Key user screening identification method and device, equipment and storage medium
By using time-series feature engineering and integrated learning agents for multimodal business data, key users are automatically identified, solving the problem of low accuracy in manual identification and improving identification efficiency and accuracy. This method is suitable for risk user screening in financial businesses.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CHINA PING AN PROPERTY INSURANCE CO LTD
- Filing Date
- 2026-01-13
- Publication Date
- 2026-05-08
AI Technical Summary
In existing technologies, the identification of key users by humans is easily affected by subjective factors, resulting in low accuracy, slow speed, and long time consumption.
By employing time-series feature engineering processing of multimodal business data and combining it with ensemble learning to complete user screening and identification intelligent agents, dynamic indicator data is obtained and user category information is identified, thereby automatically identifying key users.
It improves the efficiency and accuracy of identifying key users, reduces the misjudgment rate caused by human experience in identification, assists financial businesses in identifying risky users, and avoids insurance or claims institutions from bearing high risks.
Smart Images
Figure CN121997223A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of artificial intelligence technology and is applied to the scenario of screening and identifying key users when signing target business agreements. It relates to a method, apparatus, device and storage medium for screening and identifying key users. Background Technology
[0002] With the rapid development of artificial intelligence technology, intelligent agent technology is being widely used in various industries and fields. For example, financial business automation is being implemented, and more and more financial business processes can be completed via the Internet.
[0003] To ensure the security and focus of financial transactions conducted online, it is necessary to identify key users within the user group, such as those with good credit histories or substantial assets. Currently, the method for screening key users involves manual review and verification of the transaction request data after the user submits the request. Manual identification of key users is not only susceptible to subjective factors leading to low accuracy, but also time-consuming, resulting in slow identification speed. Summary of the Invention
[0004] The purpose of this application is to provide a method, apparatus, device, and storage medium for screening and identifying key users, so as to improve the efficiency and accuracy of key user identification.
[0005] In a first aspect, embodiments of this application provide a method for screening and identifying key users, which employs the following technical solution: A method for screening and identifying key users includes the following steps: Obtain multimodal business data provided by the target user; The multimodal business data is subjected to time-series feature engineering to obtain dynamic indicator data that incorporates business time nodes; The dynamic indicator data is input into the user screening and identification intelligent agent that has completed the integrated learning; Obtain the user category information output by the user screening and identification agent; Based on the user category information, identify whether the target user is a key user.
[0006] Secondly, this application also provides a key user screening and identification device, which adopts the following technical solution: A key user screening and identification device includes: The business data acquisition module is used to acquire multimodal business data provided by the target user; The dynamic indicator data acquisition module is used to perform time-series feature engineering processing on the multimodal business data to obtain dynamic indicator data that incorporates business time nodes. The dynamic indicator data input module is used to input the dynamic indicator data into the user screening and identification intelligent agent that has completed integrated learning; The user category information output acquisition module is used to acquire the user category information output by the user screening and identification intelligent agent. The key user identification module is used to identify whether the target user is a key user based on the user category information.
[0007] Thirdly, embodiments of this application also provide a computer device that adopts the technical solution described below: A computer device includes a memory and a processor, wherein the memory stores computer-readable instructions, and the processor executes the computer-readable instructions to implement the steps of the key user screening and identification method described above.
[0008] Fourthly, embodiments of this application also provide a computer-readable storage medium, which adopts the technical solutions described below: A computer-readable storage medium storing computer-readable instructions, which, when executed by a processor, implement the steps of the key user screening and identification method described above.
[0009] Compared with the prior art, the embodiments of this application have the following main advantages: The key user screening and identification method described in this application can be widely applied to scenarios where key users are screened and identified during the signing of target business agreements. It involves acquiring multimodal business data provided by the target user; performing time-series feature engineering to obtain dynamic indicator data incorporating business time nodes; inputting this data into a user screening and identification intelligent agent; obtaining user category information; and identifying whether the target user is a key user based on the user category information. For example, during insurance contract signing or claims review, this key user screening and identification method can assist insurance or claims-related financial businesses in identifying high-risk users in advance, preventing underwriting or claims institutions from incurring high business risks or losses. Furthermore, the user screening and identification intelligent agent, completed using ensemble learning, employs automated artificial intelligence processing for key user screening and identification, which not only improves the efficiency of key user identification but also reduces the misjudgment rate caused by relying solely on human experience, thereby improving the accuracy of key user identification. Attached Figure Description
[0010] To more clearly illustrate the solutions in this application, the accompanying drawings used in the description of the embodiments of this application will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0011] Figure 1 This is an exemplary system architecture diagram to which this application can be applied; Figure 2 This is a flowchart of an embodiment of a key user screening and identification method according to this application; Figure 3 yes Figure 2 A flowchart of a specific embodiment of step 202 shown; Figure 4 This is a flowchart of a specific embodiment of the user screening and identification intelligent agent integrated learning in the key user screening and identification method described in this application; Figure 5 yes Figure 4 A flowchart of a specific embodiment of step 404 shown; Figure 6 This is a flowchart of a specific embodiment of the key user screening and identification method described in this application, which constructs the mapping relationship between differentiated dynamic indicator data and user category output nodes; Figure 7 This is a flowchart of a specific embodiment of the key user screening and identification method described in this application, which uses a user screening and identification intelligent agent to identify user categories; Figure 8 This is a schematic diagram of one embodiment of a key user screening and identification device according to this application; Figure 9 This is a schematic diagram of the structure of one embodiment of the computer device according to this application. Detailed Implementation
[0012] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the terminology used herein in the specification of the application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application; the terms "comprising" and "having," and any variations thereof, in the specification, claims, and foregoing drawings of this application, are intended to cover non-exclusive inclusion. The terms "first," "second," etc., in the specification, claims, or foregoing drawings of this application are used to distinguish different objects, not to describe a particular order.
