Data classification method, device, computer equipment and storage medium

Abstract

The present application relates to a data classification method, apparatus, computer device, storage medium, and computer program product. The method comprises: determining, based on the attribute information and field data of a target field in a business system, more than one field characteristic factor corresponding to the target field; combining the field characteristic factors according to the characteristic combination conditions corresponding to at least two classification methods to obtain field characteristics corresponding to each classification method; classifying the target field according to the field characteristics corresponding to each classification method, obtaining preliminary classification results corresponding to each classification method; fusing the preliminary classification results to obtain a field classification result for the target field; and obtaining the data level of the data corresponding to the target field based on the field classification result mapping. This method can improve the processing efficiency of data classification.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a data classification method, apparatus, computer equipment, storage medium, and computer program product. Background Art

[0002] With the development of computer technology, people generate a large amount of data in the process of work, life, and entertainment, such as personal information such as name, age, ID number, mobile phone number, social chat data, bank transaction data, etc. To ensure that various types of data are effectively protected and legally used, and to ensure data security, various types of data can be classified into security levels and protected according to the corresponding security level, such as setting different confidentiality levels and encrypting them using different encryption algorithms.

[0003] Currently, when classifying data to determine its level, manual classification is often performed based on historical forms, or rules are used to identify and classify data with obvious characteristics. The data classification processing covers limited scenarios and requires repeated confirmation of the classification results, resulting in low data classification processing efficiency. Summary of the Invention

[0004] Based on this, it is necessary to provide a data classification method, device, computer equipment, storage medium and computer program product that can improve the efficiency of data classification processing in order to address the above technical problems.

[0005] A data classification method, comprising:

[0006] Determine more than one field characteristic factor corresponding to the target field based on attribute information and field data of the target field in the business system;

[0007] According to the feature combination conditions corresponding to at least two classification methods, the field feature factors are combined to obtain the field features corresponding to each classification method;

[0008] According to each classification method, classify the target field according to the field characteristics corresponding to the corresponding classification method, and obtain the preliminary classification results corresponding to each classification method;

[0009] Based on the preliminary classification results, the field classification results of the target field are obtained;

[0010] The data level of the data corresponding to the target field is obtained based on the field classification result mapping.

[0011] A data classification device, comprising:

[0012] A characteristic factor determination module, configured to determine more than one field characteristic factor corresponding to a target field based on attribute information and field data of the target field in the business system;

[0013] A feature factor combination module is used to combine field feature factors according to feature combination conditions corresponding to at least two classification methods to obtain field features corresponding to each classification method;

[0014] The field preliminary classification module is used to classify the target field according to each classification method and the field characteristics corresponding to the corresponding classification method, and obtain the preliminary classification results corresponding to each classification method;

[0015] The classification result fusion module is used to fuse the preliminary classification results to obtain the field classification results of the target field;

[0016] The data level determination module is used to obtain the data level of the data corresponding to the target field based on the field classification result mapping.

[0017] A computer device includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0018] Determine more than one field characteristic factor corresponding to the target field based on attribute information and field data of the target field in the business system;

[0019] According to the feature combination conditions corresponding to at least two classification methods, the field feature factors are combined to obtain the field features corresponding to each classification method;

[0020] According to each classification method, classify the target field according to the field characteristics corresponding to the corresponding classification method, and obtain the preliminary classification results corresponding to each classification method;

[0021] Based on the preliminary classification results, the field classification results of the target field are obtained;

[0022] The data level of the data corresponding to the target field is obtained based on the field classification result mapping.

[0023] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the following steps:

[0024] Determine more than one field characteristic factor corresponding to the target field based on attribute information and field data of the target field in the business system;

[0025] According to the feature combination conditions corresponding to at least two classification methods, the field feature factors are combined to obtain the field features corresponding to each classification method;

[0026] According to each classification method, classify the target field according to the field characteristics corresponding to the corresponding classification method, and obtain the preliminary classification results corresponding to each classification method;

[0027] Based on the preliminary classification results, the field classification results of the target field are obtained;

[0028] The data level of the data corresponding to the target field is obtained based on the field classification result mapping.

[0029] A computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the computer program implements the following steps:

[0030] Determine more than one field characteristic factor corresponding to the target field based on attribute information and field data of the target field in the business system;

[0031] According to the feature combination conditions corresponding to at least two classification methods, the field feature factors are combined to obtain the field features corresponding to each classification method;

[0032] According to each classification method, classify the target field according to the field characteristics corresponding to the corresponding classification method, and obtain the preliminary classification results corresponding to each classification method;

[0033] Based on the preliminary classification results, the field classification results of the target field are obtained;

[0034] The data level of the data corresponding to the target field is obtained based on the field classification result mapping.

[0035] The above-mentioned data classification method, apparatus, computer equipment, storage medium and computer program product determine more than one field characteristic factor based on the attribute information and field data of the target field in the business system, combine the field characteristic factors according to the characteristic combination conditions corresponding to at least two classification methods, and classify the target field according to the field characteristics corresponding to each classification method according to each classification method, fuse the preliminary classification results corresponding to each classification method to obtain the field classification result of the target field, and obtain the data level of the data corresponding to the target field based on the field classification result mapping. In the data classification process, the target field is classified by fusing the preliminary classification results of the target field classified by multiple classification methods, so as to obtain the data level of the corresponding data according to the field classification result mapping of the target field. This can cover the applicable scenarios of various classification methods, avoid repeated confirmation of the classification results, and improve the processing efficiency of data classification. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1A diagram of an application environment of a data classification method in one embodiment;

[0037] Figure 2 1 is a flow chart of a data classification method according to an embodiment;

[0038] Figure 3 A schematic diagram of a process for correcting a field classification result of a target field in one embodiment;

[0039] Figure 4 is a flow chart of a data classification method according to another embodiment;

[0040] Figure 5 A schematic diagram of an interface for creating a field in one embodiment;

[0041] Figure 6 A schematic diagram of an interface showing recommended field classifications in one embodiment;

[0042] Figure 7 Schematic diagram of a process for classifying a target field in one embodiment;

[0043] Figure 8 A schematic diagram of field kinship relationships in one embodiment;

[0044] Figure 9 A schematic diagram of an interface for hierarchical processing of a data table in one embodiment;

[0045] Figure 10 is a structural block diagram of a data classification device in one embodiment;

[0046] Figure 11 FIG. 1 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0048] The data classification method provided in this application can be applied to Figure 1In the application environment shown. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store data that the server 104 needs to process. The data storage system can be integrated on the server 104, or it can be placed on the cloud or other network servers. The terminal 102 can log in to the client of the business system, and generate data that needs to be graded in the business system through business processing in the client, such as various business data tables. The terminal sends the business data table to the server 104, so that the server 104 extracts the target field in the business system from the business data table, determines more than one field feature factor based on the attribute information and field data of the target field, combines the field feature factors according to the feature combination conditions corresponding to at least two classification methods, and classifies the target field according to the field features corresponding to each classification method according to each classification method. The preliminary classification results corresponding to each classification method are fused to obtain the field classification result of the target field, and the data level of the data corresponding to the target field is obtained based on the field classification result mapping, thereby realizing the hierarchical processing of each data in the business data table. In addition, when the server 104 has obtained various business data and stored the business data in the data storage system, the server 104 can also directly determine the data that needs to be hierarchically processed from the data storage system, and implement hierarchical processing of the data based on the target field according to the corresponding target field of the data, thereby eliminating the need for direct participation of the terminal 102.

[0049] Terminal 102 may include, but is not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices may include smart speakers, smart TVs, smart air conditioners, and smart car devices. Portable wearable devices may include smart watches, smart bracelets, and head-mounted devices. Server 104 may be implemented as a standalone server or a server cluster consisting of multiple servers.

[0050] In one embodiment, Figure 2 As shown, a data classification method is provided, which is applied to Figure 1 The following steps are used as an example to illustrate the server in the example:

[0051] Step 202 : Based on the attribute information and field data of the target field in the business system, determine more than one field characteristic factor corresponding to the target field.

[0052] Among them, the business system refers to a system that provides business services, such as various systems like financial business systems, online shopping business systems, video business systems, social business systems, etc. During the process of the business system providing corresponding business services, various business data will be generated, such as transaction flow data, social relationship data, video browsing record data, and so on. The business system needs to manage various business systems. For example, it can manage the data corresponding to each business user in the form of business data tables, or manage the data corresponding to various business services. A field refers to the common attribute of the same type of data in the business system. For example, for the name data of each user in the business system, the data can be the specific surnames and names of each user, and the field to which this type of data belongs is the "name" field, that is, the "name" field corresponds to the name data of each user. The target field is the field to which the data to be classified belongs. For example, for the data in the Nth row of a business data table, the field to which the data in the Nth row of the business data table belongs can be determined, and the field to which the data in the Nth row belongs is determined as the target field.

[0053] In various business systems, the same type of data is often uniformly managed through fields, such as various types of data like names, contact information, addresses, real estate status, etc. The attribute information is the information related to the attributes of the target field, such as various attributes like the name, description, permissions, expiration date, creation time, cancellation time, etc. of the target field. The field data is the data corresponding to the target field, that is, the specific data corresponding to the field. In a specific application, if the target field is the license plate number, the attribute information of the target field can include the name of the target field "license plate number", and the description of the target field "the license plate number of the vehicle under the user's name, including motor vehicles and vehicles with license plate numbers such as motorcycles", and the field data of the target field can be the license plate number data specifically corresponding to each user. For example, Zhang San's is "Yue AXXXXX", Li Si's is "Yue BXXXXX", and Zhao Wu's is "Yue BXRRRRX", etc. That is, the attribute information of the target field is related to the target field itself and may not involve the specific data of the user. It describes the target field itself; while the field data is the various specific data belonging to the target field, including the data belonging to the target field generated by each business user in the business system.

