A sequence data modeling method and system based on a recurrent neural network

Abstract

The present invention discloses a method and system for sequence data modeling based on a recurrent neural network, belonging to the technical field of sequence data modeling. The method includes: determining each reserve requirement in real time, and sorting out the training data corresponding to each benchmark model according to each reserve requirement and the corresponding benchmark model of the recurrent neural network; setting the function introduction data corresponding to each benchmark model; establishing a model library, and establishing a training and adjustment model according to the model library; analyzing the user demand data uploaded by the user to determine the user application model, and loading and installing the user application model on the user side; identifying each sequence data, analyzing each sequence data through the user application model to obtain an analysis result, and displaying the analysis result to the user; when the user adjusts the analysis result, identifying the optimized result after the user adjustment and integrating it into the application optimization data; optimizing and adjusting the user application model through the application optimization data.

Description

Technical Field

[0001] The present invention belongs to the technical field of sequence data modeling, and specifically relates to a method and system for sequence data modeling based on a recurrent neural network. Background Art

[0002] In today's information society, sequence data is everywhere, such as stock prices, meteorological data, human health monitoring data, etc. The analysis and modeling of these data are of great value to individual users and enterprises. However, the existing sequence data modeling technologies mainly rely on manual establishment by professionals, which not only consumes a large amount of time and energy, but also requires modelers to have profound professional knowledge and practical experience.

[0003] For the vast majority of individual users, although they have strong needs for sequence data analysis and modeling, due to the lack of professional knowledge and skills, it is often difficult for them to establish corresponding models by themselves. At the same time, since individual users are often reluctant to invest a large amount of time and money in a non-essential model, this further limits the popularization, promotion and application of sequence data modeling technologies.

[0004] In view of this situation, there is an urgent need in the market for a technical solution that can lower the threshold of sequence data modeling and improve the modeling efficiency. However, most of the existing solutions still focus on providing advanced modeling tools for professionals and fail to fully consider the needs and characteristics of individual users.

[0005] Therefore, the present invention proposes a sequence data modeling system based on a recurrent neural network, aiming to solve the above problems. Summary of the Invention

[0006] In order to solve the problems existing in the above solutions, the present invention provides a method and system for sequence data modeling based on a recurrent neural network.

[0007] The object of the present invention can be achieved by the following technical solutions:

[0008] A sequence data modeling system based on a recurrent neural network, comprising a platform side and a user side;

[0009] The platform side includes a model module, a requirements analysis module and an upgrade module;

[0010] The model module is used to determine each reserve requirement in real time, and according to each of the reserve requirements and the corresponding benchmark model of the recurrent neural network, the benchmark model is used to analyze the corresponding sequence data; organize the training data corresponding to each of the benchmark models; set the function introduction data corresponding to each of the benchmark models;

[0011] Build a model library, and input various reserve requirements, function introduction data, benchmark models, and training data into the model library for storage;

[0012] Build a training and adjustment model based on the model library, and the training and adjustment model is used to adjust the training data of the corresponding benchmark model according to the sequence material data of the user.

[0013] Further, the method for determining the reserve requirement is as follows:

[0014] Real-time identify various types of sequence data; the platform sets the conversion coefficients of each type of sequence data for different users;

[0015] Screen each type of sequence data to determine each type of sequence to be screened;

[0016] Real-time identify the range of potential users corresponding to each type of sequence to be screened, and count the potential quantities of each potential user within the range of potential users;

[0017] Mark each potential user within the range of potential users as i, where i = 1, 2,..., n, and n is a positive integer; mark the potential quantity of each potential user as Li;

[0018] According to the formula Calculate the screening values of each type of sequence to be screened;

[0019] In the formula: PA is the screening value; λi represents the conversion coefficient corresponding to the corresponding potential user;

[0020] Set the reserve requirement according to the type of sequence to be screened whose screening value is greater than the threshold X1.

[0021] The demand analysis module is used to analyze the user demand data uploaded by the user to determine the user application model, and the user application model is obtained by optimizing and adjusting the corresponding benchmark model according to the user demand data; load and install the user application model on the user side.

