering fragmented information to obtain public fragmented information sets and private fragmented information clusters includes: Based on preset public fields, fragmented information is traversed and matched. Fragmented information that fully matches the public fields is filtered into a public paragraph set, while fragmented information that does not fully match is filtered into a private element set. The public paragraph set is used to store several identical fragmented information paragraphs. Each fragmented information paragraph consists of several public fields. The private element set is used to store several private fields that cannot be matched with the public fields. Based on fragmented information segments, a public fragmented information set is constructed. Private fields with the same content are encapsulated into a private queue. The call address of the private queue is set, and the call address is filled in the corresponding position of the private field encapsulated in the original private element set. The call addresses of all private queues corresponding to each near-end cache space, as well as other private fields that have not been encapsulated, are clustered into their own private fragmented information clusters. The questionnaire data acquisition module is used to collect corresponding questionnaire feature data for each profile questionnaire; a front-end data embedding point is set for each profile questionnaire, and a back-end processing program is set. The back-end processing program consists of several processing threads. The processing threads are used to process a front-end task. When the back-end processing program determines that the available resources in the cloud are insufficient, it performs task message communication with the processed front-end task to obtain the corresponding releaseable resources, and schedules the releaseable resources for the latest front-end task to use. When task messages are exchanged, real-time message points and mirror message points are set according to time sequence, and a message channel is built. Based on the judgment result of whether the releasable resources accumulated by single or multiple time sequences meet the requirements, the system adaptively selects local parsing or packaged upload to the cloud for parsing. The questionnaire data analysis module constructs several questionnaire circles based on questionnaire feature data, performs data analysis on each questionnaire circle, and matches and connects several questionnaire circles based on the analysis results to obtain information focus areas for targeted questionnaire push.

2. The insurance industry-specific questionnaire generation and data collection and analysis system based on a mini-program as described in claim 1, characterized in that, Managing multiple users' individual mini-programs and uploading cached data from these mini-programs to a search engine platform, followed by the process of aggregating fragmented information based on this data on the search engine platform, includes: The remote management space is used to manage access operations on the mini-program platform, cache the content of the access operations as data information, establish a remote address for the remote management space, divide the data information into several information fragments based on the preset field length, and build a fragment relationship library based on historical insurance policies to define several fragment matching relationships. Several fragment matching relationships are used as the search benchmark data of the search engine platform. All information fragments and fragment matching relationships are traversed, and the information fragments are aggregated and processed into several categories of fragmented information. The near-end cache space creates a near-end address based on the remote address of the mini-program, associates the near-end address with the remote address to construct a mapping address set, and obtains the fragmented information corresponding to each remote address based on the mapping address set and caches it in the corresponding near-end cache space.

3. The insurance industry-specific questionnaire generation and data collection and analysis system based on a mini-program as described in claim 2, characterized in that, The process of constructing a standard questionnaire based on public fragmented information sets includes: Create a blank questionnaire, fill it with public fragmented information, and process the blank questionnaire into a regular questionnaire. At the end of the blank questionnaire, set a corresponding number of jump links based on the number of private fragmented information clusters, create questionnaire expansion templates for each jump link, and process the private fragmented information cluster of each jump link into several questionnaire items. Each questionnaire item corresponds to a type of insurance business-related data.

4. The insurance industry-specific questionnaire generation and data collection and analysis system based on a mini-program as described in claim 3, characterized in that, The process of transforming a regular questionnaire into a profile questionnaire based on private fragmented information clusters includes: The questionnaire expansion template has several data fill points. Each data fill point is used to process data related to a certain type of insurance business. Several questionnaire items are entered into the corresponding data fill points on the questionnaire expansion template according to their respective types. The insurance business related data corresponding to all data fill points are used as their respective profile data. Each questionnaire expansion template with all profile data is merged with the original regular questionnaire to form its own profile questionnaire.