[0013] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.
[0014] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.
[0015] like Figure 1 As shown, system architecture 100 may include terminal device 101, network 102, and server 103. Terminal device 101 may be a laptop 1011, tablet 1012, or mobile phone 1013. Network 102 is used as a medium to provide a communication link between terminal device 101 and server 103. Network 102 may include various connection types, such as wired, wireless communication links, or fiber optic cables.
[0016] Users can use terminal device 101 to interact with server 103 via network 102 to receive or send messages, etc. Various communication client applications can be installed on terminal device 101, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social media platform software, etc.
[0017] Terminal device 101 can be various electronic devices with a display screen and support web browsing. In addition to laptops 1011, tablets 1012, or mobile phones 1013, terminal device 101 can also be an e-book reader, an MP3 player (Moving Picture Experts Group Audio Layer III), an MP4 player (Moving Picture Experts Group Audio Layer IV), a laptop computer, and a desktop computer, etc.
[0018] Server 103 can be a server that provides various services, such as a backend server that provides support for the pages displayed on terminal device 101.
[0019] It should be noted that the key user screening and identification method provided in this application embodiment is generally executed by a server, and correspondingly, a key user screening and identification device is generally installed in the server.
[0020] It should be understood that Figure 1 The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, and servers can be included.
[0021] Continue to refer to Figure 2 The diagram illustrates a flowchart of an embodiment of a key user screening and identification method according to this application. The key user screening and identification method includes the following steps: Step 201: Obtain multimodal business data provided by the target user.
[0022] In this embodiment, the target users include users who are about to sign an insurance contract and users who are about to file a claim; correspondingly, the multimodal business data includes data from the insurance contract signing materials and data from the claim application materials. Specifically, the multimodal business data varies depending on the business being handled by the target user.
[0023] Step 202: Perform time-series feature engineering on the multimodal business data to obtain dynamic indicator data that incorporates business time nodes.
[0024] In this embodiment, the multimodal business data is subjected to time-series feature engineering processing to obtain dynamic indicator data that incorporates business time nodes. For example, based on the business time nodes, dynamic indicator data such as the number of days between insurance application and underwriting, and the number of times the application materials need to be supplemented are calculated.
[0025] Step 203: Input the dynamic indicator data into the user screening and identification agent that has completed the integrated learning.
[0026] Specifically, the dynamic indicator data is directly input into the user screening and identification intelligent agent that has completed integrated learning. The AI agent's automated identification function identifies the user's corresponding category. These categories include key users and ordinary users, high-risk users and normal users, as well as users categorized by different membership levels. These categories vary depending on the actual business scenario. For example, when applying for insurance, users can be categorized into high-risk users and normal users based on their credit history; and into key users and ordinary users based on their asset holdings.
[0027] Step 204: Obtain the user category information output by the user screening and identification agent.
[0028] Step 205: Identify whether the target user is a key user based on the user category information.
[0029] Specifically, when identifying whether a target user is a key user based on the user category information output by the user screening and identification agent, the identification can be performed directly based on the user category information association table corresponding to key users. This will not be elaborated on in detail here.
[0030] In this embodiment, the key user screening and identification method can be widely applied to scenarios where key users are screened and identified during the signing of target business agreements, such as when signing insurance contracts or reviewing claims. This key user screening and identification method can assist financial businesses such as insurance or claims in identifying high-risk users in advance, thus preventing underwriting or claims institutions from bearing high business risks or losses. Here, the user screening and identification intelligent agent, which is completed using ensemble learning, adopts an automated artificial intelligence processing method to screen and identify key users. This not only improves the efficiency of key user identification but also reduces the misjudgment rate of business identification caused by relying solely on human experience to identify key users, thereby improving the accuracy of key user identification.