[0054] Field characteristic factors are determined based on the attribute information and field data of the target field. Specifically, characteristic factors corresponding to the target field can be determined based on the attribute information and field data, and the field characteristic factors corresponding to the target field can be obtained by combining the characteristic factors. Field characteristic factors are elements that construct the field characteristics of the target field. Specifically, characteristic factors can be extracted from the field characteristic factors and combined to form the field characteristics of the target field, thereby enabling classification processing of the target field based on the field characteristics of the target field. In a specific implementation, corresponding characteristic factors can be constructed based on the attribute information and field data, and the characteristic factors of the target field can be obtained based on the constructed characteristic factors. For example, for field data, the field data can be directly used as the data characteristic factor, and the field characteristic factor can be obtained based on the data characteristic factors. Alternatively, the field data can be further divided into field sub-data, and corresponding data characteristic factors can be constructed for each field sub-data. The field characteristic factor can be obtained based on the data characteristic factors corresponding to each field sub-data. If there are more than one field characteristic factors, the field characteristic factors can be combined as needed to construct field characteristics with different expressiveness, thereby accurately classifying the target field based on the corresponding field characteristics using different classification methods.

[0055] Specifically, when classifying data, the server determines the attribute information and field data of the target field to which the data to be classified belongs, and determines more than one field characteristic factors corresponding to the target field based on the attribute information and field data of the target field. In a specific implementation, for example, when the server classifies data in a data table in a business system, the server determines the target field to which the data to be classified belongs, and obtains the attribute information and field data of the target field in the business system. Based on the obtained attribute information and field data, the server determines more than one field characteristic factors corresponding to the target field. The field characteristic factors serve as construction elements of field characteristics and are used to construct field characteristics corresponding to the target field.

[0056] Step 204 : combining the field feature factors according to the feature combination conditions corresponding to at least two classification methods to obtain the field features corresponding to each classification method.

[0057] The classification method refers to the method used to classify the target field. Different classification methods can be based on different characteristics of the target field. When classifying the target field using different classification methods, it is necessary to construct the field characteristics required by the corresponding classification method so that the field characteristics of the target field can be classified using the corresponding classification method to achieve classification of the target field. Classifying the target field can specifically include classifying the target field into standard fields. Standard fields can be pre-defined standard names for fields in data classification. By classifying the target field, the target field is matched to the corresponding standard field, and the corresponding data of the target field is hierarchically processed according to the level mapping relationship corresponding to the standard field. For example, if the target field "name" corresponds to the user's name, the target field can be classified into the standard field "name" to achieve hierarchical processing of the data corresponding to the target field "name" according to the level mapping relationship of the standard field "name".

[0058] There are at least two classification methods, so that the target field can be classified and processed separately through multiple classification methods. The feature combination condition is based on the field feature factor corresponding to the target field, and constructs the conditions of the field features of the target field required by the classification method when performing classification. When classifying the target field, each classification method can use different field features of the target field. The field features are based on the field feature factors corresponding to the target field. The field features can be combined through the feature combination conditions corresponding to the classification method to form field features suitable for classification processing by the corresponding classification method. In specific implementation, the feature combination conditions corresponding to each classification method can be determined based on historical classification data. For example, for classification method A, accurate classification of the corresponding field can be achieved based on field feature α, and field feature α is obtained by combining field feature factor 1 and field feature factor 2. The feature combination condition corresponding to classification method A can be set to form field feature α by combining field feature factor 1 and field feature factor 2. For another example, for classification method B, accurate classification of the corresponding field can be achieved based on field feature β, and field feature β is obtained by combining field feature factor 1 and field feature factor 4. The feature combination condition corresponding to classification method B can be set to form field feature β by combining field feature factor 1 and field feature factor 4. Generally, different fields are suitable for different classification methods, that is, different classification methods have different effects on the classification of various fields. When a field is unknown, the field can be classified separately through multiple classification methods. The classification results of various classification methods can be combined to achieve classification of the field, thereby ensuring the accuracy of the field classification processing.

[0059] Specifically, after obtaining more than one field characteristic factor corresponding to the target field, the server determines at least two classification methods and obtains the characteristic combination conditions corresponding to each classification method. The characteristic combination conditions are pre-set based on the corresponding classification method. The characteristic combination conditions define the field characteristics of the target field required by the classification method when the target field is classified by the corresponding classification method. The server combines the field characteristic factors according to the characteristic combination conditions corresponding to each classification method. Specifically, the server can extract the target characteristic factor from the field characteristic factor according to the characteristic combination conditions, and generate the field characteristics corresponding to the classification method based on the target characteristic factor. The classification method can classify the target field based on the field characteristics. In specific applications, when combining the field characteristic factors according to the characteristic combination conditions to obtain the corresponding field characteristics, the number of field characteristic factors defined by the characteristic combination conditions can be one or more than one. For example, feature combination condition 1 can be to extract field feature factor X from each field feature factor as the target feature factor, and obtain field features based on field feature factor X; feature combination condition 2 can be to extract field feature factor X, field feature factor Y and field feature Z from each field feature factor as the target feature factor, and obtain field features based on field feature factor X, field feature factor Y and field feature Z.

[0060] Step 206 , classify the target field according to each classification method and the field characteristics corresponding to the corresponding classification method, and obtain preliminary classification results corresponding to each classification method.

[0061] The preliminary classification result is the classification result obtained by classifying the target field according to each classification method based on the field characteristics corresponding to each classification method. At least two preliminary classification results can be obtained by classifying the target field using at least two classification methods based on the classification characteristics corresponding to each classification method. The preliminary classification results obtained are obtained by classifying different field characteristics based on different classification methods, and the preliminary classification results can be the same or different. Specifically, the preliminary classification result may include a classification result that classifies the target field into a standard field, such as classifying the target field into standard field 1 for classification method 1 and classifying the target field into standard field 3 for classification method 2.

[0062] Specifically, after obtaining the field features corresponding to each classification method, the server classifies the target field according to each classification method and the field features corresponding to the corresponding classification method, thereby obtaining the preliminary classification results corresponding to each classification method. For example, if the at least two classification methods include three classification methods, namely, a field data matching method, a field text classification method, and a neural network model classification method, then the first field features corresponding to the field data matching method can be classified according to the field data matching method to obtain a first preliminary classification result corresponding to the field data matching method; the second field features corresponding to the field text classification method can be classified according to the field text classification method to obtain a second preliminary classification result corresponding to the field text classification method; the third field features corresponding to the neural network model classification method can be classified according to the neural network model classification method to obtain a third preliminary classification result corresponding to the neural network model classification method, that is, the preliminary classification results for classifying the target field include a first preliminary classification result, a second preliminary classification result, and a third preliminary classification result, and the field classification result of the target field can be further determined based on the first preliminary classification result, the second preliminary classification result, and the third preliminary classification result.

[0063] Step 208: perform fusion based on the preliminary classification results to obtain the field classification result of the target field.

[0064] The field classification result is the classification result obtained after classifying the target field. Specifically, it may include matching the target field to a standard field, that is, determining the standard field corresponding to the target field. For example, for the target field "Date", the corresponding field classification result may be the standard field "Date of Birth". Therefore, the target field is classified into the standard field, and the data corresponding to the target field is graded according to the standard field "Date of Birth".

[0065] Specifically, after obtaining the preliminary classification results corresponding to each classification method, the server fuses the preliminary classification results according to each classification method to obtain the field classification result of the target field. The fusion processing of the preliminary classification results corresponding to each classification method can be flexibly set according to actual needs, such as fusion based on confidence or fusion in a weighted manner. For example, the server can determine the confidence corresponding to the preliminary classification results corresponding to each classification method, and determine the preliminary classification result with the highest confidence as the field classification result of the target field. For another example, the server can determine the classification weight corresponding to each classification method, and perform weighted processing on the preliminary classification results corresponding to each classification method based on the classification weight to obtain a weighted classification result, and determine the field classification result of the target field based on the weighted classification result.

[0066] Step 210: Obtain the data level of the data corresponding to the target field based on the field classification result mapping.

[0067] The data level refers to the level of the data corresponding to the target field. The data level can be set according to the actual needs of data classification in the business system, such as the data security level, the data confidentiality level, the data weight level, etc.

[0068] Specifically, after obtaining the field classification result for the target field, the server maps the target field to obtain the data level of the data corresponding to the target field based on the field classification result. Specifically, the server can perform level mapping on the target field based on the field classification result to determine the level corresponding to the target field, and then determine the data level of the data corresponding to the target field based on the level corresponding to the target field. For example, the level corresponding to the target field can be used as the data level of the data corresponding to the target field. In specific applications, the level mapping relationship between the data levels corresponding to the target field obtained based on the field classification result mapping can be set according to actual needs. For example, in a financial business system, the security level of the standard field "transaction amount" can be pre-set to high sensitivity. If the field classification result of the target field is the standard field "transaction amount", that is, the common attribute of all data corresponding to the target field is also the standard field "transaction amount", then the target field can be considered to match the standard field "transaction amount". The server can map the level of the target field to the level of the standard field "transaction amount", that is, map the target field to the high sensitivity level, and further determine that the data level of the data corresponding to the target field is also the high sensitivity level, thereby achieving hierarchical processing of the data corresponding to the target field.