[0022] Further, the method for analyzing the user demand data uploaded by the user includes:

[0023] Identify the sequence material data and function requirements in the user demand data, and match the corresponding benchmark model and training data from the model library according to the function requirements;

[0024] Through the training and adjustment model, adjust and analyze the sequence material data and the training data to obtain the adjusted training data, marked as training adjustment data;

[0025] Display the training adjustment data to the user, and the user checks the displayed training adjustment data to obtain the training optimization data;

[0026] Optimize and adjust the benchmark model with the training optimization data to obtain a user application model.

[0027] The upgrade module is used to perform model upgrade adjustment, receive the user's upgrade requirements in real time, and upgrade and adjust the user's application model according to the upgrade requirements.

[0028] The user side includes a requirement module, a data analysis module, and an upgrade requirement module;

[0029] The requirement module is used for the user to set and upload user requirement data, and the user requirement data includes sequence material data and functional requirements.

[0030] Furthermore, the method for the user to set user requirement data includes:

[0031] The platform party presets a requirement tutorial. The user determines the types of sequence data to be analyzed according to the requirement tutorial, counts each sequence data according to the types of sequence data, and the user marks the corresponding processing results for each sequence data and integrates them into sequence material data;

[0032] Set functional requirements; integrate the functional requirements and the sequence material data into user requirement data.

[0033] Furthermore, before the user marks the processing results for each sequence data, conduct a verification analysis on each sequence data and set corresponding rejection marks for the sequence data that fails the verification.

[0034] Furthermore, the method for conducting a verification analysis on each sequence data includes:

[0035] Establish a data verification model, and the expression of the data verification model is:

[0036]

[0037] In the formula: (s, f) is the input data, s is the sequence data; f is the type of sequence data; the output data is the data verification value NH(s, f), and the data verification value is 1 or 0;

[0038] Identify each sequence data and the type of sequence data, analyze each sequence data and the type of sequence data through the data verification model, and obtain the data verification value of each sequence data;

[0039] When the data verification value is 0, it is evaluated that the corresponding sequence data fails the verification;

[0040] When the data verification value is 1, it is evaluated that the corresponding sequence data passes the verification.

[0041] The data analysis module is used to analyze the sequence data, identify each sequence data, analyze each sequence data through the user application model, obtain the corresponding analysis results, and display the analysis results to the user;

[0042] When the user adjusts the analysis results, identify the adjusted analysis results of the user, mark them as optimized results, and integrate the corresponding sequence data, analysis results, and optimized results into application optimization data; optimize and adjust the user application model through the application optimization data;

[0043] When the user does not adjust the analysis results, no corresponding operations are performed.

[0044] Furthermore, the method for optimizing and adjusting the user application model through application optimization data includes:

[0045] Establish an optimization library, send each application optimization data to the optimization library for storage; identify the number of application optimization data stored in the optimization library in real time. When the number of application optimization data reaches the threshold X2, perform homogeneous expansion on each application optimization data stored in the optimization library to obtain equivalent expansion data corresponding to each application optimization data;

[0046] Send optimization expansion information to the user. When the preset time arrives, optimize and adjust the user application model through each application optimization data and equivalent expansion data; delete the application optimization data and equivalent expansion data in the optimization library.

[0047] The upgrade requirement module is used to set corresponding upgrade requirements when the user has upgrade requirements and send the upgrade requirements to the platform side.

[0048] A sequence data modeling method based on a recurrent neural network, the method includes:

[0049] Determine each reserve requirement in real time, and organize the training data corresponding to each benchmark model according to each reserve requirement and the corresponding benchmark model of the recurrent neural network; set the function introduction data corresponding to each benchmark model;

[0050] Establish a model library, input each reserve requirement, function introduction data, benchmark model, and training data into the model library for storage; establish a training adjustment model according to the model library;

[0051] Analyze the user requirement data uploaded by the user to determine the user application model, and load and install the user application model on the user side;

[0052] Identify each sequence data, analyze each sequence data through the user application model, obtain the analysis results, and display the analysis results to the user;

[0053] When the user adjusts the analysis result, identify the adjusted analysis result of the user, mark it as the optimized result, and integrate the corresponding sequence data, analysis result and optimized result into the application optimization data; optimize and adjust the user application model through the application optimization data;

[0054] When the user has an upgrade requirement, set the corresponding upgrade requirement, and the platform party upgrades and adjusts the user's application model according to the upgrade requirement.