[0031] In this embodiment, multimodal business data provided by the target user is acquired; time-series feature engineering is performed to obtain dynamic indicator data incorporating business time nodes; this data is then input into the user screening and identification agent; user category information is obtained; and the target user is identified as a key user based on the user category information. This method can be widely applied in scenarios where key users are screened and identified during target business signing, such as when signing insurance contracts or reviewing claims. This key user screening and identification method can assist financial businesses like insurance or claims in identifying high-risk users in advance, preventing underwriting or claims institutions from incurring high business risks or losses. Furthermore, the user screening and identification agent, completed using ensemble learning, employs automated artificial intelligence processing for key user screening and identification, which not only improves the efficiency of key user identification but also reduces the misjudgment rate caused by relying solely on human experience, thereby improving the accuracy of key user identification.
[0032] Continue to refer to Figure 3 , Figure 3 yes Figure 2 A flowchart of a specific embodiment of step 202 shown includes: Step 301: Perform preliminary analysis on the multimodal service data to obtain the time span information involved in the generation process of the multimodal service data; Specifically, for example, multimodal business data from insurance application to claims settlement can be preliminarily analyzed to obtain information on the time span involved in the entire business data from application, underwriting, claims review, claims payment, and claims completion.
[0033] Step 302: Perform time-series decomposition on the multimodal service data based on the time span information; Specifically, the multimodal business data is decomposed into time sequences based on different business time nodes in the time span information.
[0034] Step 303: Mark the business time nodes corresponding to the time-series decomposition results as the dynamic indicator data.
[0035] Continue to refer to Figure 4 In some specific implementations, a step of performing integrated learning of user screening and identification agents is included before step 203. Figure 4 This is a flowchart of a specific embodiment of the user screening and identification agent ensemble learning in the key user screening and identification method described in this application, including: Step 401: Obtain multimodal business data of historical batch users and construct the training dataset used for ensemble learning; Specifically, for example, we can obtain multimodal business data of historical batches of insurance users to construct the training dataset used for ensemble learning.
[0036] Step 402: Based on the user classification label corresponding to each user, perform pre-classification processing on the training dataset to obtain the number of pre-classified categories and the pre-classification processing result of the training dataset; Specifically, since these are historical users, their business behavior is already known to a certain extent. Therefore, user classification labels can be pre-set for each user, and the training dataset can be pre-classified using the user classification labels.
[0037] Step 403: Input the training dataset into an ensemble learning model based on the improved AdaBoost algorithm, wherein the improved AdaBoost algorithm computationally integrates gradient boosting tree and logistic regression algorithms. In this embodiment, an ensemble learning model based on an improved AdaBoost algorithm is used for the model selected during ensemble learning training. Specifically, conventional AdaBoost ensemble learning integrates different weak classifiers to form a final strong classifier based on multiple rounds of iterative training, thereby achieving the training of the ensemble learning model. However, this application introduces gradient boosting trees and logistic regression algorithms on the basis of conventional AdaBoost ensemble learning. By using gradient boosting trees and logistic regression algorithms, when integrating different weak classifiers to form a final strong classifier, samples with greater differences receive more attention and correction, thereby reducing or ignoring the impact of weak noise on the ensemble learning results and improving the efficiency of ensemble learning.
[0038] Step 404: Set the user category output node in the output layer of the ensemble learning model according to the number of pre-classified categories in the training dataset; Specifically, based on prior classification experience, setting user category output nodes plays a certain supervisory role, thus limiting the classification output and classification categories of the ensemble learning model under certain quantitative criteria when training the ensemble learning model.
[0039] Step 405: Using the improved AdaBoost algorithm in the ensemble learning model, the training samples in the training dataset are classified and partitioned in multiple rounds to obtain the actual classification and partitioning results for each round. Specifically, the training dataset is input into an ensemble learning model based on the improved AdaBoost algorithm for ensemble learning training, and the actual classification results are obtained in each iteration.
[0040] Step 406: Compare the differences between the actual classification results and the pre-classification results in each round. Step 407: When the difference meets the preset conditions for completion of integrated learning, the user screening and identification agent with completed integrated learning is obtained.
[0041] Specifically, the pre-classification results are used as the validation domain for ensemble learning training to verify the final ensemble learning results, ensuring the high availability of the final user screening and identification agent.
[0042] Continue to refer to Figure 5 , Figure 5 yes Figure 4 A flowchart of a specific embodiment of step 404 shown includes: Step 501: Count the number of pre-classified categories in the training dataset; Step 502: Set an equal number of user category output nodes in the output layer of the ensemble learning model according to the number of pre-classified categories.