[0069] In a specific application, after obtaining the data level of the data corresponding to the target field based on the field classification result mapping, if the target field belongs to a field in the data table, each field in the data table can be traversed to determine the data level of the data corresponding to each field in the data table, and the level of the data table can be determined based on the data level of the data corresponding to each field in the data table, thereby achieving hierarchical processing of the data table. For example, if data table S1 includes three fields with a level of three, the table level of data table S1 can be determined to be level three; for another example, if the highest level among the data levels of the data corresponding to each field in data table S2 is level five, the table level of data table S2 can be determined to be the highest level among the data levels of the data corresponding to each field, that is, the table level of data table S2 can be determined to be level five.

[0070] In the above-mentioned data classification method, more than one field characteristic factor is determined based on the attribute information and field data of the target field in the business system, the field characteristic factors are combined according to the characteristic combination conditions corresponding to at least two classification methods, and the target field is classified according to the field characteristics corresponding to each classification method according to each classification method. The preliminary classification results corresponding to each classification method are fused to obtain the field classification result of the target field, and the data level of the data corresponding to the target field is obtained based on the field classification result mapping. In the data classification process, the target field is classified by fusing the preliminary classification results of the target field using multiple classification methods, so as to obtain the data level of the corresponding data based on the field classification result mapping of the target field. This can cover the applicable scenarios of various classification methods, avoid repeated confirmation of the classification results, and improve the processing efficiency of data classification.

[0071] In one embodiment, field feature factors are combined according to feature combination conditions corresponding to at least two classification methods to obtain field features corresponding to each classification method, including: determining at least two classification methods; determining target feature factors corresponding to each classification method from field feature factors according to feature combination conditions corresponding to at least two classification methods; and obtaining field features corresponding to each classification method based on the target feature factors corresponding to each classification method.

[0072] The classification method refers to the method used to classify the target field. Different classification methods are based on the characteristics of the target field and the classification method. For example, when classifying a target field, if classification method 1 requires classification based on field characteristic A of the target field, while classification method 2 requires classification based on field characteristic B, then field characteristic A corresponding to classification method 1 and field characteristic B corresponding to classification method 2 must be constructed. Classification method 1 classifies the target field based on field characteristic A, while classification method 2 classifies the target field based on field characteristic B. The field characteristics of the target field are constructed based on field characteristic factors. Specifically, field characteristic factors can be combined according to the requirements of the classification method to obtain the field characteristics of the target field. The need for combining field characteristic factors to obtain corresponding field characteristics for various classification methods can be described using the classification method's feature combination conditions. Feature combination conditions are set in accordance with the classification method. Different classification methods can have different feature combination conditions. These feature combination conditions allow field characteristic factors to be combined to obtain field characteristics that meet the classification processing requirements of the classification method. Specifically, if a certain classification method is implemented through a neural network model, the feature combination condition corresponding to the classification method can be a condition for constructing field features as input to the neural network model based on a combination of field feature factors. The field feature factors can be combined through the feature combination condition to construct corresponding field features, and the field features are input into the neural network model so that the neural network model classifies the target field based on the input field features to obtain a preliminary classification result corresponding to the classification method.

[0073] The target characteristic factor is a characteristic factor determined from the field characteristic factors to construct the field characteristic based on the characteristic combination conditions corresponding to the classification method. The number of target characteristic factors can be one or more. When the target characteristic factor is one, it indicates that the characteristic combination conditions corresponding to the classification method determine that the field characteristic is composed of only one field characteristic factor, that is, the corresponding field characteristic is constructed using only one field characteristic factor.

[0074] Specifically, after obtaining more than one field characteristic factor corresponding to the target field, the server determines at least two classification methods. Each classification method can be configured based on actual needs, and the number of classification methods is greater than or equal to two, so that the target field can be classified using at least two classification methods. Each classification method can utilize different field characteristics of the target field when classifying the target field, such as classification based on field characteristics of different dimensions. Specifically, classification can be based on field characteristics constructed based on different field characteristic factors. In a specific implementation, the server can also determine at least two classification methods based on the business system to which the target field belongs. Based on the business system to which the target field belongs and the data types involved in the business system, the server can determine a classification method suitable for hierarchical processing of various types of data in the business system, thereby determining at least two classification methods. In addition, the server can also determine at least two classification methods based on data classification processing requirements. For example, based on the accuracy requirement of the data classification, the server can select at least two classification methods that meet the accuracy requirement. For example, based on the processing efficiency requirement of the data classification, the server can select at least two classification methods that meet the processing efficiency requirement. The server can also determine at least two classification methods that meet both accuracy and processing efficiency requirements.

[0075] After determining at least two classification methods, the server obtains feature combination conditions corresponding to each classification method. The feature combination conditions are pre-configured based on the corresponding classification method. In specific applications, different classification methods require obtaining different field features for each field. Based on the field features targeted by each classification method when classifying various fields, the feature combination conditions corresponding to each classification method are determined. Based on the feature combination conditions corresponding to each classification method, the server determines the target feature factors corresponding to the corresponding classification method from the field feature factors. Specifically, the server screens the feature factors in the field feature factors based on the feature combination conditions corresponding to the classification method to select the target feature factors required for constructing field features suitable for classification processing using that classification method. Different feature combination conditions can be set for each classification method, and the target feature factors determined by the feature combination conditions can be different. After determining the target feature factors corresponding to each classification method, the server constructs the field features corresponding to each classification method based on the target feature factors corresponding to each classification method. In specific implementations, after determining the target feature factors corresponding to a classification method, the server can combine the target feature factors based on the feature combination conditions corresponding to that classification method to obtain the field features corresponding to that classification method. In addition, the server may also directly determine the field characteristics corresponding to the classification method based on the target characteristic factor, such as directly using the target characteristic factor as the field characteristics corresponding to the classification method.

[0076] In this embodiment, the server determines the target characteristic factor from more than one field characteristic factors corresponding to the target field according to the characteristic combination conditions corresponding to at least two determined classification methods, and obtains the field characteristics corresponding to the corresponding classification method based on the target characteristic factor. Therefore, the server can construct the field characteristics required for the corresponding classification method according to the characteristic combination conditions corresponding to the field characteristic factors according to multiple classification methods, so as to classify the target field according to the corresponding field characteristics according to the classification method, thereby realizing classification of the target field through multiple classification methods, which is beneficial to covering the applicable scenarios of various classification methods, avoiding repeated confirmation of the grading results, and thus beneficial to improving the processing efficiency of data grading.

[0077] In one embodiment, the classification method includes a field data matching method, and the target feature factor includes a field data factor; based on the target feature factor corresponding to each classification method, the field feature corresponding to each classification method is obtained, including: determining that the field data factor is the first field feature corresponding to the field data matching method.

[0078] The field data matching method refers to a method for classifying a target field by performing text matching based on the field data corresponding to the target field. Field data refers to the data corresponding to the target field, that is, the specific data corresponding to the field. The field data matching method is a classification method that performs text matching on the data content corresponding to the target field. When performing text matching on field data, text matching on field data can be performed based on a regular matching algorithm to classify the target field. A regular matching algorithm performs text matching on field data using regular expressions. A regular expression is a set of characters that describes the characteristics of a string. A regular expression can be composed of ordinary strings or very complex, often using special characters to represent a range of characters, repeated occurrences, or context. By performing text matching on the regular expression with the field data corresponding to the target field, classification processing of the target field can be achieved. The target feature factor includes the field data factor, which is a feature factor determined based on the field data of the target field. In a specific implementation, the field data factor can be directly derived from the field data of the target field, such as by directly using the field data of the target field as the field data factor. Alternatively, the field data of the target field can be preprocessed, such as by performing deduplication and redundancy preprocessing, and the field data factor can be derived based on the preprocessing results. The first field feature is the field feature required to classify the target field based on the field data matching method.

[0079] Specifically, the at least two classification methods obtained by the server include a field data matching method. Based on the feature combination condition corresponding to the field data matching method, the target feature factor determined from the field feature factors includes the field data factor. After obtaining the field data factor, the server directly determines the field data factor as the first field feature corresponding to the field data matching method. That is, the server directly uses the field data factor as the field feature required by the field data matching method.

[0080] Furthermore, according to each classification method, the target field is classified according to the field characteristics corresponding to the corresponding classification method to obtain the preliminary classification results corresponding to each classification method, including: through the field data matching method, the target field is classified based on the first field characteristics to obtain the preliminary classification results corresponding to the field data matching method.

[0081] The preliminary classification result is a classification result obtained by classifying the target field according to the first field feature in accordance with the field data matching method. Specifically, after obtaining the first field feature corresponding to the field data matching method, the server classifies the target field based on the first field feature using the field data matching method. For example, the server may classify the target field based on the first field feature using a regular matching algorithm to obtain a preliminary classification result corresponding to the field data matching method. In a specific application, after obtaining the preliminary classification result corresponding to the field data matching method, the server may also perform verification processing on the preliminary classification result. After the verification passes, the preliminary classification result corresponding to the field data matching method is obtained.

[0082] In this embodiment, for the field data matching method, the server determines the field data factor in the field feature factor as the first field feature corresponding to the field data matching method, and uses the field data matching method to classify the target field based on the first field feature, thereby obtaining a preliminary classification result corresponding to the field data matching method. Using the field data matching method to classify the target field based on the first field feature determined from the field data of the target field ensures that the field data matching method can accurately classify the target field based on the field data of the target field, thereby avoiding repeated confirmation of classification results applicable to the field data matching method, thereby improving data classification processing efficiency.

[0083] In one embodiment, a target field is classified based on a first field feature through a field data matching method to obtain a preliminary classification result corresponding to the field data matching method, including: obtaining a field data matching condition corresponding to the field data matching method; performing regular matching on the first field feature and the field data matching condition to obtain a regular matching result; when the regular matching result passes the matching verification, obtaining a preliminary classification result corresponding to the field data matching method according to the regular matching result.

