Refined data processing method based on artificial intelligence

Abstract

The present invention discloses a refined data processing method based on artificial intelligence, which relates to the field of data processing and comprises the following steps: setting a data processing platform, setting management areas according to the distribution of users in the data processing platform, and setting corresponding data management nodes according to the management areas; obtaining historical data information in each management area and data information to be processed of corresponding users; obtaining feature data information of the historical data information by the data management node, constructing a feature data set, and generating a data processing model; performing feature analysis on the obtained data information to be processed, obtaining the corresponding feature data information to be processed, matching the obtained feature data information to be processed with the feature data set, and selecting the corresponding data processing model according to the matching result to perform analysis and processing on the data; the present invention reduces the overfitting of the data processing model in the process of processing data, and improves the accuracy in the process of data processing.

Description

Technical Field

[0001] The present invention relates to the field of data processing, and in particular to a refined data processing method based on artificial intelligence. Background Art

[0002] With the development of modern society, people's demand for data is getting higher and higher, and the speed and accuracy of data processing are also getting higher and higher. Traditional data processing methods mainly include manual processing and automatic processing. However, both methods have some obvious limitations, such as manual processing is slow and prone to errors, and automatic processing cannot meet the processing needs of complex business scenarios. The emergence of artificial intelligence provides a new solution for data processing;

[0003] The data processing method, device, equipment and medium with publication number CN112269805B discloses a data processing method, a data processing device, an electronic device and a computer-readable storage medium. Among them, the data processing method includes: obtaining digitized customer data corresponding to the original customer data; labeling the digitized customer data based on preset label rules to obtain labeled customer data, wherein the preset label rules include: preset static label rules, real-time rule label rules and dynamic fuzzy label rules; profiling the labeled customer data to determine the target customer data for application in refined marketing. By having preset label rules that have preset static label rules, real-time rule label rules and dynamic fuzzy label rules at the same time, it is possible to label multiple types of data in the business system, thereby accurately obtaining potential customer data to facilitate more accurate and refined marketing strategies. In addition, by processing raw data through multi-channel aggregation, the data source is greatly expanded.

[0004] At present, artificial intelligence technology can process data quickly and accurately through the analysis and mining of large amounts of data, thereby improving the efficiency and accuracy of data processing; however, in the process of setting the corresponding data processing model through artificial intelligence, overfitting and insufficient generalization ability are prone to occur, thereby reducing the accuracy of the data processing process; therefore, how to avoid overfitting and insufficient generalization ability of artificial intelligence in the data processing process, resulting in insufficient accuracy in the data processing process, is a problem we need to solve. To this end, a refined data processing method based on artificial intelligence is now provided. Summary of the invention

[0005] In order to solve the above technical problems, the purpose of the present invention is to provide a refined data processing method based on artificial intelligence.

[0006] The purpose of the present invention can be achieved by the following technical solution: A refined data processing method based on artificial intelligence comprises the following steps:

[0007] Step S1: Setting up a data processing platform, setting up a management area according to the distribution of users in the data processing platform, and setting up corresponding data management nodes according to the management area;

[0008] Step S2: Obtain historical data information in the management area corresponding to each data management node and data information to be processed of the corresponding user, and mark and store the obtained historical data information and data information to be processed;

[0009] Step S3: The data management node analyzes and processes the obtained historical data information, obtains feature data information of the historical data information, constructs a feature data set according to the feature data information, analyzes and trains the feature data set based on a deep learning algorithm, generates a data processing model, and generates a corresponding data processing node;

[0010] Step S4: Perform feature analysis on the obtained data information to be processed, obtain the corresponding feature data information to be processed, match the obtained feature data information to be processed with the feature data set in the corresponding data processing node, and if the match is successful, input it into the corresponding data processing model for analysis and processing;

[0011] Step S5: Generate federated retrieval information for the unmatched feature data information to be processed, obtain matching results between feature data sets of other data processing nodes in the data processing platform and the feature data information to be processed based on the federated retrieval information, and analyze and process the data information to be processed according to the matching results.