[0055] Compared with the prior art, the beneficial effects of the present invention are:

[0056] First of all, the present invention greatly reduces the threshold of sequence data modeling, enabling non-professional users to easily analyze and model sequence data. Traditionally, model establishment requires professionals to spend a lot of time and effort, which is unrealistic for most individual users. However, through the intelligent service of the platform party, the present invention can intelligently generate a model for personal applications according to the sequence data selected by the user, without the user having profound professional knowledge and practical experience, thus greatly expanding the application scope of sequence data modeling technology.

[0057] Secondly, the present invention improves the efficiency of sequence data modeling. Since the platform party has pre-trained models for various sequence data, users can obtain models that meet their needs in a short time without having to build them from scratch. This not only saves the user's time, but also improves the accuracy and reliability of modeling.

[0058] In addition, the present invention also provides a flexible optimization and adjustment scheme. When the user has higher requirements for the accuracy of the model, the platform party can dispatch professional personnel for optimization and adjustment. This way of providing free or low-cost use and experience for each user in the early stage not only reduces the user's threshold and cost, but also facilitates the popularization and promotion of the technology. At the same time, since users will accumulate a large amount of training materials that truly reflect their needs during long-term use, this provides rich data support for the subsequent optimization and adjustment by professional personnel, further improving the accuracy and practicality of the model.

[0059] Finally, the implementation of the present invention also helps to promote the continuous development of sequence data modeling technology. Through the intelligent service of the platform party and the optimization and adjustment by professional personnel, the model library can be continuously accumulated and optimized, improving the adaptability and generalization ability of the model. At the same time, the present invention can also stimulate more users' interest and enthusiasm for sequence data modeling technology, promoting technological innovation and development. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0061] Figure 1 This is the principle block diagram of the present invention. Detailed implementation manners

[0062] The following will clearly and completely describe the technical solutions of the present invention in combination with the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0063] As Figure 1 shown, a sequence data modeling system based on a recurrent neural network includes a platform side and a user side;

[0064] The platform side is used by the platform party and includes a model module, a requirements analysis module, and an upgrade module;

[0065] The model module is used to intelligently analyze various possible sequence data modeling requirements, which are marked as reserve requirements;

[0066] According to each reserve requirement and the recurrent neural network, a corresponding benchmark model is established, and the training data corresponding to each benchmark model is sorted out; the establishment of the benchmark model and the setting of the training data are both established and set by professionals of the platform party; the function introduction data of each benchmark model is set, that is, it is introduced how the benchmark model analyzes the sequence data and what results are obtained;

[0067] A model library is established, and each reserve requirement, function introduction data, benchmark model, and training data are input into the model library for storage;

[0068] An adjustment model for training is established through existing intelligent technologies. The adjustment model for training is used to subsequently and intelligently adjust the training data of the corresponding benchmark model according to the received sequence material data of the user; that is, the adjustment is made according to the rules, requirements, etc. corresponding to the sequence material data of the user. Through the professional capabilities of the platform party, the corresponding adjustment model for training can be established; it can be established manually by the platform party; exemplarily, the adjustment model for training can be established based on neural networks such as CNN networks and DNN networks, and the corresponding training set is established and trained manually. The training set includes input data and output data. The input data is the sequence material data and the training data of the corresponding benchmark model; the output data is the adjusted training data.

[0069] In one embodiment, the determination of the reserve requirement is determined based on the existing method.

[0070] In one embodiment, the method for determining the reserve requirement is as follows:

[0071] All kinds of sequence data are recognized in real time and marked as sequence data types; the platform party estimates the possible conversion rates of each sequence data type for different users and marks them as conversion coefficients, that is, the platform party estimates the conversion rates of each sequence data type for various users;

[0072] Combined with the already established corresponding benchmark model, each sequence type to be screened is determined;

[0073] The range of potential users corresponding to each sequence type to be screened is recognized in real time, that is, the range of people who will apply this type of sequence data; the number of each potential user within the range of potential users is statistically estimated in real time and marked as the potential quantity;

[0074] Each potential user within the range of potential users is marked as i, i = 1, 2,..., n, where n is a positive integer; the potential quantity of each potential user is marked as Li;

[0075] According to the formula Calculate the screening value of each sequence type to be screened;

[0076] In the formula: PA is the screening value; λi represents the conversion coefficient corresponding to the corresponding potential user;

[0077] Set the reserve requirement according to the sequence types to be screened whose screening value is greater than the threshold X1.