[0084] The field data matching conditions are conditions for performing text matching on the first field feature. The field data matching conditions may include various standard fields and corresponding matching conditions. By matching the first field feature with each matching condition in the field data matching conditions, a classification result for the target field is obtained based on the standard fields corresponding to the matching conditions that match consistently. For example, the target field may be classified as a standard field that matches consistently. In a specific application, the field data matching conditions may include various standard fields and corresponding regular expressions. The regular expressions may be used to perform text matching on the first field feature, thereby obtaining a preliminary classification result for the target field based on the standard fields corresponding to the regular expressions. For example, the regular expression may include "^[\w-]+(\.[\w-]+)*@([\w-]+\.)+(com|cn|net)$" to identify and classify the email field. Specifically, the field data corresponding to the target field may be matched with the regular expression. If a match is successful, the target field may be determined to be a "mailbox" field, i.e., the classification result for the target field is "mailbox."

[0085] Specifically, after obtaining a first field feature corresponding to a field data matching method, the server obtains a field data matching condition corresponding to the field data matching method. The field data matching condition can be pre-set based on actual needs. Specifically, a regular expression can be constructed based on a field with a distinct rule, and the field data matching condition is then determined based on the constructed regular expression. The server performs regular matching on the first field feature with the field data matching condition. The field data matching condition can include various standard fields and their corresponding regular expressions. By performing regular matching on the first field feature with the regular expression in the field data matching condition, a regular matching result is obtained. The server performs a matching check on the obtained regular matching result to ensure the accuracy of the regular matching. If the regular matching result passes the matching check, indicating a high degree of credibility of the regular matching result, the server then obtains a preliminary classification result corresponding to the field data matching method based on the regular matching result. Specifically, the server performs regular matching on the first field feature with the regular expression in the field data matching condition. If a match is successful, the standard field corresponding to the successfully matched regular expression can be determined as the preliminary classification result corresponding to the field data matching method, thereby achieving classification processing of the target field using the field data matching method.

[0086] In specific applications, for fields with distinct characteristics in business scenarios, field data matching can be used to classify the field data to ensure accurate classification. For example, mobile phone numbers are fixed at 11 digits, with the first digit being 1; email addresses include the suffix "@.com / cn / net" in their field data; and license plate numbers have the Chinese abbreviations of the provinces and cities. For these fields with distinct characteristics, constructing corresponding regular expressions to perform regular matching on the field data allows for accurate classification.

[0087] In this embodiment, a regular expression match is performed on the first field feature against the field data matching condition corresponding to the field data matching method. When the regular expression match result passes the match verification, a preliminary classification result corresponding to the field data matching method is obtained based on the regular expression match result. Regular expression matching is performed on the first field feature based on the field data matching condition, allowing accurate classification of fields in the business system that are suitable for regular expression matching.

[0088] In one embodiment, the classification method includes a field text classification method, and the target feature factors include a field description factor and a field name factor; based on the target feature factors corresponding to each classification method, the field features corresponding to each classification method are obtained, including: combining the field description factor and the field name factor to obtain a second field feature corresponding to the field text classification method.

[0089] The at least two classification methods determined include a field text classification method. This method classifies the target field by performing text classification based on the field description and field name corresponding to the target field. The field description describes the content of the field, such as text explaining the concept and function of the field. The field description can be edited when the corresponding field is created, or users can edit and add to the field at any time during the operation of the business system to ensure normal use of the field. The field name refers to the name of the field, which can specifically include names corresponding to various languages, such as Chinese, English, and French. Both the field description and the field name are attribute information of the field and can be obtained from the field's attribute information. The field text classification method classifies the field description and field name in the attribute information corresponding to the target field. When performing text classification on the field description and field name, the field description and field name can be classified based on a text classification model to classify the target field. If the field description and field name corresponding to the field are short, the text classification model corresponding to the field text classification method can be a short text classification model, thereby accurately classifying the target field based on the short text classification model. For example, the short text classification model can be a FastText fast text classification model, a TF-IDF (Term Frequency–Inverse Document Frequency) + XGB (eXtreme Gradient Boosting) text classification model, etc.

[0090] The target feature factors include the field description factor and the field name factor. The field description factor is a feature factor determined based on the field description of the field, and the field name factor is a feature factor determined based on the field name of the field. In specific implementations, the field description factor can be directly obtained based on the field description of the target field, and the field name factor can be obtained based on the field name of the target field, such as directly using the field description of the target field as the field description factor and the field name of the target field as the field name factor. The field description and field name of the target field can also be preprocessed, such as by performing deduplication and de-redundancy preprocessing, and then obtaining the field description factor and field name factor based on the preprocessing results. The second field feature is the field feature used when classifying the target field according to the field text classification method.

[0091] Specifically, the at least two classification methods acquired by the server include a field text classification method. Based on the feature combination conditions corresponding to the field text classification method, the target feature factors determined from the field feature factors include a field description factor and a field name factor. After obtaining the field description factor and the field name factor, the server combines the field description factor and the field name factor to obtain a second field feature corresponding to the field data matching method. That is, after combining the field description factor and the field name factor, the server obtains the field feature required for classifying the target field using the field text classification method.

[0092] Furthermore, according to each classification method, the target field is classified according to the field characteristics corresponding to the corresponding classification method to obtain the preliminary classification results corresponding to each classification method, including: through the text classification model corresponding to the field text classification method, the target field is classified based on the second field characteristics to obtain the preliminary classification results corresponding to the field text classification method.

[0093] Among them, the preliminary classification result is the classification result obtained by classifying the target field according to the second field feature according to the field text classification method. Specifically, after obtaining the second field feature corresponding to the field text classification method, the server classifies the target field based on the second field feature through the field text classification method, such as by using the text classification model corresponding to the field text classification method to classify the target field based on the second field feature, and obtain the preliminary classification result corresponding to the field text classification method. Specifically, after obtaining the second field feature corresponding to the field text classification method, the server determines the text classification model corresponding to the field text classification method. The text classification model is pre-trained based on historical business data according to actual needs. The server classifies the target field based on the second field feature through the queried text classification model. Specifically, the second field feature can be used as the input of the text classification model, that is, the second field feature is input into the text classification model, and the text classification model classifies the target field according to the second field feature, and outputs the preliminary classification result corresponding to the target field.

[0094] In this embodiment, for the field text classification method, the server determines the target feature factor based on the field description factor and the field name factor in the field feature factor, combines the field description factor and the field name factor to obtain a second field feature corresponding to the field text classification method, and classifies the target field based on the second field feature using the field text classification method to obtain a preliminary classification result corresponding to the field text classification method. Utilizing the field text classification method, the target field is classified based on the second field feature determined by the field description and field name of the target field, and the target field is text-classified based on the attribute information of the target field. This helps avoid repeated confirmation of the classification results applicable to the field text classification method, thereby improving the processing efficiency of data classification.

[0095] In one embodiment, the target field is classified based on the second field feature through a text classification model corresponding to the field text classification method to obtain a preliminary classification result corresponding to the field text classification method, including: performing vector conversion on the second field feature through a text classification model corresponding to the field text classification method to obtain a word vector; and performing linear classification on the target field based on the word vector through the text classification model to obtain a preliminary classification result corresponding to the field text classification method.

[0096] The text classification model can be pre-trained based on historical data from the business system. Based on the input attribute information of the target field, specifically the field description and field name, the text classification model classifies the target field and outputs a preliminary classification result for the target field. If the field description and field name are short, the text classification model corresponding to the field text classification method can be a short text classification model, thereby accurately classifying the target field based on the short text classification model. For example, the short text classification model can be a FastText fast text classification model or a TF-IDF (Term Frequency–Inverse Document Frequency) + XGB (eXtreme Gradient Boosting) text classification model. The word vector is the result of transforming the second field features into a vector. By transforming the second field features into a vector, an input vector suitable for classification by the text classification model can be obtained. The text classification model then performs linear classification of the target field based on the word vector obtained from the vector transformation. This can be implemented using a linear classifier within the text classification model.

[0097] Specifically, after obtaining the second field feature corresponding to the field text classification method, the server queries the text classification model corresponding to the field text classification method. The text classification model is pre-trained based on historical data in the business system according to actual needs. The server uses the second field feature as input to the text classification model, and inputs the second field feature into the text classification model, so that the text classification model performs vector conversion on the second field feature to obtain a word vector. Furthermore, the text classification model performs linear classification on the target field based on the word vector obtained by the vector conversion. Specifically, the word vector can be classified by a linear classifier in the text classification model to obtain a preliminary classification result corresponding to the field text classification method.

[0098] In specific applications, the type of text classification model corresponding to the field text classification method can be set according to actual needs. For example, if there is only one text classification model, such as a short text classification model, the target field can be classified using the short text classification model. Alternatively, if there are more than one text classification model, the target field can be classified using each of the multiple text classification models, and the classification results of the various text classification models can be combined to obtain a preliminary classification result corresponding to the field text classification method. Regarding the selection of the text classification model corresponding to the field text classification method, multiple text classification models constructed using different algorithms can be pre-trained based on historical data in the business system. Based on the accuracy of the various text classification models in classifying the field, the text classification model corresponding to the field text classification method can be determined. For example, the text classification model with the highest field classification accuracy can be selected as the text classification model corresponding to the field text classification method. In specific implementations, text classification models constructed using various algorithms can also be selected as the text classification model corresponding to the field text classification method, and the target field can be classified using multiple text classification models. The classification results of the various text classification models can then be combined to obtain a preliminary classification result corresponding to the field text classification method. For example, you can set classification weights for various text classification models and perform a weighted fusion of the classification results based on the classification weights to obtain a preliminary classification result corresponding to the field text classification method. In specific implementations, if the rough type corresponding to the target field has been obtained, you can select the corresponding text classification model based on the rough type of the target field and perform field classification on the target field to ensure the accuracy of the field classification results.

