Data intelligent updating iteration method and device, equipment and medium

Abstract

The invention relates to the technical field of medical treatment and data processing, and discloses an intelligent data updating iteration method, device and equipment and a medium. The method comprises the steps that a basic database is obtained, feature values in the basic database are extracted, a feature value retrieval model is established, initial feature values in the feature value retrieval model are extracted, and the initial feature values are obtained; the initial feature value is set as an initial regression case, the initial regression case is trained based on a step length method to obtain a simplified regression case, the simplified regression case iterates the initial regression case, and the simplified regression case is regressed to the feature value retrieval model. According to the method, the regression case extraction accuracy is improved, the regression duration is shortened, the algorithm execution efficiency is improved, and the characteristics and requirements of different regression cases are better adapted, so that the generalization ability and robustness of the algorithm are improved while the accuracy is kept, the data sampling frequency and the iteration speed can be controlled, and the method is suitable for large-scale popularization and application. And the calculation efficiency is improved while the data precision is ensured.

Description

Technical Field

[0001] The present invention relates to the fields of medical technology and data processing technology, and in particular to a data intelligent update and iteration method, device, equipment and medium. Background Art

[0002] With the expansion of business, the continuous iteration of requirements and the upgrading and transformation of systems, the switch between the old and new systems has not been completed. The functional coupling of the new and old links has led to frequent releases of various projects, and the manpower investment required for joint debugging is very huge. In the past, iterative processing was located before each release, mainly selecting cases of historical function points and related coordination system function points from the regression use case library for regression testing and verification. Due to the large number of cases involved according to the functional modules, regression takes a long time. Summary of the invention

[0003] The present invention provides a data intelligent update and iteration method, device, computer equipment and medium to solve the technical problems that iterative regression involves a large number of cases and takes a long time to regress.

[0004] In a first aspect, a data intelligent update and iteration method is provided, comprising:

[0005] Obtain a basic database, extract characteristic values ​​from the basic database and establish a characteristic value retrieval model;

[0006] Extract the initial eigenvalues ​​in the eigenvalue retrieval model and set the initial eigenvalues ​​as the initial regression cases;

[0007] Based on the step-size method, the initial regression cases are trained to obtain the simplified regression cases;

[0008] The reduced regression cases are iterated on the initial regression cases, and the reduced regression cases are regressed on the feature value retrieval model.

[0009] In a second aspect, a data intelligent update and iteration device is provided, comprising:

[0010] An extraction module is used to obtain a basic database, extract characteristic values ​​in the basic database and establish a characteristic value retrieval model;

[0011] An extraction module, used for extracting initial eigenvalues ​​in the eigenvalue retrieval model and setting the initial eigenvalues ​​as initial regression cases;

[0012] A training module, used to train the initial regression cases based on the step-size method to obtain the simplified regression cases;

[0013] The iteration module is used to iterate the simplified regression cases to the initial regression cases, and to regress the simplified regression cases to the feature value retrieval model.

[0014] In a third aspect, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned data intelligent update and iteration method when executing the computer program.

[0015] In a fourth aspect, a computer-readable storage medium is provided, which stores a computer program, and when the computer program is executed by a processor, the steps of the above-mentioned data intelligent update and iteration method are implemented.

[0016] In the scheme implemented by the above-mentioned data intelligent update and iteration method, device, computer equipment and storage medium, the basic database can be obtained, the eigenvalues ​​in the basic database can be extracted and the eigenvalue retrieval model can be established, the initial eigenvalues ​​in the eigenvalue retrieval model can be extracted, and the initial eigenvalues ​​can be set as the initial regression case, the initial regression case can be trained based on the step length method to obtain a simplified regression case, the simplified regression case can be iterated to the initial regression case, and the simplified regression case can be regressed to the eigenvalue retrieval model. This can improve the accuracy of regression case extraction, shorten the regression time, improve the algorithm execution efficiency, better adapt to the characteristics and requirements of different regression cases, thereby maintaining accuracy while improving the generalization ability and robustness of the algorithm, and can also control the frequency of data sampling and the speed of iteration, and improve the calculation efficiency while ensuring data accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative labor.