[0078] The requirement analysis module is used to analyze the user requirement data uploaded by the user, identify the sequence material data and functional requirements in the user requirement data, and match the corresponding benchmark model and training data from the model library according to the functional requirements; through training adjustment, the model adjusts and analyzes the sequence material data and training data to obtain the adjusted training data, which is marked as training adjustment data, and the training adjustment data is displayed to the user. The user can check the displayed training adjustment data according to the requirements, or can choose not to check, but the accuracy after checking is higher. If not checked, the training is optimized according to the subsequent adjustment data of the user; the checked training adjustment data is marked as training optimization data, and the benchmark model is optimized and adjusted through the training optimization data, and the adjusted benchmark model is marked as the user application model and loaded at the user side.

[0079] The upgrade module is used to upgrade and adjust the user application model of the user, that is, the professional personnel of the platform party perform targeted optimization and upgrade on the user application model of the user to improve its analysis accuracy.

[0080] The user side is used for users to use, and can be in the form of an app, a website, etc.; it includes a requirement module, a data analysis module and an upgrade requirement module;

[0081] The requirement module is used for users to set and upload user requirement data. The user requirement data includes relevant data such as sequence material data and functional requirements; the detailed process is as follows:

[0082] Preset the corresponding user requirement data template, that is, the corresponding tutorial. The user determines the types of sequence data to be analyzed according to the requirement tutorial. The user counts each sequence data according to the type of sequence data, marks the corresponding processing results for each sequence data, and integrates them into sequence material data; set the functional requirements; integrate the functional requirements and sequence material data into user requirement data.

[0083] In one embodiment, because in the actual application process, it often occurs that the user adds sequence data of non - this sequence data type to the sequence material data, which affects the analysis accuracy of the subsequent benchmark model. Therefore, in this embodiment, each sequence data determined by the user is checked, and the process is as follows:

[0084] Establish a data check model. The data check model is used to judge whether the sequence data belongs to the corresponding sequence data type. The expression of the data check model is In the formula: (s, f) is the input data, s is the sequence data; f is the sequence data type; the output data is the data check value NH(s, f), and the data check value is 1 or 0;

[0085] Analyze each sequence data and the sequence data type through the data check model to obtain the data check value of each sequence data;

[0086] Mark and eliminate each sequence of data with a data verification value of 0.

[0087] The data analysis module is used to analyze the corresponding sequence data at the user's end, identify each sequence of data, analyze each sequence of data through the user application model, obtain the corresponding analysis results, and display the analysis results to the user.

[0088] When the user adjusts the analysis results, identify the adjusted analysis results of the user, mark them as optimized results, and integrate the corresponding sequence data, analysis results, and optimized results into application optimization data; optimize and adjust the user application model through the application optimization data.

[0089] When the user does not adjust the analysis results, no corresponding operation is performed.

[0090] In one embodiment, optimizing and adjusting the user application model through the application optimization data is a real-time adjustment, that is, the user application model is adjusted in real time according to the obtained application optimization data.

[0091] In one embodiment, the method of optimizing and adjusting the user application model through the application optimization data is to optimize and adjust the user application model based on the existing technology according to the application optimization data.

[0092] In one embodiment, the method of optimizing and adjusting the user application model through the application optimization data includes:

[0093] Establish an optimization library, send each application optimization data to the optimization library for storage, and the user can view the optimization in real time; identify the quantity of the stored application optimization data in real time. When the quantity of the application optimization data reaches the threshold X2, perform a similar expansion on each of the stored application optimization data in the optimization library to obtain the equivalent expansion data corresponding to each application optimization data, that is, expand and simulate according to the sequence data in the application optimization data to form a sequence data that can be regarded as equivalent but has different data, and generate the corresponding analysis results based on the generated sequence data according to the optimization results; the above expansion steps can be implemented based on the existing technology. For example, the platform party establishes a corresponding equivalent expansion model based on neural networks such as CNN networks or DNN networks, and establishes a corresponding training set for training through manual means. The training set includes input data and output data. The input data is each application optimization data, and the output data is the equivalent expansion data corresponding to each application optimization data; analyze through the equivalent expansion model to obtain the equivalent expansion data corresponding to each application optimization data; send optimization expansion information to the user, indicating that the user application model will be optimized and adjusted after a preset time. During this period, the user can view each application optimization data and equivalent expansion data and can adjust the equivalent expansion data.