[0099] In this embodiment, the second field feature is sequentially vectorized and linearly classified through the text classification model corresponding to the field data matching method to obtain a preliminary classification result corresponding to the field text classification method. The second field feature is subjected to field classification based on the text classification model corresponding to the field text classification method, and the fields in the business system that are suitable for the text classification model can be accurately classified.

[0100] In one embodiment, the classification method includes a neural network model classification method, and the target feature factor includes a field data factor and a field description factor; based on the target feature factor corresponding to each classification method, the field feature corresponding to each classification method is obtained, including: combining the field data factor and the field description factor to obtain a third field feature corresponding to the neural network model classification method.

[0101] The at least two classification methods identified include a neural network model classification method. This method uses a pre-trained artificial neural network model to classify the target field based on the field data and field description corresponding to the target field. Field data refers to the data corresponding to the target field, that is, the specific data corresponding to the field. The field description describes the content of the field, such as text explaining the concept and function of the field. The field description can be edited when the corresponding field is created, or can be edited and added by the user at any time during the operation of the business system to facilitate normal use of the field. The field description is attribute information of the field and can be obtained from the field's attribute information. The neural network model classification method classifies the target field based on the field data corresponding to the target field and the field description in the attribute information. When classifying the target field based on the field data and field description, the field data and field description can be classified based on the field classification network model, thereby achieving classification of the target field. The field classification network model is a network model pre-trained using historical data from the business system using a machine learning algorithm, such as an artificial neural network algorithm or a deep learning algorithm. Machine learning (ML) is a multidisciplinary field that encompasses probability theory, statistics, approximation theory, convex analysis, and algorithmic complexity theory. It specifically studies how computers can simulate or implement human learning behaviors to acquire new knowledge or skills and reorganize existing knowledge structures to continuously improve their performance. Machine learning is at the core of artificial intelligence and the fundamental way to make computers intelligent. Its applications span all areas of AI. Machine learning and deep learning typically include techniques such as artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and self-learning.

[0102] The target feature factor includes a field data factor and a field description factor. The field data factor is a feature factor determined based on the field data of the target field. In specific implementation, the field data factor can be obtained directly based on the field data of the target field, such as directly using the field data of the target field as the field data factor; the field description factor is a feature factor determined based on the field description of the field. In specific implementation, the field description factor can be obtained directly based on the field description of the target field, such as directly using the field description of the target field as the field description factor; the field description of the target field can also be preprocessed, such as after deduplication and redundancy preprocessing, the field description factor is obtained based on the preprocessing result. The third field feature is the field feature used when classifying the target field according to the neural network model classification method.

[0103] Specifically, the at least two classification methods acquired by the server include a neural network model classification method. Based on the feature combination conditions corresponding to the neural network model classification method, the target feature factor determined from the field feature factors includes a field data factor and a field description factor. After obtaining the field data factor and the field description factor, the server combines the field data factor and the field description factor to obtain a third field feature corresponding to the neural network model classification method. That is, after the server combines the field data factor and the field description factor, it obtains the field feature required for classifying the target field using the neural network model classification method.

[0104] Furthermore, according to each classification method, the target field is classified according to the field characteristics corresponding to the corresponding classification method to obtain the preliminary classification results corresponding to each classification method, including: using the field classification network model corresponding to the neural network model classification method, classifying the target field based on the third field characteristics to obtain the preliminary classification results corresponding to the neural network model classification method.

[0105] Among them, the preliminary classification result is the classification result obtained by classifying the target field according to the third field feature according to the neural network model classification method. Specifically, after obtaining the third field feature corresponding to the neural network model classification method, the server classifies the target field based on the third field feature through the neural network model classification method, such as the field classification network model corresponding to the neural network model classification method, classifying the target field based on the third field feature, and obtaining the preliminary classification result corresponding to the neural network model classification method. The field classification network model can be pre-trained through a machine learning algorithm, such as an LSTM (Long-Short Term Memory) model, a BiLSTM (Bi-directional Long Short-Term Memory) model, a GRU (Gated Recurrent Unit) model, and a TextCNN (Text Convolutional Neural Network) model.

[0106] Specifically, after obtaining the third field feature corresponding to the neural network model classification method, the server determines the field classification network model corresponding to the neural network model classification method. The field classification network model is pre-trained based on historical business data according to actual needs through a machine learning algorithm. The server classifies the target field based on the third field feature through the queried field classification network model. Specifically, the third field feature can be used as the input of the field classification network model, that is, the third field feature is input into the field classification network model, and the field classification network model classifies the target field according to the third field feature and outputs the preliminary classification result corresponding to the target field. In specific implementation, if the preliminary type corresponding to the target field has been obtained, the corresponding field classification network model can be selected based on the preliminary type of the target field to perform field classification on the target field to ensure the accuracy of the field classification result.

[0107] In this embodiment, for the neural network model classification method, the server determines a target feature factor from the field data factor and the field description factor in the field feature factor, combines the field data factor and the field description factor to obtain a third field feature corresponding to the neural network model classification method, and classifies the target field based on the third field feature using the neural network model classification method to obtain a preliminary classification result corresponding to the neural network model classification method. Using the neural network model classification method to classify the target field based on the third field feature determined from the field data and field description of the target field helps avoid repeated confirmation of classification results applicable to the neural network model classification method, thereby improving data classification processing efficiency.

[0108] In one embodiment, Figure 3 As shown, the data classification method also includes a process of correcting the field classification result of the target field, specifically including:

[0109] Step 302: Determine the source field corresponding to the target field based on the field lineage relationship of the target field.

[0110] Among them, the blood relationship of data refers to the connection formed between data during the process of generation, processing, circulation and extinction. Common blood relationships include the blood relationship of data tables and the blood relationship of data table fields, which respectively express which tables a table is calculated from and which fields of which tables a field is processed from. The field blood relationship describes the relationship between the generation, processing, circulation and extinction of each field. The source field is the field that generates the target field, that is, the source field to which the target field belongs. For example, for target field A, it is obtained by summing the data between field B and field C, that is, for the data in target field A, it is obtained by summing the data corresponding to field B and the data corresponding to field C. Then, it can be determined that target field A originates from field B and field C, that is, the source fields corresponding to target field A are field B and field C.

[0111] Specifically, the server can obtain the field lineage relationship of the target field. The field lineage relationship records the generation, processing, circulation and extinction relationship of the target field. According to the field lineage relationship, the source field corresponding to the target field can be determined. For example, the source of the target field can be traced to determine the source field corresponding to the target field.

[0112] Step 304 : Based on the source field classification result corresponding to the source field, the field classification result is classified and corrected to obtain a field classification result after classification correction.

[0113] The source field classification result is the field classification result corresponding to the source field. The source field classification result can be the field classification result obtained by classifying the source field. The source field classification result can also be obtained by configuring the source field when creating the source field. Specifically, after determining the source field corresponding to the target field, the server obtains the source field classification result corresponding to the source field, and classifies and corrects the field classification result of the target field based on the source field classification result, thereby obtaining the field classification result after classification correction. The field classification result after classification correction can be used as the final field classification result of the target field.

[0114] Furthermore, mapping the field classification result to obtain the data level of the data corresponding to the target field includes: mapping the field classification result after classification correction to obtain the data level of the data corresponding to the target field.

[0115] Specifically, after classifying and correcting the field classification result of the target field using the source field classification result corresponding to the source field, the server may determine the data level of the data corresponding to the target field based on the field classification result after classification correction.

[0116] In this embodiment, the field classification results are classified and corrected according to the field lineage relationship of the target field, so as to determine the data level of the data corresponding to the target field based on the field classification result mapping after classification correction. The field lineage relationship of the target field can be used to ensure the accuracy of the field classification results, which is conducive to improving the accuracy of data classification.

[0117] In one embodiment, the data level of the data corresponding to the target field is obtained based on the field classification result mapping, including: querying the field level mapping relationship in the business system; performing level mapping based on the field level mapping relationship and the field classification result to obtain the field level corresponding to the target field; and determining the data level of the data corresponding to the target field according to the field level.

[0118] The field-level mapping relationship records the field levels corresponding to various fields in the business system. Field-level mapping relationships can be pre-configured based on the business system's requirements. This allows different field-level mapping relationships to be configured for different fields in the business system, effectively grading the data corresponding to each field. For example, in a financial business system, the field levels corresponding to various fields involved in the financial business system can be configured based on the data classification and grading specifications used in financial scenarios. The field level refers to the level of the corresponding field, which can be security, encryption, storage, and other levels. The field level is the level of the data corresponding to the field. For example, if field X has a field level of five, then the data level of each data corresponding to field X is also level five. This means that the data level of the data to be graded can be determined based on the field level of the corresponding field. Therefore, the data level of each data corresponding to a field can be uniformly determined based on the field level of the field.

[0119] Specifically, after obtaining the field classification results for the target field, the server maps the field classification results to obtain the data level of the data corresponding to the target field. Specifically, the server can query the field level mapping relationship in the business system. The field level mapping relationship includes mapping relationships between various standard fields in the business system and their corresponding field levels. The field level mapping relationship is pre-set based on the needs of the business system. After obtaining the field level mapping relationship in the business system, the server performs level mapping on the target field based on the field level mapping relationship and the field classification results, and determines the field level corresponding to the target field. Specifically, the server can match the field classification results corresponding to the target field with the various standard fields in the field level mapping relationship, and determine the field level corresponding to the target field based on the field levels corresponding to the matching standard fields, thereby achieving level mapping for the target field. After determining the field level corresponding to the target field, the server determines the data level of the data corresponding to the target field based on the field level corresponding to the target field. For example, the server can directly use the field level corresponding to the target field as the data level corresponding to the target field.