[0012] Furthermore, the process of setting up the data processing platform and the corresponding data management node therein includes:

[0013] A data processing platform is set up, in which a user entry window is set up, the user entry window is used to enter user verification information of using the data processing platform, the user verification information is verified, and it is determined whether to generate a user account of the corresponding user based on the verification result, and a management area is set according to the distribution of user accounts, and corresponding data management nodes are set according to the set management area. The data management node is used to analyze and manage data information submitted by user accounts in the corresponding management area.

[0014] Furthermore, the process of obtaining the historical data information in the management area corresponding to each data management node and the to-be-processed data information of the corresponding user includes:

[0015] The data management node is provided with a data entry window and a data collection window; the data entry window is used to enter relevant historical data information and data information to be processed for the user account in the corresponding management area; the data collection window is provided with a corresponding API interface according to the user account requirements, and the API interface is used to collect historical data information and data information to be processed corresponding to the third-party service or functional program according to the user account requirements; the historical data information includes the historical data information to be processed and the data processing results in the corresponding management area in the data processing platform.

[0016] Furthermore, the data management node analyzes and processes the obtained historical data information to obtain feature data information of the historical data information, and the process of constructing a feature data set according to the feature data information includes:

[0017] The data management node obtains historical data information obtained in the corresponding management area, performs feature extraction on the historical data information, obtains feature data information corresponding to the historical data information, classifies the historical data information according to the obtained feature data information, obtains historical data information with the same feature data information, generates a corresponding feature data set according to the same historical data information, and obtains the amount of data elements in the feature data set;

[0018] A similar element threshold is preset, and the amount of data elements in the corresponding feature data set is compared and analyzed with the similar element threshold. When the amount of data elements is greater than or equal to the similar element threshold, the feature data set corresponding to the marking result is marked as a big data feature data set; when the amount of data elements is less than the similar element threshold, a circular search link is generated according to the feature data information of the feature data set, and the circular search link is used to obtain historical data information in other data management nodes that corresponds to the same feature data information as the feature data set;

[0019] The successfully retrieved historical data information is marked, and it is determined whether the marked historical data information is in the big data feature data set in other data management nodes. If not, it is marked for backup. A backup is performed based on the retrieved historical data information and a feature data set is generated. The feature data set is marked as a discrete feature data set. If it is in the big data feature data set, the data management node is marked according to the feature data information corresponding to the historical data information.

[0020] Furthermore, the process of analyzing and training the feature data set, generating a data processing model, and generating corresponding data processing nodes includes:

[0021] Based on the deep learning algorithm, the big data feature data set in the data management node is analyzed and trained until the loss function tends to be stable, the model parameters of the data processing model are saved until it meets the preset requirements, the data processing model of the corresponding big data feature data set is output, and the data processing node is set according to the feature data information therein, and it is marked as a regional data processing node; other unlabeled feature data sets in the data management node are associated according to the node label;

[0022] The data management platform is provided with a cross-regional storage space, which is used to store discrete feature data sets, obtain the number of elements of the discrete feature data sets, compare and analyze the obtained number of elements with the similar element threshold, and when the number of elements is greater than or equal to the similar element threshold, construct a data processing model of the discrete feature data sets based on a deep learning algorithm, set data processing nodes according to the feature data information therein, and mark them as cross-regional data processing nodes.

[0023] Furthermore, the process of performing feature analysis on the obtained data information to be processed, obtaining the corresponding feature data information to be processed, and matching the obtained feature data information to be processed with the feature data sets in each data processing node includes:

[0024] The data information to be processed obtained in the data management node is obtained, and feature analysis is performed on the obtained data information to be processed, and feature data information to be processed corresponding to the data information to be processed is obtained, and the obtained feature data information to be processed is matched with feature data information corresponding to feature data sets of each regional data processing node in the data management node; if there is only one successful matching result, the feature data information to be processed is sent to the successfully matched data processing node; if there are two or more successful matching results, the feature data information to be processed is marked as abnormal, and data abnormality information is generated and sent to the data processing platform; if there is no successful matching result, a cross-regional processing application is generated for the feature data information to be processed, and it is sent to the cross-regional storage space, and the feature data information to be processed in the cross-regional processing application is matched with the feature data information corresponding to the feature data set of the cross-regional data processing node, and the corresponding data processing model is selected according to the successful matching result to analyze and process the data information to be processed.