[0018] Figure 1 This is a schematic diagram of an application environment of a data intelligent update and iteration method in one embodiment of the present invention;

[0019] Figure 2 It is a flow chart of a method for intelligent data update iteration in one embodiment of the present invention;

[0020] Figure 3 It is a structural diagram of a data intelligent updating and iteration device in one embodiment of the present invention;

[0021] Figure 4 is a schematic diagram of a structure of a computer device in one embodiment of the present invention;

[0022] Figure 5 It is another structural schematic diagram of a computer device in one embodiment of the present invention. DETAILED DESCRIPTION

[0023] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0024] The data intelligent update and iteration method provided by the embodiment of the present invention can be applied in the following aspects: Figure 1 In an application environment, the client communicates with the server through a network. The server can obtain a basic database, extract eigenvalues ​​in the basic database and establish a eigenvalue retrieval model, extract initial eigenvalues ​​in the eigenvalue retrieval model, and set the initial eigenvalues ​​as initial regression cases, train the initial regression cases based on the step-length method to obtain a simplified regression case, iterate the simplified regression case to the initial regression case, and regress the simplified regression case to the eigenvalue retrieval model. In the present invention, for complex insurance entities such as medical insurance business, an entity alignment optimization data intelligent update and iteration solution can be used, first by obtaining a basic database, extracting eigenvalues ​​in the basic database and establishing a eigenvalue retrieval model, extracting initial eigenvalues ​​in the eigenvalue retrieval model, and setting the initial eigenvalues ​​as initial regression cases, training the initial regression case based on the step-length method to obtain a simplified regression case, iterating the simplified regression case to the initial regression case, and regressing the simplified regression case to the eigenvalue retrieval model. The accuracy of regression case extraction is improved, the effect of regression time is shortened, the algorithm execution efficiency is improved, and the characteristics and requirements of different regression cases are better adapted to improve the generalization ability and robustness of the algorithm while maintaining accuracy. The frequency of data sampling and the speed of iteration can also be controlled to improve the computing efficiency while ensuring data accuracy. Among them, the client can be but is not limited to various personal computers, laptops, smart phones, tablet computers and portable wearable devices. The server can be implemented with an independent server or a server cluster composed of multiple servers. The present invention is described in detail below through specific embodiments.

[0025] See also Figure 2 As shown, Figure 2 A flowchart of a data intelligent update and iteration method provided by an embodiment of the present invention includes the following steps:

[0026] S10: Obtain the basic database, extract the characteristic values ​​in the basic database and establish characteristic value retrieval

[0027] Model;

[0028] ·In this embodiment, the basic database is used to store business-related labels and data. In this embodiment, in order to solve the technical problem of a large number of cases involved in iterative regression, the basic database is classified to facilitate the overall planning of the data content in the basic database. By dividing the data in the basic database into different categories, the data can be organized and managed more efficiently. The classification of data in the basic database helps to better understand the structure and content of the data in the basic database, making the subsequent storage and retrieval of data in the basic database simpler and faster. Through the classified data in the basic database, bottlenecks and problems in the data implementation process can be quickly discovered, and corresponding improvement measures can be taken to improve work efficiency. Data intelligent update and iteration are applicable to both financial and medical scenarios.