5. The insurance industry-specific questionnaire generation and data collection and analysis system based on a mini-program as described in claim 4, characterized in that, The process of collecting corresponding questionnaire feature data for each profile questionnaire includes: Establish data sharing between front-end data tracking points and processing threads; The front-end data tracking system monitors the profile questionnaires and all interactive behaviors on the questionnaires in real time. All relevant data obtained in real time is packaged as a front-end task and transmitted to the back-end processing program. The back-end processing program allocates idle processing threads and obtains available resources on the cloud server to process the data of the front-end task, thereby obtaining the questionnaire feature data of each profile questionnaire. The questionnaire feature data includes questionnaire metadata and behavioral sequence data.

6. The insurance industry-specific questionnaire generation and data collection and analysis system based on a mini-program as described in claim 5, characterized in that, The process by which the backend processor obtains available resources to process data from the frontend task includes: When a front-end task accesses a back-end processing program, the back-end processing program simultaneously processes data from several front-end tasks and determines whether the available resources on the cloud server can sustain the data processing for the latest front-end task. If so, no action will be taken; If not, then the task messages of several front-end tasks that are already in data processing are communicated with each other. Based on the task message communication, the amount of resources that can be released is obtained, and the releaseable resources are returned to the cloud server. The back-end processing program then calls the available resources that meet the needs of the latest front-end task, or directly schedules the releaseable resources to the latest front-end task.

7. The insurance industry-specific questionnaire generation and data collection and analysis system based on a mini-program as described in claim 6, characterized in that, The process of task message communication includes: Based on the time sequence, several real-time message points are set, and a number of mirror message points are set for each real-time message point corresponding to the time sequence. A message channel is built by connecting the real-time message points and mirror message points under each time sequence for message communication, and the amount of resources that can be released for all front-end tasks that are processing data under the corresponding time sequence is obtained. If the amount of releasable resources in a given time sequence directly meets the amount of available resources required by the latest front-end task, then the releasable resources in the corresponding time sequence will be directly scheduled to the latest front-end task. The data processing program will then perform local parsing of the releasable resources at the location of the front-end task and execute data processing for the latest front-end task. If the amount of releaseable resources in a given time series is insufficient to meet the available resources required by the latest front-end task, then the amount of releaseable resources in the next time series is obtained by adding the amount of releaseable resources to the current time series, until the total amount of resources meets the requirements. The releaseable resources of several time series are then packaged into a resource package and returned to the cloud server for cloud parsing. The parsed data is then sent to the latest front-end task based on its own cloud address. The data processing program directly calls the cloud address of the parsed data to complete the data processing of the front-end task.

8. The insurance industry-specific questionnaire generation and data collection and analysis system based on a mini-program as described in claim 7, characterized in that, The process of conducting data analysis based on questionnaire feature data, constructing several questionnaire circles based on the results of the data analysis, and then targeting and pushing questionnaires to specific information focus areas within these questionnaire circles includes: Group several users with the same questionnaire metadata in the questionnaire feature data into one user cluster, and group several users with the same behavioral sequence data into another user cluster, and label them as Private Domain User Cluster 1 and Private Domain User Cluster 2 respectively. Repeat the grouping of all questionnaire feature data to obtain several Private Domain User Cluster 1 and Private Domain User Cluster 2. Each private domain user cluster is used to represent the user group corresponding to different questionnaire metadata categories; Each private domain user cluster is used to represent user groups corresponding to different behavioral sequence data categories; Each user in private domain user cluster 1 and private domain user cluster 2 is assigned a separate questionnaire circle. The questionnaire circles corresponding to private domain user cluster 1 and private domain user cluster 2 are respectively denoted as the first type of questionnaire circle and the second type of questionnaire circle. Based on the questionnaire feature data of each user's corresponding profile questionnaire, the common area of ​​each user in the first type of questionnaire circle and the second type of questionnaire circle is located and used as the information focus area of ​​the corresponding user. Subsequent insurance-related questionnaires are then pushed to the information focus area.

Images