[0025] Furthermore, the process of generating federated search information for unmatched feature data information to be processed includes:

[0026] The data information to be processed that has not been successfully matched by the regional data processing nodes and the cross-regional data processing nodes and its corresponding feature data information to be processed are obtained in the data management node, and federal retrieval information is generated based on the information, and the federal retrieval information is used to traverse the regional data processing nodes in other data management nodes according to the corresponding feature data information to be processed, and the data management nodes associated with the feature data sets are traversed preferentially, and the federal retrieval information is sent to the other data management nodes in sequence until there is a successfully matched regional data processing node or the traversal of other data management nodes in the data processing platform is completed.

[0027] Further, the matching results of the federated search information in other data management nodes are obtained. If the matching is successful, the corresponding regional data processing node selects the corresponding data processing model to analyze and process the data information to be processed in the federated search information; if there is no regional data processing node with a successful matching, the data information to be processed is marked as random noise data;

[0028] A feature data set is generated according to the feature data information to be processed of the random noise data and stored.

[0029] Compared with the prior art, the present invention has the following beneficial effects:

[0030] 1. By setting management areas according to the distribution of users in the data processing platform, setting corresponding data management nodes according to the management areas, and analyzing and processing the data to be processed in each management area according to the different data management nodes, the efficiency of the data processing process is improved to a certain extent, and by retrieving the historical data information of the same feature data information in each data management node, a discrete feature data set is constructed according to the retrieval results. By setting the discrete feature data set, a small amount of historical data information distributed in different management areas is used to construct a data processing model, thereby improving the utilization rate of system resources;

[0031] 2. Set corresponding data processing models according to different feature data sets, extract features from the data to be processed, obtain the feature data to be processed and perform matching analysis with each feature data set, and select the corresponding data processing model according to the matching analysis results to analyze and process it, thereby realizing the fine division of the data processing models corresponding to the data to be processed, thereby avoiding overfitting and insufficient generalization ability of a data processing model;

[0032] 3. By setting the federated retrieval information to match the feature data sets in other data management nodes, the corresponding data processing model is selected according to the matching results to analyze and process the obtained data information to be processed. This not only improves the security of data information to a certain extent, but also can effectively utilize the processing resources of other data management nodes. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 This is a schematic diagram of the refined data processing method based on artificial intelligence in an embodiment of the present application. DETAILED DESCRIPTION

[0034] like Figure 1 As shown, the refined data processing method based on artificial intelligence includes the following steps:

[0035] Step S1: Setting up a data processing platform, setting up a management area according to the distribution of users in the data processing platform, and setting up corresponding data management nodes according to the management area;

[0036] Step S2: Obtain historical data information in the management area corresponding to each data management node and data information to be processed of the corresponding user, and mark and store the obtained historical data information and data information to be processed;

[0037] Step S3: The data management node analyzes and processes the obtained historical data information, obtains feature data information of the historical data information, constructs a feature data set according to the feature data information, analyzes and trains the feature data set based on a deep learning algorithm, generates a data processing model, and generates a corresponding data processing node;

[0038] Step S4: Perform feature analysis on the obtained data information to be processed, obtain the corresponding feature data information to be processed, match the obtained feature data information to be processed with the feature data set in the corresponding data processing node, and if the match is successful, input it into the corresponding data processing model for analysis and processing;

[0039] Step S5: Generate federated retrieval information for the unmatched feature data information to be processed, obtain matching results between feature data sets of other data processing nodes in the data processing platform and the feature data information to be processed based on the federated retrieval information, and analyze and process the data information to be processed according to the matching results.