[0043] Specifically, by counting the number of pre-classified categories in the training dataset and setting an equal number of user category output nodes according to the number of pre-classified categories, the user screening and recognition agent that is finally integrated and learned conforms to the current classification settings, thereby improving the usability of the agent.
[0044] In this embodiment, two different verification methods can be freely selected for ensemble learning verification: a sampling-based verification method and a full verification data-based verification method.
[0045] Specifically, the sampling-based verification method, which involves sequentially comparing the differences between the actual classification results and the pre-classification results in each round, includes: obtaining the actual classification results of the current round; sampling several groups of pre-classification results from the pre-classification results; identifying whether all of the several groups of pre-classification results belong to a subset of the actual classification results of the current round; if any of the several groups of pre-classification results does not belong to a subset of the actual classification results of the current round, then the preset ensemble learning completion condition is not met, and the next round of obtaining actual classification results and comparing differences continues; subsequently, the step of obtaining the user screening and identification agent with completed ensemble learning when the differences meet the preset ensemble learning completion condition includes: if all of the several groups of pre-classification results belong to a subset of the actual classification results of the current round, then the preset ensemble learning completion condition is met; obtaining the classification processing parameters corresponding to the actual classification results of the current round, and setting the strong classifier corresponding to the classification processing parameters as the user screening and identification agent with completed ensemble learning.
[0046] Specifically, the verification method based on full-scale verification data includes the step of sequentially comparing the differences between the actual classification results and the pre-classification processing results in each round, comprising: obtaining the actual classification results of the current round; using a full-scale comparison method to identify the consistency between the actual classification results of the current round and the pre-classification processing results; if the consistency does not exceed a preset similarity threshold, then the preset ensemble learning completion condition is not met, and the acquisition of the actual classification results and difference comparison of the next round continues; subsequently, the step of obtaining the user screening and identification agent with ensemble learning completed when the difference meets the preset ensemble learning completion condition includes: if the consistency exceeds a preset similarity threshold, then the preset ensemble learning completion condition is met; obtaining the classification processing parameters corresponding to the actual classification results of the current round, and setting the strong classifier corresponding to the classification processing parameters as the user screening and identification agent with ensemble learning completed.
[0047] In this embodiment, by providing two different verification methods that can be freely selected during the verification of ensemble learning, the accuracy of the user screening and identification agent obtained after the ensemble learning is fully guaranteed.
[0048] Continue to refer to Figure 6 In some specific implementations, after step 407, a step of constructing the mapping relationship between the differentiated dynamic indicator data and the user category output nodes is also included. Figure 6This is a flowchart of a specific embodiment of the key user screening and identification method described in this application, which constructs the mapping relationship between differentiated dynamic indicator data and user category output nodes, including: Step 601: Obtain the dynamic indicator data for the user business data in all categories of the pre-classification processing results; Specifically, the dynamic indicator data acquisition can be performed on the user business data in all categories of the pre-classification processing results, or the dynamic indicator data acquisition can be performed according to the acquisition methods provided in steps 301 to 302.
[0049] Step 602: Through comprehensive comparison, determine the differentiated dynamic indicator data corresponding to each of the classification categories; Specifically, the comprehensive comparison includes first summarizing the dynamic indicator data contained in each category, and then, based on the summarization results, conducting pairwise comparisons to ultimately identify the differentiated dynamic indicator data between different categories.
[0050] Step 603: Based on the differentiated dynamic indicator data corresponding to each of the categories and the user category output nodes corresponding to each of the categories, construct the mapping relationship between the differentiated dynamic indicator data and the user category output nodes.
[0051] Specifically, after establishing the mapping relationship between differentiated dynamic indicator data and user category output nodes, when using the user screening and identification agent completed by the ensemble learning to identify user categories, it is only necessary to identify the differentiated dynamic indicator data to determine the user category of the target user.
[0052] Continue to refer to Figure 7 In some specific implementations, after step 203, a step of user category identification using a user screening and identification agent is also included. Figure 7 This is a flowchart of a specific embodiment of the key user screening and identification method described in this application, which utilizes a user screening and identification intelligent agent for user category identification, including: Step 701: Identify the differentiated dynamic indicator data contained in the dynamic indicator data; Step 702: Based on the mapping relationship between the differentiated dynamic indicator data and the user category output nodes, select the corresponding user category output nodes as the user category target output nodes. Step 703: Determine the user category information based on the output of the user category target output node.