[0029] ·5 Specifically, the basic database includes at least a requirement classification table, a use case classification table, and a feature value relationship table. The requirement classification table is established by sorting out the requirement documents in the insurance policy system and setting label classification according to the fields of unique document number, requirement summary, requirement proposer, requirement coordinator, and business identification label. For example, the data sample is: insurance policy preservation-insurance total policy return-insurance policy system-financial documents

[0030] System-Refund; sort out the unique use case number, use case category, use case test point, and use case in the policy system 10 Detailed information, field settings for use case identification tags, label classification, establish a use case classification table, and store it in the basic database, for example, insurance policy-voucher verification-verify policy refund voucher changes-case step expected results-involved refund; associate the relationship between each requirement in the requirement classification table and the use case type in the use case classification table, and establish a feature value relationship table, store the feature value relationship table in the basic database, the feature value relationship table mainly

[0031] It stores the relationship between each requirement and use case type, including: business identification label and use case identification label 15 Signature, and the average verification workload for each use case type, for example: Refund-involving refund-0.5 man-days;

[0032] Policy cancellation-non-fee-related scenario-0.2 person-days.

[0033] After the basic database is established, the feature values ​​in the basic database are extracted. This extraction is to extract all the feature values ​​in the basic database and establish a feature value retrieval model for all the extracted feature values.

[0034] The eigenvalue retrieval model facilitates subsequent regression calculations and avoids repeated use of large model data. 20 This solves the problem of poor program performance caused by excessive data during calculation, which is helpful for the subsequent data in the basic database.

[0035] The retrieval of data becomes simpler and faster, and the bottlenecks and problems in the data implementation process can be quickly discovered, and corresponding improvement measures can be taken to improve work efficiency. Among them, the characteristic value includes at least: unique document number, requirement summary, requirement proposer, requirement joint debugging party, business identification tag, unique use case number, use case category, use case test point, use case details information, and use case identification tag.

[0036] S20: extracting the initial eigenvalue in the eigenvalue retrieval model, and setting the initial eigenvalue as the initial regression case;

[0037] It should be noted that in the eigenvalue retrieval model, the eigenvalues ​​of the training regression are randomly extracted and set as the initial eigenvalues, where the initial eigenvalues ​​serve as the initial parameter values ​​in the subsequent step method calculations, and multiple initial eigenvalues ​​form the initial regression case. Random extraction can avoid selection bias and ensure that the initial eigenvalues ​​can simply represent all eigenvalues. By randomly extracting eigenvalues, the characteristics of the population can be more accurately reflected, the authenticity of subsequent training results can be increased, and it helps to ensure the accuracy and reliability of the training results. Randomly extracted eigenvalues ​​usually have higher statistical power, which means that they can more accurately detect real differences or relationships in the population, and also help avoid data leakage. By ensuring that the initial eigenvalues ​​are not seen by the model during the training process, the generalization ability of the model can be evaluated. Random extraction can simplify the data processing process because smaller, more representative data sets can be processed more easily, which can save time and computing resources while still obtaining meaningful analysis results.

[0038] S30: Train the initial regression case based on the step-size method to obtain a simplified regression case;

[0039] It should be noted that the step size method mainly refers to the adjustment amplitude taken each time the model data is updated during the optimization process. The step size controls the speed at which the algorithm approaches the optimal solution during gradient descent. The step size is mainly the amplitude of parameter update at each iteration during the optimization process. Among them, the step size method is divided into three categories: fixed step size method, adaptive step size method, and line search method.

[0040] The fixed step size method is that in each iteration, the step size is kept at a fixed value, which is suitable for some simple optimization problems or when the nature of the problem allows the use of a fixed step size. However, the fixed step size method may not achieve the optimal convergence speed and effect in all cases.

[0041] The adaptive step size method dynamically adjusts the step size according to the performance of the algorithm during the iteration process. This method is usually more flexible and efficient, and the appropriate step size is selected according to the specific situation of the problem and the state of the algorithm. The following are some common adaptive step size methods: Gradient-based adaptive step size method: This method adjusts the step size according to the size and direction of the gradient. For example, when the gradient is large, a smaller step size can be selected to avoid oscillation; when the gradient is small, a larger step size can be selected to accelerate convergence; Adaptive step size method based on historical information: This method uses historical gradient information to adjust the step size. For example, the AdaGrad algorithm adjusts the step size of each parameter according to the sum of the squares of the historical gradients, so that the step size is larger in the direction of the smaller gradient and smaller in the direction of the larger gradient; Adaptive step size method based on momentum: This method combines the concept of momentum and adjusts the step size by considering the gradient direction of the previous iterations. For example, both RMSProp and Adam algorithms use the concept of momentum and adjust the step size according to the weighted average of the sum of the squares of the historical gradients.