[0040] It should be further explained that, in the specific implementation process, the data processing platform is set, the management area is set according to the distribution of users in the data processing platform, the corresponding data management node is set according to the management area, and the data management node analyzes and processes the data of the corresponding users in the management area, including:

[0041] Setting up a data processing platform, which is used to perform refined analysis and processing on the data information provided by users on the platform;

[0042] The data processing platform is provided with a user entry window, and the user entry window is used to enter the user verification information of using the data processing platform, and the user verification information includes the user's basic identity information, the user's IP information and the user's qualification certification information, and the obtained user verification information is verified and processed, and it is determined whether to generate a user account of the corresponding user according to the verification processing result. If the verification processing result is passed, the user is granted the right to enter the data processing platform, and the user account and the corresponding login password are set for the user; if the verification processing result is not passed, the user is not granted the right to enter the data processing platform, and the user verification information is rejected;

[0043] It should be further explained that, in the specific implementation process, the user basic identity information corresponding to the user verification information includes the user's name, user ID number and the mobile phone number of the user who completed the real-name authentication, and the user's IP information includes IP information, geographic location information and network type information;

[0044] The data processing platform divides the area according to the IP information of the user of the user account, sets the corresponding management area according to the IP information of the user, marks the obtained management area, and stores the management area according to the marking result;

[0045] The data processing platform sets corresponding data management nodes according to the marking results of each management area, and the data management nodes are used to analyze and manage the data information submitted by the user accounts in the management area of ​​the corresponding marking results.

[0046] It should be further explained that, in the specific implementation process, the process of obtaining the historical data information in the management area corresponding to each data management node and the to-be-processed data information of the corresponding user, and marking and storing the obtained historical data information and to-be-processed data information includes:

[0047] The data management node is provided with a data entry window and a data collection window, and the data entry window and the data collection window are used to obtain data information in the corresponding management area, and the specific implementation process includes:

[0048] The data entry window is used to obtain historical data information and pending data information related to user accounts in the corresponding management area, and mark the obtained historical data information and pending data information according to the entered user accounts;

[0049] The data collection window is provided with a corresponding API interface according to the user account, and the API interface is used to integrate third-party services or functional programs into the application program in the data processing platform according to the needs of the corresponding user account, realize real-time monitoring of the third-party services or functional programs, obtain corresponding historical data information and data information to be processed, and mark the obtained data information;

[0050] It should be further explained that, in the specific implementation process, the marking results of the historical data information and the data information to be processed obtained in the data entry window and the data collection window include the corresponding user account and collection time. The collected data information is deduplicated according to the marking results, and the consistent data information is eliminated. After the elimination is completed, the remaining data information is temporarily stored.

[0051] It should be further explained that, in the specific implementation process, the data management node analyzes and processes the obtained historical data information, obtains the characteristic data information of the historical data information, constructs a characteristic data set according to the characteristic data information, analyzes and trains the characteristic data set based on the deep learning algorithm, generates a data processing model, and generates a corresponding data processing node. The process includes:

[0052] Analyze and process the historical data information in the data management node, verify the data integrity corresponding to the obtained historical data information, and determine whether the corresponding historical data information meets the verification result; if it meets the verification result, analyze and process the historical data information; if it does not meet the verification result, do not analyze and process the historical data information;

[0053] Acquire historical data information that meets the verification result, perform feature extraction on the historical data information, and acquire feature data information corresponding to the historical data information, wherein the feature data information includes feature identification classification, feature text classification, and feature value classification of the historical data information type; classify and process the corresponding historical data information in the data management node according to the feature data information, acquire historical data information with the same feature data information, and generate a corresponding feature data set according to the same historical data information, wherein the feature data set includes data information to be processed and analysis results corresponding to multiple historical data information with the same feature data information at historical moments, and mark the feature data set according to the feature data information of the historical data information therein, and acquire the amount of data elements in the feature data set corresponding to each marking result;

[0054] A similar element threshold is preset, and the amount of data elements in the feature data set corresponding to each marking result is compared and analyzed with the similar element threshold. When the amount of data elements is greater than or equal to the similar element threshold, the feature data set corresponding to the marking result is marked as a big data feature data set; when the amount of data elements is less than the similar element threshold, a circular retrieval link is generated based on the feature data set, and the circular retrieval link obtains the historical data information corresponding to the feature data set in other data management nodes; the successfully retrieved historical data information is marked accordingly, and it is determined whether the marked historical data information is in the big data feature data set in other data management nodes. If not, it is marked for backup, and a feature data set is generated based on the retrieved historical data information. The feature data set is marked as a discrete feature data set. If it is in the big data feature data set, the data management node is marked according to the feature data information corresponding to the historical data information.