[0053] In this embodiment, the key user screening and identification method can be widely applied to scenarios where key users are screened and identified during the signing of target business agreements. It involves acquiring multimodal business data provided by the target user; performing time-series feature engineering to obtain dynamic indicator data incorporating business time nodes; inputting this data into the user screening and identification intelligent agent; obtaining user category information; and identifying whether the target user is a key user based on the user category information. For example, during insurance contract signing or claims review, this key user screening and identification method can assist insurance or claims-related financial businesses in identifying high-risk users in advance, preventing underwriting or claims institutions from incurring high business risks or losses. Furthermore, the user screening and identification intelligent agent, completed using ensemble learning, employs automated artificial intelligence processing for key user screening and identification, which not only improves the efficiency of key user identification but also reduces the misjudgment rate caused by relying solely on human experience, thereby improving the accuracy of key user identification.
[0054] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence (AI) refers to the theories, methods, technologies, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.
[0055] Foundational technologies for artificial intelligence generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies mainly encompass computer vision, robotics, biometrics, speech processing, natural language processing, and machine learning / deep learning.
[0056] In this embodiment, the key user screening and identification method can be widely applied to scenarios where key users are screened and identified during the signing of target business agreements. It involves acquiring multimodal business data provided by the target user; performing time-series feature engineering to obtain dynamic indicator data incorporating business time nodes; inputting this data into the user screening and identification intelligent agent; obtaining user category information; and identifying whether the target user is a key user based on the user category information. For example, during insurance contract signing or claims review, this key user screening and identification method can assist insurance or claims-related financial businesses in identifying high-risk users in advance, preventing underwriting or claims institutions from incurring high business risks or losses. Furthermore, the user screening and identification intelligent agent, completed using ensemble learning, employs automated artificial intelligence processing for key user screening and identification, which not only improves the efficiency of key user identification but also reduces the misjudgment rate caused by relying solely on human experience, thereby improving the accuracy of key user identification.
[0057] Further reference Figure 8 As a response to the above Figure 2 The implementation of the method shown in this application provides an embodiment of a key user screening and identification device, which is similar to... Figure 2 Corresponding to the method embodiments shown, this device can be specifically applied to various electronic devices.
[0058] like Figure 8 As shown, the key user screening and identification device 800 described in this embodiment includes: a business data acquisition module 801, a dynamic indicator data acquisition module 802, a dynamic indicator data input module 803, a user category information output acquisition module 804, and a key user identification module 805. Wherein: The business data acquisition module 801 is used to acquire multimodal business data provided by the target user; The dynamic indicator data acquisition module 802 is used to perform time-series feature engineering processing on the multimodal business data to obtain dynamic indicator data that incorporates business time nodes; it is also used to acquire the dynamic indicator data for user business data in all categories of the pre-classification processing results respectively. The dynamic indicator data input module 803 is used to input the dynamic indicator data into the user screening and identification intelligent agent that has completed integrated learning; User category information output acquisition module 804 is used to acquire user category information output by the user screening and identification intelligent agent; The key user identification module 805 is used to identify whether the target user is a key user based on the user category information.
[0059] This application obtains multimodal business data provided by target users; performs time-series feature engineering to obtain dynamic indicator data incorporating business time nodes; inputs this data into a user screening and identification intelligent agent; obtains user category information; and identifies whether the target user is a key user based on the user category information. It can be widely applied in scenarios where key users are screened and identified during target business signing, such as when signing insurance contracts or reviewing claims. This key user screening and identification method can assist insurance or claims and other financial businesses in identifying high-risk users in advance, preventing underwriting or claims institutions from incurring high business risks or losses. Furthermore, the user screening and identification intelligent agent, completed using ensemble learning, employs automated artificial intelligence processing for key user screening and identification, which not only improves the efficiency of key user identification but also reduces the misjudgment rate caused by relying solely on human experience, thereby improving the accuracy of key user identification.
[0060] In this embodiment, the dynamic indicator data acquisition module 802 includes a time span information parsing unit, a time-series decomposition unit, and a dynamic indicator data generation unit. Wherein: The time span information parsing unit is used to perform preliminary parsing of the multimodal business data to obtain the time span information involved in the generation process of the multimodal business data; A time-series decomposition unit is used to perform time-series decomposition on the multimodal service data based on the time span information; The dynamic indicator data generation unit is used to mark the business time nodes corresponding to the time-series decomposition results as the dynamic indicator data.
[0061] In this embodiment, the key user screening and identification device 800 further includes a training dataset construction module, a pre-classification processing module, a learning and training input module, an output node setting module, an actual classification result acquisition module, a difference comparison module, and a user screening and identification agent acquisition module. Wherein: The training dataset building module is used to acquire multimodal business data of historical batch users and build the training dataset used for ensemble learning. The pre-classification processing module is used to pre-classify the training dataset according to the user classification label corresponding to each user, so as to obtain the number of pre-classified categories and the pre-classification processing result of the training dataset; The learning training input module is used to input the training dataset into an ensemble learning model based on the improved AdaBoost algorithm, wherein the improved AdaBoost algorithm computationally integrates gradient boosting tree and logistic regression algorithms. The output node setting module is used to set the user category output node in the output layer of the ensemble learning model according to the number of pre-classified categories in the training dataset; The actual classification result acquisition module is used to perform multiple rounds of iterative classification on the training samples in the training dataset using the improved AdaBoost algorithm in the ensemble learning model, and obtain the actual classification result for each round. The difference comparison module is used to compare the differences between the actual classification results and the pre-classification results in each round. The user screening and identification agent acquisition module is used to obtain the user screening and identification agent that has completed integrated learning when the difference meets the preset integrated learning completion conditions.