[0042] The line search method is a more accurate step size calculation method. It updates the parameters by searching for an optimal step size in each iteration. The line search method usually includes two types: exact line search and inexact line search. Exact line search: Exact line search aims to find the step size that minimizes the objective function value by solving a one-dimensional optimization problem, that is, finding a parameter α that minimizes f(x_k+αd_k), where x_k is the current parameter value, d_k is the search direction, and α is the step size. Inexact line search: Inexact line search aims to find a step size that satisfies certain conditions, rather than the optimal step size. These conditions usually include Armijo conditions, Wolfe conditions, or Goldstein conditions. The computational cost of inexact line search is low, and it can usually improve the efficiency of the algorithm while ensuring the convergence of the algorithm.

[0043] Specifically, in this embodiment, the line search method is mainly used, by finding a step size α so that the objective function f(x+αd) reaches the minimum value in a given search direction d, that is, finding a parameter α that minimizes the objective function value. By determining the eigenvalue retrieval direction and the step size factor extracted from the eigenvalue relationship table, based on the step size method, cyclic training is performed in combination with the eigenvalue retrieval direction, the step size factor and the initial regression case until the training result reaches a preset threshold, and a streamlined regression case is obtained. Exact line search can find a step size that minimizes the objective function value (or approximately minimizes it within a given accuracy), so that the algorithm has better convergence in theory. Exact line search is usually suitable for optimization problems that require high step size accuracy.

[0044] Parameter α Search Method: Determine a search direction d, which can be the gradient direction (for gradient descent method), the negative gradient direction (for the steepest descent method), or any other valid search direction. Select an initial regression case [α_min, α_max] that contains the optimal step size. The initial regression case can be determined according to the specific situation of the problem and the characteristics of the algorithm. Within the initial regression case, continuously narrow the search interval through a certain method (such as interpolation method, golden section method, Fibonacci search, etc.) until a step size α_opt that meets a certain accuracy requirement is found. This step size should make the objective function f(x + α_optd) reach the minimum value (or approximate minimum value) in the given search direction d.

[0045] Specifically, the interpolation method selects a certain number of points within the initial regression case of the search, calculates the eigenvalues of these points, and constructs an interpolation polynomial using these eigenvalues to approximate the objective function. Then, determine the optimal step size by solving the minimum point of the interpolation polynomial. Commonly used interpolation methods include: one-point quadratic interpolation method, two-point quadratic interpolation method, three-point quadratic interpolation method, two-point cubic interpolation method, etc. One-point quadratic interpolation method: Also known as Newton's method, it uses eigenvalue and first derivative information to construct an interpolation polynomial and solve its minimum point. Two-point quadratic interpolation method: Includes the secant method, which uses the function values of two points to construct an interpolation polynomial and determines a new search direction or step size by solving the root of the interpolation polynomial. Three-point quadratic interpolation method, two-point cubic interpolation method, etc.: These methods use more points to construct interpolation polynomials of higher degrees to improve the accuracy of approximation. However, it should be noted that too high a degree of the interpolation polynomial may lead to numerical instability or the generation of multiple extreme points, so it needs to be carefully selected.