[0055] Setting a regional storage space according to each data management node, wherein the regional storage space is used to store a large data feature data set obtained in a corresponding management area;

[0056] A cross-region storage space is set in the data management platform, and the cross-region storage space is used to store discrete feature data sets obtained by corresponding circular retrieval links.

[0057] Based on the deep learning algorithm, the big data feature data set in each data management node is analyzed and trained, and a data processing model for analyzing and processing historical data information is constructed. The big data feature data set is divided into a training set and a validation set. The training set is analyzed and trained based on the deep learning algorithm until the loss function tends to be stable, and the model parameters of the data processing model are saved. The obtained validation set is input into the data processing model, and the validation result is output. If it meets the preset requirements, the data processing model is saved. If it does not meet the preset requirements, the data storage model is retrained until it meets the preset requirements; the corresponding data processing node is set according to the obtained data processing model, and the data processing node is marked. The marking result includes the feature data set corresponding to the data processing node and the data management node to which it belongs, and the data processing node is marked as a regional data processing node;

[0058] Obtain a discrete feature data set in a cross-regional storage space, obtain the number of elements in the discrete feature data set, compare and analyze the obtained number of elements with a similar element threshold, and when the number of elements is greater than or equal to the similar element threshold, construct a data processing model for the discrete feature data set based on a deep learning algorithm, perform marking processing according to the data processing model, and mark the data processing node as a cross-regional data processing node;

[0059] When the number of elements is less than the similar element threshold, the discrete feature data set is stored.

[0060] It should be further explained that, in the specific implementation process, the process of performing feature analysis on the obtained data information to be processed, obtaining the corresponding feature data information to be processed, matching the obtained feature data information to be processed with the feature data set in the corresponding data processing node, and if the match is successful, inputting it into the corresponding data processing model for analysis and processing includes:

[0061] Acquire the data information to be processed obtained in the data management node, perform verification processing on the obtained data information to be processed, and determine the accuracy of the data information to be processed;

[0062] Performing feature analysis on the data information to be processed that meets the verification processing, obtaining the feature data information to be processed of the data information to be processed, matching the obtained feature data information to be processed with the feature data sets of each data processing node in the data management node, obtaining the matching result of the feature data information to be processed, and if there is only one matching result, sending the feature data information to be processed to the regional data processing node with successful matching, the regional data processing node obtains the corresponding feature data information to be processed, inputs it into the data processing model, and outputs the processing result of the feature data information to be processed;

[0063] If there are two or more successful matching results, the feature data information to be processed is marked as abnormal, and data abnormality information is generated and sent to the data processing platform;

[0064] If there is no successful match result, a cross-regional processing application is generated for the feature data information to be processed, the obtained cross-regional processing application is sent to the cross-regional storage space, the obtained cross-regional processing application is matched with the cross-regional data processing nodes corresponding to each discrete feature data set in the cross-regional storage space, and the corresponding cross-regional data processing node is obtained according to the matching result; if the match is successful, the corresponding cross-regional data processing node is obtained, and the corresponding cross-regional data processing node analyzes and processes the data information to be processed corresponding to the feature data information to be processed.

[0065] It should be further explained that, in a specific implementation process, the process of generating federated retrieval information for the unmatched feature data information to be processed, obtaining matching results of feature data sets of other data processing nodes in the data processing platform with the feature data information to be processed based on the federated retrieval information, and analyzing and processing the data information to be processed according to the matching results includes:

[0066] Obtain unprocessed feature data information that has not been matched in the cross-regional storage space, generate federal search information based on the obtained feature data information to be processed, the federal search information includes the corresponding feature data information to be processed, the data information to be processed, and the federal search application information, the federal search application information is the search anti-counterfeiting information that is associated with other data management nodes in the data processing platform, generate key information based on the feature data information to be processed and the search anti-counterfeiting information, encrypt the federal search application information based on the obtained key information, send the encrypted federal search application information to other data management nodes in sequence, and give priority to traversing the data management nodes that are associated with the feature data set;

[0067] Other data management nodes match each data processing node in turn according to the obtained federated search application information. The data processing node generates key information according to the characteristic data information in its characteristic data set, and decrypts the federated search application information according to the generated key information. If the corresponding data processing node fails to decrypt successfully, the federated search application information will not be analyzed and processed.