[0062] In this embodiment, the output node setting module includes a pre-classified category quantity statistics unit and an output node setting unit. Wherein: The pre-classification category count unit is used to count the number of pre-classification categories in the training dataset; The output node setting unit is used to set an equal number of user category output nodes in the output layer of the ensemble learning model according to the number of pre-classified categories.
[0063] In this embodiment, the difference comparison module includes a first difference comparison unit and a second difference comparison unit. Wherein: The first difference comparison unit is used to obtain the actual classification result of the current round; it is also used to sample several groups of pre-classification results from the pre-classification processing results; it is also used to identify whether the several groups of pre-classification results all belong to the classification subset of the actual classification result of the current round; it is also used to, if there are pre-classification results among the several groups of pre-classification results that do not belong to the classification subset of the actual classification result of the current round, then the preset ensemble learning completion condition is not met, and the next round of actual classification result acquisition and difference comparison is continued. The second difference comparison unit is used to obtain the actual classification result of the current round; it is also used to identify the consistency between the actual classification result of the current round and the pre-classification processing result by using a full comparison method; and it is also used to continue to obtain the actual classification result and difference comparison of the next round if the consistency does not exceed the preset similarity threshold, which does not meet the preset ensemble learning completion condition.
[0064] In this embodiment, the user screening and identification agent acquisition module includes a first user screening and identification agent acquisition unit and a second user screening and identification agent acquisition unit. Wherein: The first unit for obtaining the user screening and identification agent is used to satisfy the preset ensemble learning completion condition if all of the several pre-classification results belong to the classification subset of the actual classification results of the current round; it is also used to obtain the classification processing parameters corresponding to the actual classification results of the current round, and set the strong classifier corresponding to the classification processing parameters as the user screening and identification agent that has completed the ensemble learning. The second user screening and identification agent acquisition unit is used to satisfy the preset ensemble learning completion condition if the consistency exceeds the preset similarity threshold; it is also used to obtain the classification processing parameters corresponding to the actual classification result of the current round, and set the strong classifier corresponding to the classification processing parameters as the user screening and identification agent that has completed the ensemble learning.
[0065] In this embodiment, the key user screening and identification device 800 further includes a differentiated dynamic indicator data determination module and a mapping relationship construction module. Wherein: The Differentiated Dynamic Indicator Data Determination Module is used to determine the differentiated dynamic indicator data corresponding to each of the various categories through comprehensive comparison. The mapping relationship construction module is used to construct the mapping relationship between the differentiated dynamic indicator data and the user category output nodes based on the differentiated dynamic indicator data corresponding to each category and the user category output nodes corresponding to each category.
[0066] In this embodiment, the key user screening and identification device 800 further includes a differentiated dynamic indicator data identification module, a target output node screening module, and a user category information determination module. Wherein: A differentiated dynamic indicator data identification module is used to identify differentiated dynamic indicator data contained in the dynamic indicator data; The target output node filtering module is used to filter out the corresponding user category output nodes as user category target output nodes based on the mapping relationship between differentiated dynamic indicator data and user category output nodes. The user category information determination module is used to determine user category information based on the output results of the user category target output node.
[0067] 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 instructing related hardware through computer-readable instructions. These computer-readable instructions can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the methods described above. The aforementioned storage medium can be a non-volatile storage medium such as a magnetic disk, optical disk, or read-only memory (ROM), or random access memory (RAM).
[0068] It should be understood that although the steps in the flowcharts of the accompanying figures are shown sequentially as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the accompanying figures may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.
[0069] To address the aforementioned technical problems, embodiments of this application also provide a computer device. Please refer to [link / reference needed]. Figure 9 , Figure 9 This is a basic structural block diagram of the computer device in this embodiment.
[0070] The computer device 9 includes a memory 9a, a processor 9b, and a network interface 9c that are interconnected via a system bus. It should be noted that... Figure 9 Only a computer device 9 with component memory 9a, processor 9b, and network interface 9c is shown. However, it should be understood that it is not required to implement all the components shown, and more or fewer components may be implemented instead. Those skilled in the art will understand that the computer device described herein is a device capable of automatically performing numerical calculations and / or information processing according to pre-set or stored instructions. Its hardware includes, but is not limited to, microprocessors, application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), digital signal processors (DSPs), embedded devices, etc.