[0120] In this embodiment, level mapping is performed based on the field level mapping relationship pre-configured in the business system and the field classification result of the target field, and the data level of the data corresponding to the target field is obtained based on the field level corresponding to the determined target field. Therefore, level mapping can be performed based on the needs of different business systems based on the field classification result of the target field to determine the data level of the corresponding data. When the field level mapping relationship in the business system changes, it is only necessary to redetermine the field level of the field based on the changed field level mapping relationship, and there is no need to reclassify the repeated fields, which is beneficial to improving the processing efficiency of data grading.

[0121] In one embodiment, the data classification method also includes: when a field creation request sent by a terminal is received, a field editing template is sent to the terminal; when the attribute information of the field to be created is obtained through the field editing template, the field is classified according to the attribute information of the field to be created, and a recommended classification result corresponding to the field to be created is obtained; the recommended classification result is sent to the terminal; the recommended classification result is used to instruct the terminal to determine the field category of the field to be created based on the recommended classification result; the field category of the field to be created is used to determine the data level of the data corresponding to the field to be created.

[0122] Among them, the terminal is connected to the server in communication, can access the business system, generate business data in the process of business services, and can also configure various fields for the business system. The field creation request is used to request the server to create a new field. In a specific application, a user with the authority to create a new field in the business system can send a field creation request to the server through the terminal to request the creation of a new field. The field editing template is the template used when creating a new field in the business system. The terminal can edit and configure the field editing template to realize the field creation process. The attribute information is information related to the attributes of the target field, such as the name, description, authority, validity period, creation time, cancellation time and other attributes of the target field. The recommended classification result is the field classification result obtained by performing field classification processing on the field to be created based on the attribute information of the field to be created. Specifically, the attribute information can be used to classify the field to be created through a variety of classification methods to obtain a recommended classification result. The recommended classification result is used to recommend the field category corresponding to the field to be created to the terminal, thereby simplifying the terminal user's configuration operation on the field to be created, which is conducive to improving the processing efficiency of the new field.

[0123] Specifically, when a server receives a field creation request from a terminal, indicating that the terminal has triggered the need to create a new field in a business system, the server retrieves a field editing template. This template can be configured based on the business system, meaning different business systems can have different field editing templates. The server then returns the retrieved template to the terminal. After the terminal receives the field editing template from the server, the user can edit the template to create the new field. The server monitors the user's input in the field editing template. Upon obtaining attribute information of the field to be created from the field editing template, such as the field name and description, the server pre-classifies the field to be created based on the obtained attribute information, obtains a recommended classification result for the field to be created, and sends the recommended classification result to the terminal, displaying it in the field classification section of the field editing template. The terminal user can then set a field category for the field to be created based on the recommended classification result in the field classification section of the field editing template. This allows the user to configure the field category corresponding to the field during the field creation process, thus enabling classification of the field to be created. The data level of the data corresponding to the field to be created can be determined based on the field category. Specifically, the server may obtain the data level of the data corresponding to the field to be created based on the field category mapping of the field to be created.

[0124] In this embodiment, the server can respond to the field creation request sent by the terminal and create a new field by sending a field editing template. When obtaining the attribute information of the field to be created, the server can classify the field to be created according to the attribute information of the field to be created, and send the obtained recommended classification result to the terminal to instruct the terminal to determine the field category of the field to be created according to the recommended classification result. The field category of the field to be created can be used to determine the data level of the data corresponding to the field to be created, thereby realizing pre-classification processing of the data corresponding to the field to be created when creating a new field in the business system, thereby improving the processing efficiency of data classification.

[0125] In one embodiment, the data classification method also includes: in response to a field creation operation triggered in a business system, displaying a field editing template for the field to be created; in the field classification item in the field editing template, displaying a recommended classification result corresponding to the field to be created; the recommended classification result is obtained by field classification based on the attribute information of the field to be created, and the attribute information of the field to be created is obtained by the editing operation on the field editing template; in response to the editing operation on the recommended classification result, displaying the field category of the field to be created in the field classification item; the field category is used to determine the data level of the data corresponding to the field to be created.

[0126] In this embodiment, the user can trigger the processing of creating a new field in the terminal. Among them, the field creation operation is a triggering operation of creating a new field in the business system. Specifically, a user with the authority to create a new data field can create a control for the field in the business system to trigger the field creation operation. The field editing template is the template used when creating a new field in the business system. The terminal can edit and configure the field editing template to realize the field creation processing. The field classification item is an operation item for configuring the field category of the field to be created. The recommended classification result is the field classification result obtained by performing field classification processing on the field to be created based on the attribute information of the field to be created. Specifically, the attribute information can be used to classify the field to be created through a variety of classification methods to obtain a recommended classification result. The attribute information is information related to the attributes of the target field, such as the name, description, authority, validity period, creation time, cancellation time and other attributes of the target field. The recommended classification result is used to recommend the field category corresponding to the field to be created to the terminal, thereby simplifying the terminal user's configuration operation on the field to be created, which is conducive to improving the processing efficiency of the new field.

[0127] Specifically, in response to a field creation operation triggered in a business system by a user with field creation permission, the terminal displays a field editing template for the field to be created. The field editing template can be obtained by a server. Specifically, the terminal can send a field creation request to the server to request the server to issue the field editing template. The field classification item in the field editing template displays recommended classification results for the field to be created. These recommended classification results are obtained by classifying the field based on the attribute information of the field to be created, which is obtained by editing the field editing template. If a user enters attribute information for the field to be created, such as a description or name, while editing the field editing template, the server can classify the field based on the attribute information and send the recommended classification results to the terminal for display in the field classification item of the field editing template. The user can edit the recommended classification results. In response to the user's editing operation, the terminal displays the field category of the field to be created in the field classification item. The field category determines the data level of the data corresponding to the field to be created. Specifically, the server may obtain the data level of the data corresponding to the field to be created based on the field category mapping of the field to be created.

[0128] In this embodiment, the user can trigger the processing of creating a new field through the terminal. In the field classification item in the field editing template, the recommended classification results obtained by classifying the field according to the attribute information of the field to be created are displayed to assist the user in configuring the field category of the field to be created. The configured field category of the field to be created can be used to determine the data level of the data corresponding to the field to be created, so that when a new field is created in the business system, the data corresponding to the field to be created is pre-classified, thereby improving the processing efficiency of data classification.

[0129] In one embodiment, the data classification method further includes: when the field to be created is used as the target field, and the segment classification result determined according to the target field does not match the field category of the field to be created, sending field classification exception information for the field to be created to the terminal.

[0130] Among them, the field classification exception information is used to prompt that the current field classification result of the field to be created is abnormal and needs to be corrected, so as to ensure the accuracy of the field classification result corresponding to the newly created field. Specifically, the server can use the field to be created as the target field to classify the field to be created and obtain the field classification result corresponding to the field to be created. The server matches the field classification result corresponding to the field to be created with the field category determined during the creation process of the field to be created. If the two do not match, it indicates that the field category determined during the creation process of the field to be created is abnormal. The server generates field classification exception information for the field to be created and sends the field classification exception information to the terminal to prompt the terminal to promptly correct the field classification result of the field to be created, thereby ensuring the accuracy of the field classification results corresponding to each field in the business system, which is conducive to the accuracy of grading the corresponding data of the field.

[0131] In this embodiment, the field to be created is used as the target field, and the target field is classified to obtain the field classification result of the field to be created. If the field classification result of the field to be created does not match the field category of the field to be created, it indicates that the field category configured when creating the field may be incorrect. The server then sends field classification exception information to the terminal to prompt the terminal to correct the field classification result of the field to be created, thereby ensuring the accuracy of the field classification results corresponding to each field in the business system, which is conducive to the accuracy of grading the corresponding data of the field.

[0132] In one embodiment, based on the attribute information and field data of the target field in the business system, more than one field characteristic factors corresponding to the target field are determined, including: determining the target field in the business system; obtaining the field description, field name and field data of the target field; and obtaining more than one field characteristic factors corresponding to the target field based on the field description, field name and field data.

[0133] The target field is the field in the business system that contains the data that requires data classification processing. The field description describes the content of the field, such as text that explains the concept and function of the field. The field description can be edited when the corresponding field is created, or users can edit and add to the field at any time during the operation of the business system to ensure normal use of the field. The field name refers to the name of the field, which can include names corresponding to various languages, such as Chinese, English, and French. The field data is the data corresponding to the target field, that is, the specific data corresponding to the field.

[0134] Specifically, the server determines the target field in the business system, and specifically can extract the fields corresponding to the data that needs to be processed by data classification from each field in the business system as the target field. For example, from the data table of the business system, each field in the data table can be used as the target field one by one to perform data classification processing on the data corresponding to each field in the data table. After determining the target field, the server obtains the field description, field name and field data corresponding to the target field, and obtains more than one field characteristic factors corresponding to the target field based on the obtained field description, field name and field data. Specifically, the server can obtain the field description factor based on the field description, obtain the field name factor based on the field name, obtain the field data factor based on the field data, and obtain the field characteristic factor corresponding to the target field based on the field description factor, field name factor and field data factor.