[0046] The golden section method approximates the optimal solution by continuously narrowing the interval containing the minimum point within the initial regression case of the search. Only one function value needs to be calculated in each iteration, so the calculation efficiency is relatively high. The specific steps of the golden section method are as follows: 1. Determine the initial regression case [a, b] and the allowable error ε; 2. Calculate the golden section ratio t = (√5 - 1) / 2 ≈ 0.618; 3. Take two initial points p and q within the initial regression case [a, b], where p = a + (1 - t)(b - a) and q = a + t(b - a); 4. Calculate the function values f(p) and f(q) at the initial points p and q; 5. Update the initial regression case [a, b] according to the size relationship of the function values. If f(p) < f(q), the new search interval is [a, q]; otherwise, the new initial regression case is [p, b] (i.e., the refined regression case is [p, b]); 6. Repeat steps 3 to 5 until the length of the initial regression case is less than the allowable error ε or the predetermined number of iterations is reached.

[0047] Fibonacci search is determined by using the Fibonacci sequence. Fibonacci search is similar to the golden section method and is also an optimization method based on interval contraction. The split point of the initial regression case is determined by using the Fibonacci sequence, and the optimal solution is approached by continuously shrinking the initial regression case. The specific steps of Fibonacci search are similar to the golden section method, except that the selection of the split point is determined by the Fibonacci sequence.

[0048] In one implementation of this embodiment, assuming that the model threshold is 1 person-day, the optimal step length method indicator formula χ k+1 = x k +α k d k The initial point x0 is the business identification label corresponding to this demand, which is the selected parameter, and the step factor χ k That is, the business identification label corresponding to all the subordinate use case identification labels is searched in all the eigenvalue relationship tables. The training is based on the optimal step length method indicator formula and the initial regression case is repeatedly cycled to make the indicator result decrease and converge. With the continuous extraction of cases to be regressed, when χ k+1 When the sum of the overall regression workload is close to 1, the iteration is stopped and a new batch of regression cases is obtained. Then the total execution workload required for the new batch of regression cases is 1 man-day, that is, one regression working day, and the new batch of regression cases (streamlined regression cases) all correspond to the requirements of this iteration and include cases of related adjusters.

[0049] S40: iterate the simplified regression case to the initial regression case, and regress the simplified regression case to the feature value retrieval model.

[0050] According to the regression results, the technical problems can be gradually solved by repeated calculations, updating and iterating, and optimizing data. Through continuous updating and iteration, the algorithm can gradually adjust the eigenvalue retrieval model data to better fit the data, thereby improving the prediction accuracy of the eigenvalue retrieval model. This involves a large amount of data input, model training, and result evaluation, and continuously optimizes the model performance through iteration. Furthermore, through iterative testing and adjustment of algorithm parameters, the execution efficiency of the algorithm can be gradually optimized, computing time and resource consumption can be reduced, and the performance and practicality of the algorithm can be significantly improved.

[0051] After the step S40, the present embodiment further includes: extracting simplified feature labels of the simplified regression model and extracting initial feature labels of the initial regression case; judging whether the scope of the initial feature labels completely covers the simplified feature labels; if completely covered, the simplified regression model training is successful; if partially covered or not covered at all, the simplified regression model training fails, and the step of re-executing the step of training the initial regression case based on the step length method to obtain the simplified regression case.

[0052] Specifically, the model self-corrects: If the business identification tag passed in is brand new or other abnormal situations, the model will always return 0, resulting in failure to extract the initial regression case. At this time, the above formula needs to be adjusted. The alternative is to identify the label based on the historical use case, perform idempotent collision within a certain timeout period, and stop when more than 1 output is obtained. For example, the eigenvalue relationship table does not have the business identification tag [policyholder change], but the use case identification tag contains [policyholder new insurance policy] and [policyholder cross-order discount]. By disassembling the newly added business identification tags, preliminary calculations can obtain several regression cases related to the policyholder for recommendation. Then manual intervention is required to correct the label and determine whether it is valuable to regress.

[0053] After the step S40, this embodiment further includes: extracting initial feature labels of the initial regression cases; querying the feature value retrieval model for redundant regression cases with feature value labels other than the initial regression cases according to the feature labels, wherein the redundant regression cases do not include the initial regression cases; and eliminating redundant regression cases.