[0068] If the decryption is successful, the corresponding data processing model in the data processing node analyzes and processes the data information to be processed in the federal search application information;

[0069] If the data processing nodes in all data management nodes in the data processing platform fail to decrypt successfully, the federated search application information is returned to the corresponding data management node, a processing failure message is generated, and the data information to be processed is marked as random noise data;

[0070] It should be further explained that, in the specific implementation process, the data information to be processed marked as random noise data is stored in the cross-region storage space, and the feature data information to be processed of the random noise data is used to generate a corresponding feature data set, until the number of elements in the feature data set is greater than or equal to the similar element threshold, then a corresponding data processing model is constructed for the feature data set based on the deep learning algorithm;

[0071] The feature data information to be processed, which is marked as random noise data, is fed back and analyzed and processed by the staff to obtain the analysis and processing results.

[0072] 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 skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A refined data processing method based on artificial intelligence, characterized in that: The following steps are involved: Step S1: Setting up a data processing platform, setting up a management area according to the distribution of users in the data processing platform, and setting up corresponding data management nodes according to the management area; Step S2: Obtain historical data information in the management area corresponding to each data management node and data information to be processed of the corresponding user, and mark and store the obtained historical data information and data information to be processed; Step S3: The data management node analyzes and processes the obtained historical data information, obtains feature data information of the historical data information, constructs a feature data set according to the feature data information, analyzes and trains the feature data set based on a deep learning algorithm, generates a data processing model, and generates a corresponding data processing node; Step S4: Perform feature analysis on the obtained data information to be processed, obtain the corresponding feature data information to be processed, match the obtained feature data information to be processed with the feature data set in the corresponding data processing node, and if the match is successful, input it into the corresponding data processing model for analysis and processing; Step S5: generating federated retrieval information for the unmatched feature data information to be processed, obtaining matching results between the feature data sets of other data processing nodes in the data processing platform and the feature data information to be processed based on the federated retrieval information, and analyzing and processing the data information to be processed according to the matching results; The process of analyzing and training the feature data set, generating a data processing model, and generating corresponding data processing nodes includes: Based on the deep learning algorithm, the big data feature data set in the data management node is analyzed and trained until the loss function tends to be stable, the model parameters of the data processing model are saved until it meets the preset requirements, the data processing model of the corresponding big data feature data set is output, and the data processing node is set according to the feature data information therein, and it is marked as a regional data processing node; the corresponding feature data set is associated with the regional data processing node in other data management nodes according to the node mark; The data processing platform is provided with a cross-regional storage space, and the cross-regional storage space is used to store discrete feature data sets, obtain the number of elements of the discrete feature data sets, compare and analyze the obtained number of elements with the similar element threshold, and when the number of elements is greater than or equal to the similar element threshold, construct a data processing model of the discrete feature data sets based on a deep learning algorithm, set data processing nodes according to the feature data information therein, and mark them as cross-regional data processing nodes.

2. The method for refined data processing based on artificial intelligence according to claim 1, characterized in that: The process of setting up the data processing platform and the corresponding data management node therein includes: A data processing platform is set up, in which a user entry window is set up, the user entry window is used to enter user verification information of using the data processing platform, the user verification information is verified, and it is determined whether to generate a user account of the corresponding user based on the verification result, and a management area is set according to the distribution of user accounts, and corresponding data management nodes are set according to the set management area. The data management node is used to analyze and manage data information submitted by user accounts in the corresponding management area.