[0135] In this embodiment, based on the field description, field name and field data of the target field, more than one field characteristic factor corresponding to the target field is obtained, so that the target field can be classified using the corresponding field characteristics through a variety of different classification methods based on the field description, field name and field data of the target field. The target field can be further classified by integrating the preliminary classification results of the target field classified by multiple classification methods to obtain the data level of the corresponding data according to the field classification result mapping of the target field. It can cover the applicable scenarios of various classification methods, avoid repeated confirmation of the grading results, and improve the processing efficiency of data grading.

[0136] This application also provides an application scenario, which applies the above-mentioned data classification method. Specifically, the application of the data classification method in this application scenario is as follows:

[0137] With the continuous development of computers, the importance of various data types has become increasingly apparent, and data has gradually become an asset. Data assets refer to data resources, recorded physically or electronically, owned or controlled by individuals or enterprises and capable of generating future economic benefits for the enterprise. Data security has become a hot topic, and the country has enacted relevant laws to regulate data processing activities and ensure data security. Data security refers to taking necessary measures to ensure that data is effectively protected and legally used, as well as the ability to maintain a continuous state of security. Ensuring security throughout the entire data processing process requires data collection, storage, use, processing, transmission, provision, and disclosure.

[0138] For the financial industry, the People's Bank of China has issued various industry standards for data security classification within financial business systems, regulating data security management within financial institutions. Protecting user data security is a paramount responsibility for financial institutions, but it also presents challenges. These include determining what core data is, where is it stored, the integrity and compliance of data reporting, and compliance with data transmission, use, and sharing regulations. Data security classification is fundamental to building data lifecycle security capabilities. It allows for precise identification of sensitive data locations, facilitating the development of appropriate protection strategies. Data security classification involves quantifying the security classification results of data tables to identify the locations of top secret, confidential, highly sensitive, moderately sensitive, and low-sensitivity data tables, paving the way for the implementation of targeted data protection policies.

[0139] Traditional data security classification processing involves manually organizing historical forms and driving manual selection of classification targets. However, this approach is inefficient, has varying understandings of classification standards, and cannot guarantee the accuracy of conclusions. Furthermore, there are rule-based methods that use rules to identify fields with very obvious characteristics, such as ID cards, mobile phone numbers, and email addresses. However, these methods cover limited scenarios and make decisions based on a limited understanding of the data, making it easy to miss highly sensitive tables and limiting the scope of security protection. Algorithms are used to label the classification information of a portion of tables, and classification models are used to implement data security classification processing. The security classification of some tables is labeled as samples, and the security classification of other tables is predicted. This approach has poor interpretability and cannot effectively trace the security classification results. Furthermore, in scenarios where security targets change, work updates and migrations cannot be completed in a lightweight manner, impacting the efficiency of data classification processing.

[0140] Based on this, the data classification method provided in this embodiment is as follows: Figure 4 As shown, for the data tables in the financial business system, the data classification process is divided into two steps, including classifying the fields in the data table and classifying the data table, that is, first completing the classification of each field in the data table, maintaining the classification and classification strategy in combination with the specification requirements, and finally completing the output of the classification results. Specifically, the metadata in the financial business system includes incremental data tables and stock data tables. The incremental data table is a newly added data table, and the stock data table is an already stored data table. The incremental data in the incremental data table can be completed with the assistance of the model, and the classification of the stock data in the stock data table can be completed by the model. Specifically, as Figure 4As shown, for incremental data tables, during the table creation process, field classification recommendations are made for each field. The data table creator determines the corresponding field classification results for each field in the incremental data table based on the recommended field classifications, and security classifications are performed on the incremental data table based on the field classification results. For existing data tables, field classifications are performed based on both field data and field attributes, and the field classification results are corrected based on field lineage relationships. Security classifications are then performed on existing data tables based on the corrected field classification results. Data lineage relationships refer to the connections formed between data during the process of generation, processing, flow, and extinction. Common examples include data table lineage relationships and data table field lineage relationships, which respectively express which tables a table is calculated from and which fields in which tables a field is processed from.

[0141] Furthermore, field classification requires determining the field's classification target. This can be developed based on the business's data scenarios, as well as national and industry standards. For example, in a financial scenario, field information at data classification decision points can be prioritized based on financial industry regulations and standards to determine the field's classification target. For example, the field's classification target can be shown in Table 1 below.

[0142] Table 1

[0143]

[0144]

[0145] Data in financial scenarios can include basic personal information and financial information. This information can be further refined into various field classification targets. For example, basic personal information includes gender, nationality, mobile phone number, social media account, and other field classification targets. Fields in a data table are classified into pre-set classification targets, thereby standardizing and normalizing field names. Based on the field classification results, security classification of the data table can be implemented.

[0146] Furthermore, for the hierarchical processing of incremental data tables, a security classification selection entry for fields is provided during the data table creation process, so that the user who creates the data table can select the classification information to which the specific field belongs. At the same time, the classification of the corresponding field can be recommended in combination with the Chinese name and English name of the field, thereby improving the processing efficiency of data classification. Moreover, in order to avoid the situation where there is a large-scale misselection or wrong selection of classification during the data table creation process, the model post-verification capability can be configured for some fields with strong regularity. For example, during the data table creation process, the user selects the "gender" field, but the subsequent data content and distribution seriously do not match the characteristics of this field. For example, the data content of this field is not a binary value of male or female, but includes various types of numerical values ​​such as numbers, then an alarm and an abnormal reminder can be issued to prompt the user to make corrections based on the security classification of the field.

[0147] In a specific application, when a user creates a new data table, he can create a data table with four fields, including update time, check time, registration time, and card information. Specifically, the user can trigger the data table creation operation, and the terminal will display the data table creation interface. In this data table creation interface, the user can edit the properties of each field in the data table, such as the field's serial number, Chinese name, English name, concept, whether it is auto-increment, field security classification, etc. Figure 5 As shown in the table creation interface, the user creates a table that includes four fields: update time, check time, registration time, and card information, and edits the Chinese and English names of each field. Figure 6 As shown, after the user edits the Chinese name and English name of the field, the recommended field classification corresponding to the field can be displayed in the field security classification item corresponding to the field in the data table creation interface. For example, for the update time field, the recommended field classifications include behavior occurrence channel, behavior occurrence time, behavior record, behavior label, etc. The user can directly select the target classification from the recommended field classification to edit the field security classification, or the user can continue to set it by himself without adopting the recommended field classification. Furthermore, post-verification can be performed on fields that do not adopt the recommended field classification to ensure the accuracy of the security classification of each field. In addition, post-verification processing can also be performed on each field in the incremental data table to ensure the accuracy of the security classification of each field in the incremental data table.

[0148] For the hierarchical processing of stock data tables, field classification can be achieved through a combination of models and strategies. Figure 7As shown, a target field to be classified is identified from an existing data table. Regular expression matching and verification are performed on the field data of the target field to obtain a first preliminary classification result for the target field. The field description and field name of the target field are input into a short text classification model to obtain a second preliminary classification result for the target field. The field description and field data of the target field are input into a pre-trained neural network classification model to obtain a third preliminary classification result for the target field. The first, second, and third preliminary classification results are integrated to obtain a preliminary classification result for the target field. For example, the one with the highest confidence among the first, second, and third preliminary classification results can be used as the preliminary classification result for the target field. The preliminary classification result is corrected based on field lineage relationships to obtain a field classification result for the target field. Based on this field classification result, the level of the target field is determined, thereby achieving hierarchical processing of the data corresponding to the target field and the existing data table.

[0149] In the process of performing regular matching on the field data of the target field and performing verification processing to obtain the first preliminary classification result of the target field, a regular expression can be pre-constructed, as shown in Table 2 below.

[0150] Table 2

[0151]

[0152] The field data of the target field is matched with each regular expression in the recognition rule. If the match is successful, it can be determined that the target field belongs to the field corresponding to the successfully matched regular expression, thereby realizing classification processing of the target field based on the field data of the target field. However, when the content of the target field is used, that is, when the field data is classified by regular expressions, the accuracy is high in scenarios where the field data has obvious characteristics, but it cannot be migrated to more scenarios. For example, if the field data is the number 3, it is impossible to determine whether it is the payment amount, age, number of payment orders, or other information based on the field data alone. That is, the applicable scenarios for field classification processing through regular matching are limited.

[0153] The field description and field name of the target field are input into a short text classification model to obtain a second preliminary classification result for the target field. The text length of the field description and field name is limited, with a high proportion of fields being less than 20 characters long. The short text classification model can effectively classify the target field based on the field description and field name. In a specific application, the target field can be classified using the FastText text classification model based on the field description and field name. Specifically, the field description and field name can be used as input to the FastText text classification model, which converts the input field description and field name into word vectors. After averaging, a linear classifier performs linear classification on the averaged word vectors, and a classification result for the target field is obtained based on the linear classification results. Furthermore, the output of a statistical language-based n-gram model and the output of h-softmax (hierarchical softmax) can be used as additional features for the target field, and combined with the word vectors to classify the target field, further improving the accuracy of the target field classification. The FastText text classification model has low model complexity and fast training speed, ensuring efficient data classification.

[0154] In specific implementation, the input corresponding to the FastText text classification model can be determined based on the analysis of the results of the classification processing of historical fields. For example, three groups of input features can be constructed, namely field description + field name + field data, field description + field name, and field description. The three groups of input features corresponding to the historical fields are respectively input into the FastText text classification model for classification, and based on the classification results, it is determined which of the three groups of input features the FastText text classification model corresponds to. In specific applications, for the FastText text classification model, when the input features are composed of field description + field name, the accuracy of the corresponding classification is the highest, so it is determined that the input features corresponding to the FastText text classification model are the field description + field name corresponding to the field. Generally speaking, for data related to enterprises and individuals, the accuracy of field classification through the FastText text classification model is high, while the recognition accuracy of transaction data is low, that is, the FastText text classification model is suitable for classifying fields corresponding to enterprise and some personal data. Based on this, after pre-determining the rough type of the target field, such as enterprise and personal data fields or transaction data fields, determine whether to use the FastText text classification model to classify the target field according to the rough type of the target field, or set the classification weight of the FastText text classification model to ensure the accuracy of field classification.