[0054] Specifically, after iterative extraction of the initial regression cases, the remaining regression cases under the [Cancelation Refund] business identification label are redundant cases and need to be eliminated. The model threshold can be continuously adjusted to dynamically extract batch regression cases based on the manpower input of the regressive work and the size of the iterative version content.

[0055] In each subsequent requirement review and case writing, the changes and impact points will be labeled and stored in the database. As the basic database is continuously enriched, this embodiment can be used as a tool to extract related impact points and recommend regression cases before each launch, which not only controls the manpower input, but also can quickly extract and process the regression cases with the highest correlation, ensuring efficiency while ensuring the quality of the launch. Each system performs regression according to its own experience and understanding. When encountering joint debugging requirements, it often either regresses the entire system or omits regression function points due to lack of time, resulting in uncontrollable online quality and very easy production accidents. After using this tool, although there is an additional manpower maintenance cost for the preliminary preparation work, the left shift of the test makes the subsequent verification coverage of function points more comprehensive, greatly reducing the risk of related system impact and improving the quality of product iteration testing.

[0056] The data intelligent update and iteration method provided by the present invention can be applied to data update of replacement of old and new systems in various application scenarios or in the scope of medical insurance business. For example, in the application field of data intelligent update and iteration, the update of the system often involves a large number of cases and the regression takes a long time. The data intelligent update and iteration method can improve the accuracy of regression case extraction and shorten the regression time.

[0057] It can be seen that in the above scheme, for complex insurance entities such as medical insurance business, firstly, by obtaining the basic database, extracting the eigenvalues ​​in the basic database and establishing the eigenvalue retrieval model, extracting the initial eigenvalues ​​in the eigenvalue retrieval model, and setting the initial eigenvalues ​​as the initial regression case, the initial regression case is trained based on the step length method to obtain a simplified regression case, the simplified regression case is iterated to the initial regression case, and the simplified regression case is regressed to the eigenvalue retrieval model. This improves the accuracy of regression case extraction, shortens the regression time, improves the algorithm execution efficiency, and better adapts to the characteristics and requirements of different regression cases, thereby maintaining accuracy while improving the generalization ability and robustness of the algorithm, and can also control the frequency of data sampling and the speed of iteration, and improve the calculation efficiency while ensuring data accuracy.

[0058] It should be understood that the order of execution of the steps in the above embodiment does not necessarily mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present invention.

[0059] In one embodiment, a data intelligent updating and iteration device is provided, and the data intelligent updating and iteration device corresponds one-to-one to the data intelligent updating and iteration method in the above embodiment. Figure 3 As shown, the data intelligent update and iteration device includes an extraction module 101, an extraction module 102, a training module 103, and an iteration module 104. The detailed description of each functional module is as follows:

[0060] Extraction module 101, used to obtain a basic database, extract characteristic values ​​in the basic database and establish a characteristic value retrieval model;

[0061] An extraction module 102 is used to extract the initial eigenvalues ​​in the eigenvalue retrieval model and set the initial eigenvalues ​​as initial regression cases;

[0062] A training module 103, used for training the initial regression cases based on the step length method to obtain simplified regression cases;

[0063] The iteration module 104 is used to iterate the simplified regression case to the initial regression case, and regress the simplified regression case to the feature value retrieval model.

[0064] In one embodiment, the device further comprises:

[0065] The label classification module is used to sort out the demand documents in the insurance policy system, and to set label classification according to the fields of unique document number, demand summary, demand proposer, demand coordinator, and business identification label to establish a demand classification table, and store it in the basic database.

[0066] In one embodiment, the device further comprises:

[0067] The use case classification module is used to sort out the fields of the unique use case number, use case category, use case test point, use case detail information, and use case identification tag in the insurance policy system, set label classification, establish a use case classification table, and store it in the basic database.

[0068] In one embodiment, the device further comprises:

[0069] The association module is used to associate the relationship between each requirement in the requirement classification table and the use case type in the use case classification table, and to establish a characteristic value relationship table, and store the characteristic value relationship table in the basic database.