[0094] When the preset time is reached, optimize and adjust the user application model with the application optimization data and equivalent expansion data for each application, and the corresponding equivalent expansion data can also be retained for verification; delete the application optimization data and equivalent expansion data in the optimization library.

[0095] The upgrade requirement module is used to set corresponding upgrade requirements when the user has a need to professionally upgrade the user application model, and send the corresponding upgrade requirements to the platform side.

[0096] Because the user application model in the early stage is basically provided for each user free of charge or at a low cost, when the user has a higher-precision analysis requirement, professional personnel on the platform side need to perform optimization and upgrade.

[0097] A sequence data modeling method based on a recurrent neural network, the method includes:

[0098] Determine each reserve requirement in real time, and organize the training data corresponding to each benchmark model according to each reserve requirement and the corresponding benchmark model of the recurrent neural network; set the function introduction data corresponding to each benchmark model;

[0099] Establish a model library, input each reserve requirement, function introduction data, benchmark model and training data into the model library for storage; establish a training adjustment model according to the model library;

[0100] Analyze the user demand data uploaded by the user, determine the user application model, and load and install the user application model on the user side.

[0101] Identify each sequence data, analyze each sequence data through the user application model, obtain the corresponding analysis result, and display the analysis result to the user;

[0102] When the user adjusts the analysis result, identify the adjusted analysis result of the user, mark it as the optimization result, and integrate the corresponding sequence data, analysis result and optimization result into the application optimization data; optimize and adjust the user application model with the application optimization data;

[0103] When the user has an upgrade requirement, set the corresponding upgrade requirement, and the platform side upgrades and adjusts the user's user application model according to the upgrade requirement.

[0104] The above formulas are all calculated by removing the dimension and taking their numerical values. The formula is a formula obtained by collecting a large amount of data for software simulation to be closest to the real situation. The preset parameters and preset thresholds in the formula are set by those skilled in the art according to the actual situation or obtained by a large amount of data simulation.

[0105] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical method of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A sequence data modeling system based on a recurrent neural network, characterized in that, It includes a platform side and a user side; The platform side includes a model module, a requirements analysis module, and an upgrade module; The model module is used to determine each reserve requirement in real time, and according to each reserve requirement and the corresponding benchmark model of the recurrent neural network, the benchmark model is used to analyze the corresponding sequence data; sort out the training data corresponding to each benchmark model; Set the function introduction data corresponding to each benchmark model; Establish a model library, and input each reserve requirement, function introduction data, benchmark model, and training data into the model library for storage; Establish a training adjustment model according to the model library, and the training adjustment model is used to adjust the training data of the corresponding benchmark model according to the user's sequence material data; The requirements analysis module is used to analyze the user requirement data uploaded by the user, determine the user application model, and the user application model is obtained by optimizing and adjusting the corresponding benchmark model according to the user requirement data; load and install the user application model in the user side; The upgrade module is used to perform model upgrade and adjustment, receive the user's upgrade requirement in real time, and upgrade and adjust the user's user application model according to the upgrade requirement; The user side includes a requirements module, a data analysis module, and an upgrade requirement module; The requirements module is used for the user to set and upload user requirement data, and the user requirement data includes sequence material data and function requirements; The data analysis module is used to analyze the sequence data, identify each sequence data, analyze each sequence data through the user application model, obtain the corresponding analysis result, and display the analysis result to the user; When the user adjusts the analysis result, identify the adjusted analysis result by the user, mark it as the optimization result, and integrate the corresponding sequence data, analysis result, and optimization result into the application optimization data; Optimize and adjust the user application model through the application optimization data; When the user does not adjust the analysis result, no corresponding operation is performed; The upgrade requirement module is used to set the corresponding upgrade requirement when the user has an upgrade requirement, and send the upgrade requirement to the platform side; The method for determining the reserve requirement is: Identify each sequence data type in real time; the platform party sets the conversion coefficient of each sequence data type for different users; Screen each sequence data type to determine each sequence type to be screened; Identify the potential user range corresponding to each sequence type to be screened in real time, and count the potential number of each potential user within the potential user range; Mark each potential user within the potential user range as i, i = 1, 2,..., n, where n is a positive integer; mark the potential number of each potential user as Li; According to the formula Calculate the screening values of each sequence type to be screened; In the formula: PA is the screening value; λi represents the conversion coefficient corresponding to the corresponding potential user; Set the reserve requirement according to the sequence type to be screened whose screening value is greater than the threshold X1.