3. The method for refined data processing based on artificial intelligence according to claim 2, characterized in that: The process of obtaining the historical data information in the management area corresponding to each data management node and the to-be-processed data information of the corresponding user includes: The data management node is provided with a data entry window and a data collection window; the data entry window is used to enter relevant historical data information and data information to be processed for the user account in the corresponding management area; the data collection window is provided with a corresponding API interface according to the user account requirements, and the API interface is used to collect historical data information and data information to be processed corresponding to the third-party service or functional program according to the user account requirements; the historical data information includes the historical data information to be processed and the data processing results in the corresponding management area in the data processing platform.

4. The method for refined data processing based on artificial intelligence according to claim 3 is characterized in that: The data management node analyzes and processes the acquired historical data information to obtain feature data information of the historical data information, and the process of constructing a feature data set according to the feature data information includes: The data management node obtains historical data information obtained in the corresponding management area, performs feature extraction on the historical data information, obtains feature data information corresponding to the historical data information, classifies the historical data information according to the obtained feature data information, obtains historical data information with the same feature data information, generates a corresponding feature data set according to the same historical data information, and obtains the amount of data elements in the feature data set; Preset a similar element threshold, compare and analyze the amount of data elements in the corresponding feature data set with the similar element threshold, and when the amount of data elements is greater than or equal to the similar element threshold, mark the corresponding feature data set as a big data feature data set; when the amount of data elements is less than the similar element threshold, generate a circular search link based on the feature data information of the feature data set, and the circular search link is used to obtain historical data information corresponding to the same feature data information of the feature data set in other data management nodes; The successfully retrieved historical data information is marked, and it is determined whether the marked historical data information is in the big data feature data set in other data management nodes. If not, it is marked for backup. A backup is performed based on the retrieved historical data information and a feature data set is generated. The feature data set is marked as a discrete feature data set. If it is in the big data feature data set, the data management node is marked according to the feature data information corresponding to the historical data information.

5. The method for refined data processing based on artificial intelligence according to claim 4 is characterized in that: The process of performing feature analysis on the obtained data information to be processed, obtaining the corresponding feature data information to be processed, and matching the obtained feature data information to be processed with the feature data sets in each data processing node includes: The data information to be processed obtained in the data management node is obtained, and feature analysis is performed on the obtained data information to be processed, and feature data information to be processed corresponding to the data information to be processed is obtained, and the obtained feature data information to be processed is matched with feature data information corresponding to feature data sets of each regional data processing node in the data management node; if there is only one successful matching result, the feature data information to be processed is sent to the successfully matched data processing node; if there are two or more successful matching results, the feature data information to be processed is marked as abnormal, and data abnormality information is generated and sent to the data processing platform; if there is no successful matching result, a cross-regional processing application is generated for the feature data information to be processed, and it is sent to the cross-regional storage space, and the feature data information to be processed in the cross-regional processing application is matched with the feature data information corresponding to the feature data set of the cross-regional data processing node, and the corresponding data processing model is selected according to the successful matching result to analyze and process the data information to be processed.

6. The method for refined data processing based on artificial intelligence according to claim 5, characterized in that: The process of generating federated retrieval information for unmatched feature data information to be processed includes: The data information to be processed that has not been successfully matched by the regional data processing nodes and the cross-regional data processing nodes and its corresponding feature data information to be processed are obtained in the data management node, and federal retrieval information is generated based on the information, and the federal retrieval information is used to traverse the regional data processing nodes in other data management nodes according to the corresponding feature data information to be processed, and the data management nodes associated with the feature data sets are traversed preferentially, and the federal retrieval information is sent to the other data management nodes in sequence until there is a successfully matched regional data processing node or the traversal of other data management nodes in the data processing platform is completed.

7. The method for refined data processing based on artificial intelligence according to claim 6, characterized in that: Obtaining matching results of the federated search information in other data management nodes. If the match is successful, the corresponding regional data processing node selects a corresponding data processing model to analyze and process the data information to be processed in the federated search information; If there is no successfully matched regional data processing node, the data information to be processed is marked as random noise data; A feature data set is generated according to the feature data information to be processed of the random noise data and stored.

Images