[0071] The computer device can be a desktop computer, laptop, handheld computer, or cloud server, etc. The computer device can interact with the user via a keyboard, mouse, remote control, touchpad, or voice control.
[0072] The memory 9a includes at least one type of readable storage medium, including flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 9a may be an internal storage unit of the computer device 9, such as the hard disk or memory of the computer device 9. In other embodiments, the memory 9a may also be an external storage device of the computer device 9, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the computer device 9. Of course, the memory 9a may also include both the internal storage unit and its external storage device of the computer device 9. In this embodiment, the memory 9a is typically used to store the operating system and various application software installed on the computer device 9, such as computer-readable instructions for a key user screening and identification method. In addition, the memory 9a can also be used to temporarily store various types of data that have been output or will be output.
[0073] In some embodiments, the processor 9b may be a central processing unit (CPU), controller, microcontroller, microprocessor, or other data processing chip. The processor 9b is typically used to control the overall operation of the computer device 9. In this embodiment, the processor 9b is used to execute computer-readable instructions stored in the memory 9a or to process data, such as executing computer-readable instructions for the key user screening and identification method described above.
[0074] The network interface 9c may include a wireless network interface or a wired network interface, which is typically used to establish communication connections between the computer device 9 and other electronic devices.
[0075] The computer device proposed in this embodiment belongs to the field of artificial intelligence technology and is applied in scenarios where key users are screened and identified during the signing of target business agreements. This application obtains multimodal business data provided by target users; performs time-series feature engineering processing to obtain dynamic indicator data incorporating business time nodes; inputs this data into a user screening and identification intelligent agent; obtains user category information; and identifies whether a target user is a key user based on the user category information. For example, when signing insurance contracts or reviewing claims, this key user screening and identification method can assist in insurance or claims-related financial businesses by pre-identifying high-risk users, preventing underwriting or claims institutions from incurring high business risks or losses. Furthermore, the user screening and identification intelligent agent, completed using ensemble learning, employs automated artificial intelligence processing for key user screening and identification, which not only improves the efficiency of key user identification but also reduces the misjudgment rate caused by relying solely on human experience, thereby improving the accuracy of key user identification.
[0076] This application also provides another embodiment, namely, a computer-readable storage medium storing computer-readable instructions that can be executed by a processor to cause the processor to perform the steps of the key user screening and identification method described above.
[0077] The computer-readable storage medium proposed in this embodiment belongs to the field of artificial intelligence technology and is applied to scenarios where key users are screened and identified during the signing of target business agreements. This application obtains multimodal business data provided by the target user; performs time-series feature engineering processing to obtain dynamic indicator data incorporating business time nodes; inputs this data into a user screening and identification intelligent agent; obtains user category information; and identifies whether the target user is a key user based on the user category information. For example, when signing insurance contracts or reviewing claims, this key user screening and identification method can assist insurance or claims-related financial businesses in identifying high-risk users in advance, preventing underwriting or claims institutions from incurring high business risks or losses. Furthermore, the user screening and identification intelligent agent, completed using ensemble learning, employs automated artificial intelligence processing to screen and identify key users, which not only improves the efficiency of key user identification but also reduces the misjudgment rate caused by relying solely on human experience to identify key users, thereby improving the accuracy of key user identification.
[0078] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of this application.
[0079] Obviously, the embodiments described above are only some embodiments of this application, not all embodiments. The accompanying drawings show preferred embodiments of this application, but do not limit the patent scope of this application. This application can be implemented in many different forms; rather, the purpose of providing these embodiments is to make the disclosure of this application more thorough and comprehensive. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing specific embodiments, or make equivalent substitutions for some of the technical features. Any equivalent structures made using the content of this application's specification and drawings, directly or indirectly applied to other related technical fields, are similarly within the scope of patent protection of this application. Software tools or components not belonging to this company that appear in the embodiments of this application are merely illustrative examples and do not represent actual use.
Claims
1. A method for screening and identifying key users, characterized in that, Includes the following steps: Obtain multimodal business data provided by the target user; The multimodal business data is subjected to time-series feature engineering to obtain dynamic indicator data that incorporates business time nodes; The dynamic indicator data is input into the user screening and identification intelligent agent that has completed the integrated learning; Obtain the user category information output by the user screening and identification agent; Based on the user category information, identify whether the target user is a key user.