2. The sequence data modeling system based on a recurrent neural network according to claim 1, wherein The method for analyzing the user requirement data uploaded by the user includes: Identify the sequence material data and function requirements in the user requirement data, and match the corresponding benchmark model and training data from the model library according to the function requirements; The training adjustment model is used to adjust and analyze the sequence material data and the training data to obtain the adjusted training data, which is marked as training adjustment data; The training adjustment data is displayed to the user, and the user checks the displayed training adjustment data to obtain the training optimization data; The benchmark model is optimized and adjusted by the training optimization data to obtain the user application model.

3. The sequence data modeling system based on a recurrent neural network according to claim 1, characterized in that The method for the user to set the user demand data includes: The platform party presets a demand tutorial. The user determines the types of sequence data to be analyzed according to the demand tutorial, counts each sequence data according to the types of sequence data, and the user marks the corresponding processing results for each sequence data, which are integrated into sequence material data; Set the functional requirements; integrate the functional requirements and the sequence material data into the user demand data.

4. A sequence data modeling system based on a recurrent neural network according to claim 3, wherein Before the user marks the processing results for each sequence data, check and analyze each sequence data, and set the corresponding rejection marks for the sequence data that fails the check.

5. A sequence data modeling system based on a recurrent neural network according to claim 4, wherein The method for checking and analyzing each sequence data includes: Establish a data check model, and the expression of the data check model is: ； In the formula: (s, f) is the input data, s is the sequence data; f is the type of sequence data; the output data is the data check value NH(s, f), and the data check value is 1 or 0; Identify each sequence data and the type of sequence data, and analyze each sequence data and the type of sequence data through the data check model to obtain the data check value of each sequence data; When the data check value is 0, it is evaluated that the corresponding sequence data fails the check; When the data check value is 1, it is evaluated that the corresponding sequence data passes the check.

6. A sequence data modeling system based on a recurrent neural network according to claim 1, characterized in that The method for optimizing and adjusting the user application model by applying the optimization data includes: Establish an optimization library, and send each application optimization data to the optimization library for storage; real-time identify the number of application optimization data stored in the optimization library. When the number of application optimization data reaches the threshold X2, perform homogeneous expansion on each application optimization data stored in the optimization library to obtain the equivalent expansion data corresponding to each application optimization data; Send optimization expansion information to the user. When the preset time arrives, optimize and adjust the user application model by each application optimization data and the equivalent expansion data; delete the application optimization data and the equivalent expansion data in the optimization library.

7. A method for modeling sequence data based on a recurrent neural network, characterized in that, Applied to a sequence data modeling system based on a recurrent neural network as described in any one of claims 1 to 6, the method includes: Determine each reserve requirement in real time, and according to each reserve requirement and the corresponding benchmark model of the recurrent neural network, sort out the training data corresponding to each benchmark model; set the function introduction data corresponding to each benchmark model; Establish a model library, input each reserve requirement, function introduction data, benchmark model and training data into the model library for storage; establish a training adjustment model according to the model library; Analyze the user demand data uploaded by the user to determine the user application model, and load and install the user application model on the user side; Identify each sequence data, analyze each sequence data through the user application model to obtain the analysis result, and display the analysis result to the user; When the user adjusts the analysis result, identify the adjusted analysis result of the user, mark it as the optimized result, and integrate the corresponding sequence data, analysis result, and optimized result into the application optimization data; optimize and adjust the user application model through the application optimization data; When the user has an upgrade requirement, set the corresponding upgrade requirement, and the platform party upgrades and adjusts the user's application model according to the upgrade requirement.

Images