[0070] In one embodiment, the training module 103 is specifically configured to:

[0071] Determine the eigenvalue retrieval direction and the step factor extracted from the eigenvalue relationship table;

[0072] Based on the step size method, cyclic training is performed in combination with the eigenvalue retrieval direction, step size factor and initial regression case until the training result reaches the preset threshold to obtain a streamlined regression case.

[0073] In one embodiment, the training module 103 is further configured to:

[0074] Extracting reduced feature labels for reduced regression models and extracting initial feature labels for initial regression cases;

[0075] Determine whether the scope of the initial feature labels completely covers the simplified feature labels;

[0076] If it is fully covered, the streamlined regression model training is successful;

[0077] If it is partially covered or not covered at all, the training of the streamlined regression model fails, and the steps are re-executed to train the initial regression cases based on the step size method to obtain the streamlined regression cases.

[0078] In one embodiment, the training module 103 is further configured to:

[0079] Extract initial feature labels for initial regression cases;

[0080] According to the feature label, the feature value retrieves redundant regression cases with feature value labels other than the initial regression case in the model, wherein the redundant regression cases do not include the initial regression case;

[0081] Eliminate redundant regression cases.

[0082] The present invention provides a data intelligent update and iteration device, which first obtains a basic database, extracts eigenvalues ​​from the basic database and establishes an eigenvalue retrieval model, extracts initial eigenvalues ​​in the eigenvalue retrieval model, sets the initial eigenvalues ​​as initial regression cases, trains the initial regression cases based on the step length method to obtain simplified regression cases, iterates the simplified regression cases to the initial regression cases, and regresses the simplified regression cases to the eigenvalue retrieval model. This improves the accuracy of regression case extraction, shortens the effect of regression time, improves algorithm execution efficiency, and better adapts to the characteristics and requirements of different regression cases, thereby improving the generalization ability and robustness of the algorithm while maintaining accuracy, and can also control the frequency of data sampling and the speed of iteration, and improves calculation efficiency while ensuring data accuracy.

[0083] For the specific definition of the data intelligent update and iteration device, please refer to the definition of the data intelligent update and iteration method above, which will not be repeated here. Each module in the above-mentioned data intelligent update and iteration device can be implemented in whole or in part by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0084] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 4 As shown. The computer device includes a processor, a memory, a network interface and a database connected via a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile and / or volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external client via a network connection. When the computer program is executed by the processor, it realizes the functions or steps of a data intelligent update iteration method service side.

[0085] In one embodiment, a computer device is provided. The computer device may be a client, and its internal structure diagram may be as follows: Figure 5As shown. The computer device includes a processor, a memory, a network interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external server through a network connection. When the computer program is executed by the processor, it realizes the functions or steps of a client side of a data intelligent update iteration method

[0086] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the following steps when executing the computer program:

[0087] Obtain a basic database, extract characteristic values ​​from the basic database and establish a characteristic value retrieval model;

[0088] Extract the initial eigenvalues ​​in the eigenvalue retrieval model and set the initial eigenvalues ​​as the initial regression cases;

[0089] Based on the step-size method, the initial regression cases are trained to obtain the simplified regression cases;

[0090] The reduced regression cases are iterated on the initial regression cases, and the reduced regression cases are regressed on the feature value retrieval model.

[0091] In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

[0092] Obtain a basic database, extract characteristic values ​​from the basic database and establish a characteristic value retrieval model;

[0093] Extract the initial eigenvalues ​​in the eigenvalue retrieval model and set the initial eigenvalues ​​as the initial regression cases;

[0094] Based on the step-size method, the initial regression cases are trained to obtain the simplified regression cases;

[0095] The reduced regression cases are iterated on the initial regression cases, and the reduced regression cases are regressed on the feature value retrieval model.