2. The key user screening and identification method according to claim 1, characterized in that, Before performing the step of inputting the dynamic indicator data into the user screening and identification agent completed by ensemble learning, the method further includes: Acquire multimodal business data of historical batch users to construct the training dataset used for ensemble learning; Based on the user classification label corresponding to each user, the training dataset is pre-classified to obtain the number of pre-classified categories and the pre-classification result of the training dataset; The training dataset is input into an ensemble learning model based on an improved AdaBoost algorithm, wherein the improved AdaBoost algorithm computationally integrates gradient boosting trees and logistic regression algorithms. In the output layer of the ensemble learning model, user category output nodes are set according to the number of pre-classified categories in the training dataset; The improved AdaBoost algorithm in the ensemble learning model is used to perform multiple rounds of iterative classification on the training samples in the training dataset to obtain the actual classification results in each round. The differences between the actual classification results and the pre-classification results are compared sequentially in each round. When the differences meet the preset conditions for completion of ensemble learning, the user screening and identification agent that has completed ensemble learning is obtained.
3. The key user screening and identification method according to claim 2, characterized in that, The step of setting the user category output node in the output layer of the ensemble learning model according to the number of pre-classified categories in the training dataset specifically includes: Count the number of pre-classified categories in the training dataset; Set an equal number of user category output nodes in the output layer of the ensemble learning model according to the number of pre-classified categories.
4. The key user screening and identification method according to claim 2, characterized in that, The step of sequentially comparing the differences between the actual classification results and the pre-classification results in each round includes: Obtain the actual classification result for the current round; Several groups of pre-classification results are extracted from the pre-classification results using a sampling method; Identify whether each of the pre-classification results belongs to a subset of the actual classification results in the current round; If any of the pre-classification results in the aforementioned groups does not belong to a subset of the actual classification results in the current round, then the preset ensemble learning completion condition is not met, and the acquisition and difference comparison of the actual classification results in the next round will continue. The step of obtaining the user screening and identification agent with completed ensemble learning when the differences satisfy the preset ensemble learning completion conditions includes: If all of the aforementioned pre-classification results belong to a subset of the actual classification results in the current round, then the preset ensemble learning completion condition is met. Obtain the classification processing parameters corresponding to the actual classification result of the current round, and set the strong classifier corresponding to the classification processing parameters as the user screening and recognition agent completed by the ensemble learning.
5. The key user screening and identification method according to claim 2, characterized in that, The step of sequentially comparing the differences between the actual classification results and the pre-classification results in each round includes: Obtain the actual classification result for the current round; A full comparison method is used to identify the consistency between the actual classification result of the current round and the pre-classification result; If the consistency does not exceed the preset similarity threshold, the preset ensemble learning completion condition is not met, and the next round of actual classification results acquisition and difference comparison will continue. The step of obtaining the user screening and identification agent with completed ensemble learning when the differences satisfy the preset ensemble learning completion conditions includes: If the consistency exceeds a preset similarity threshold, then the preset ensemble learning completion condition is met. Obtain the classification processing parameters corresponding to the actual classification result of the current round, and set the strong classifier corresponding to the classification processing parameters as the user screening and recognition agent completed by the ensemble learning.
6. The key user screening and identification method according to claim 2, characterized in that, After performing the step of obtaining the user screening and identification agent with completed ensemble learning when the differences satisfy the preset ensemble learning completion condition, the method further includes: The dynamic indicator data is obtained for user business data in all categories of the pre-classification results; By comprehensively comparing the data, the differentiated dynamic indicators corresponding to each of the categories were determined. Based on the differentiated dynamic indicator data corresponding to each category and the user category output nodes corresponding to each category, a mapping relationship between the differentiated dynamic indicator data and the user category output nodes is constructed.
7. The key user screening and identification method according to claim 6, characterized in that, After performing the step of inputting the dynamic indicator data into the user screening and identification agent completed by ensemble learning, the method further includes: Identify the differentiated dynamic indicator data contained in the dynamic indicator data; Based on the mapping relationship between differentiated dynamic indicator data and user category output nodes, the corresponding user category output nodes are selected as user category target output nodes; The user category information is determined by the output of the target output node for the user category.
8. A key user screening and identification device, characterized in that, include: The business data acquisition module is used to acquire multimodal business data provided by the target user; The dynamic indicator data acquisition module is used to perform time-series feature engineering processing on the multimodal business data to obtain dynamic indicator data that incorporates business time nodes. The dynamic indicator data input module is used to input the dynamic indicator data into the user screening and identification intelligent agent that has completed integrated learning; The user category information output acquisition module is used to acquire the user category information output by the user screening and identification intelligent agent. The key user identification module is used to identify whether the target user is a key user based on the user category information.
9. A computer device, characterized in that, The device includes a memory and a processor, wherein the memory stores computer-readable instructions, and the processor executes the computer-readable instructions to implement the steps of the key user screening and identification method as described in any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-readable instructions, which, when executed by a processor, implement the steps of the key user screening and identification method as described in any one of claims 1 to 7.