[0096] It should be noted that the above functions or steps that can be implemented by the computer-readable storage medium or computer device can refer to the relevant descriptions on the server side and the client side in the aforementioned method embodiment. To avoid repetition, they will not be described one by one here.

[0097] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0098] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0099] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. A data intelligent update and iteration method, characterized in that: include: Obtain a basic database, extract characteristic values ​​from the basic database and establish a characteristic value retrieval model; Extract the initial eigenvalues ​​in the eigenvalue retrieval model and set the initial eigenvalues ​​as the initial regression cases; Based on the step-size method, the initial regression cases are trained to obtain the simplified regression cases; The reduced regression cases are iterated on the initial regression cases, and the reduced regression cases are regressed on the feature value retrieval model.

2. The data intelligent update and iteration method according to claim 1, characterized in that: Before the steps of obtaining a basic database, extracting characteristic values ​​from the basic database and establishing a characteristic value retrieval model, the method includes: Sort out the demand documents in the insurance policy system, and create a demand classification table based on the unique document number, demand summary, party proposing the demand, party coordinating the demand, and business identification tag fields, and store it in the basic database.

3. The data intelligent update and iteration method according to claim 2, characterized in that: Before the steps of obtaining a basic database, extracting characteristic values ​​from the basic database and establishing a characteristic value retrieval model, the method further includes: Sort out the fields of unique use case number, use case category, use case test point, use case detail information, and use case identification tag in the insurance policy system, set up label classification, establish a use case classification table, and store it in the basic database.

4. The data intelligent update and iteration method according to claim 3, characterized in that: Before the steps of obtaining a basic database, extracting characteristic values ​​from the basic database and establishing a characteristic value retrieval model, the method further includes: Associate the relationships between the requirements in the requirement classification table and the use case types in the use case classification table, establish a characteristic value relationship table, and store the characteristic value relationship table in the basic database.

5. The data intelligent update and iteration method according to claim 4, characterized in that: The step of training the initial regression case based on the step length method to obtain a simplified regression case includes: Determine the eigenvalue retrieval direction and the step factor extracted from the eigenvalue relationship table; Based on the step size method, cyclic training is performed in combination with the eigenvalue retrieval direction, step size factor and initial regression case until the training result reaches the preset threshold to obtain a streamlined regression case.

6. The data intelligent updating and iteration method according to claim 5, characterized in that: After the steps of iterating the simplified regression case into the initial regression case and regressing the simplified regression case into the feature value retrieval model, the method further includes: Extracting reduced feature labels for reduced regression models and extracting initial feature labels for initial regression cases; Determine whether the scope of the initial feature labels completely covers the simplified feature labels; If it is fully covered, the streamlined regression model training is successful; If it is partially covered or not covered at all, the training of the streamlined regression model fails, and the steps are re-executed to train the initial regression cases based on the step size method to obtain the streamlined regression cases.

7. The data intelligent updating and iteration method according to claim 1, characterized in that: After the steps of iterating the simplified regression case into the initial regression case and regressing the simplified regression case into the feature value retrieval model, the method further includes: Extract initial feature labels for initial regression cases; According to the feature label, the feature value retrieves redundant regression cases with feature value labels other than the initial regression case in the model, wherein the redundant regression cases do not include the initial regression case; Eliminate redundant regression cases.

8. A data intelligent update and iteration device, characterized in that: include: An extraction module is used to obtain a basic database, extract characteristic values ​​in the basic database and establish a characteristic value retrieval model; An extraction module, used for extracting initial eigenvalues ​​in the eigenvalue retrieval model and setting the initial eigenvalues ​​as initial regression cases; A training module, used to train the initial regression cases based on the step-size method to obtain the simplified regression cases; The iteration module is used to iterate the simplified regression cases to the initial regression cases, and to regress the simplified regression cases to the feature value retrieval model.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the data intelligent update and iteration method as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the data intelligent update and iteration method as described in any one of claims 1 to 7 are implemented.

Images