[0155] In addition, the target field can be classified using the TF-IDF+XGB text classification model. Specifically, when determining the input corresponding to the TF-IDF+XGB text classification model, three sets of input features can be constructed: field description + field name + field data, field description + field name, and field description. These three sets of input features corresponding to the historical field are respectively input into the TF-IDF+XGB text classification model for classification. Based on the classification results, the corresponding set of input features for the TF-IDF+XGB text classification model is determined. In specific applications, the TF-IDF+XGB text classification model achieves the highest classification accuracy when the input feature is the field description. Therefore, the corresponding input feature for the TF-IDF+XGB text classification model is determined to be the field description corresponding to the field. When classifying the target field, the field description of the target field is input into the TF-IDF+XGB text classification model. The TF-IDF+XGB text classification model then performs word segmentation on the field description and calculates the TF-IDF matrix. The target field is then classified based on the XGB and TF-IDF matrices.

[0156] In the process of inputting the target field's field description and field data into a pre-trained neural network classification model to obtain the third preliminary classification result for the target field, the neural network classification model can be pre-trained with various models as needed, such as LSTM models, BiLSTM models, GRU models, and TextCNN models. In a specific application, the accuracy of the LSTM model, BiLSTM model, GRU model, and TextCNN model for classifying the target field based solely on the field description is shown in Table 3 below.

[0157] Table 3

[0158] Model LSTM BiLSTM GRU TextCNN Accuracy 0.7799 0.7442 0.772 0.7649

[0159] When classifying fields based on their field descriptions, the LSTM model has a higher accuracy rate. For different types of data, the GRU model has a higher accuracy in classifying fields corresponding to general data and transaction-type data, while the LSTM model has a higher accuracy in classifying fields corresponding to enterprise and personal data. In specific implementations, a corresponding neural network classification model can be determined based on the type of data involved in the business system to which the data table belongs to classify the target fields in the data table. Alternatively, a corresponding neural network classification model can be comprehensively determined based on whether the field data or attribute information is short text or long text, the model's complexity, the ability to mitigate long-distance dependencies, and the perspective of parallelism. Alternatively, the target fields in the data table can be directly classified using various neural network classification models, and a third preliminary classification result can be obtained based on the classification results of the various neural network classification models.

[0160] Furthermore, in the process of correcting the classification result of a field by field lineage, the source field of the target field can be determined according to the source of the target field, and the preliminary classification result can be corrected according to the field classification result of the source field to determine the field classification result of the target field. Specifically, Figure 8 As shown, the f_age field of table_c is obtained by calculating the maximum value of the f_age fields of tables table_a and table_b, and the classification results of the f_age fields of tables table_a and table_b are both "age". Based on the blood relationship of this field, it can be determined that the classification result corresponding to the f_age field of table_c should also be "age". If the classification result of the f_age field of table_c is not "age" after classification, the classification result of this field can be corrected, thereby ensuring the accuracy of the field classification based on the field blood relationship.

[0161] After obtaining the field classification results for a target field in a data table, the data table or the data in the data table can be graded based on the field classification results. For example, when only data grading is required, the level corresponding to the target field can be determined based on the field classification results for the target field in the data table, thereby determining the data level of the data corresponding to the target field based on the level corresponding to the target field. When the level of the entire data table needs to be determined, that is, when the data table needs to be graded, the field classification results for the target field can be mapped to the level corresponding to the data table. Specifically, in the context of a financial business system, the mapping relationship from field classification to table grading can be implemented in conjunction with the requirements of the grading specifications preset in the business system, thereby being implemented in the grading policy management platform. If the grading specifications preset in the business system change, the mapping relationship from field classification to table grading can be directly modified, thereby achieving dynamic expansion of the data table grading process without having to re-establish the model to categorize each field in the data table. Furthermore, after the grading process for the data table is completed, the grading results can be further confirmed by the data table grading node to ensure the accuracy of the data table grading results.

[0162] In a specific application, such as Figure 9 As shown, when adding a classification to a data table, the classification fields include "Date of Birth," "Precise Address Information," and "User Identification Code." The reference standard is the "Guidelines for Financial Data Security Classification," with an internal classification of 3 and a recommended classification of 3. The classification status is valid. After confirming the added classification, the corresponding classification levels can be set for the three fields in the data table, allowing them to be protected using the corresponding data protection methods for the classification levels, such as encrypting the data in the three fields using the corresponding encryption methods. Furthermore, the data table can also be classified as Level 3 and encrypted using the data protection methods corresponding to Level 3.

[0163] The data classification method in this embodiment, combined with business scenario characteristics and regulatory requirements, systematically and efficiently solves the efficiency and accuracy problems of data security classification and grading; moreover, the overall data classification solution is highly robust and scalable, among which the classification of data corresponding fields is a relatively stable part. By adopting a differentiation strategy based on the characteristics of the classification fields, high accuracy can be pursued, and the results of table classification can be iterated in accordance with national policies and industry specifications, forming a strategic configuration capability, facilitating iteration, and helping to improve the efficiency of data classification processing.

[0164] The present application also provides an application scenario, which applies the above-mentioned data classification method.

[0165] Specifically, the application of the data classification method in this application scenario is as follows:

[0166] For business data tables of enterprise business customers, based on the attribute information and field data of the target fields in the business data tables, more than one field characteristic factor corresponding to the target fields is determined. The field characteristic factors are combined according to the characteristic combination conditions corresponding to at least two classification methods to obtain the field characteristics corresponding to each classification method. The target fields are classified according to the field characteristics corresponding to each classification method, and preliminary classification results corresponding to each classification method are obtained. The preliminary classification results are then merged to obtain the field classification results for the target fields. The data level of the business data corresponding to the target fields in the business data tables is determined based on the field classification results of the target fields, and the business data is encrypted according to the data level of the business data. For example, in the business data table, the level of business data corresponding to field 1 is 1, the level of business data corresponding to field 2 is 3, and the level of business data corresponding to field 3 is 2. The business data corresponding to field 1 can be encrypted according to the data encryption method corresponding to level 1, the business data corresponding to field 2 can be encrypted according to the data encryption method corresponding to level 3, and the business data corresponding to field 3 can be encrypted according to the data encryption method corresponding to level 2. In this way, after the data is classified, data security is ensured according to the corresponding data levels.

[0167] It should be understood that, although the various steps in the flowcharts involved in the various embodiments described above are displayed in sequence according to the instructions of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments described above can include multiple steps or multiple stages, and these steps or stages are not necessarily executed and completed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0168] In one embodiment, Figure 10 As shown, a data classification device 1000 is provided. The device can be a software module or a hardware module, or a combination of the two to form a part of a computer device. The device specifically includes: a characteristic factor determination module 1002, a characteristic factor combination module 1004, a field preliminary classification module 1006, a classification result fusion module 1008 and a data level determination module 1010, wherein:

[0169] A characteristic factor determination module 1002 is configured to determine more than one field characteristic factor corresponding to a target field based on attribute information and field data of the target field in the business system;

[0170] The feature factor combination module 1004 is used to combine the field feature factors according to the feature combination conditions corresponding to at least two classification methods to obtain the field features corresponding to each classification method;

[0171] The field preliminary classification module 1006 is used to classify the target field according to each classification method and the field characteristics corresponding to the corresponding classification method, and obtain the preliminary classification results corresponding to each classification method;

[0172] The classification result fusion module 1008 is used to fuse the preliminary classification results to obtain the field classification result of the target field;

[0173] The data level determination module 1010 is used to obtain the data level of the data corresponding to the target field based on the field classification result mapping.

[0174] In one embodiment, the feature factor combination module 1004 includes a classification method determination module, a target feature factor determination module and a field feature acquisition module; wherein: the classification method determination module is used to determine at least two classification methods; the target feature factor determination module is used to determine the target feature factor corresponding to each classification method from the field feature factors according to the feature combination conditions corresponding to the at least two classification methods; the field feature acquisition module is used to obtain the field feature corresponding to each classification method based on the target feature factor corresponding to each classification method.

[0175] In one embodiment, the classification method includes a field data matching method, and the target feature factor includes a field data factor; the field feature acquisition module is also used to determine that the field data factor is the first field feature corresponding to the field data matching method; the field preliminary classification module 1006 is also used to classify the target field based on the first field feature through the field data matching method to obtain a preliminary classification result corresponding to the field data matching method.

[0176] In one embodiment, the field preliminary classification module 1006 is also used to obtain the field data matching conditions corresponding to the field data matching method; perform regular matching on the first field feature and the field data matching conditions to obtain a regular matching result; when the regular matching result passes the matching verification, obtain the preliminary classification result corresponding to the field data matching method based on the regular matching result.

[0177] In one embodiment, the classification method includes a field text classification method, and the target feature factors include a field description factor and a field name factor; the field feature acquisition module is also used to combine the field description factor and the field name factor to obtain a second field feature corresponding to the field text classification method; the field preliminary classification module 1006 is also used to classify the target field based on the second field feature through a text classification model corresponding to the field text classification method, and obtain a preliminary classification result corresponding to the field text classification method.

[0178] In one embodiment, the field preliminary classification module 1006 is also used to perform vector conversion on the second field feature through a text classification model corresponding to the field text classification method to obtain a word vector; and to perform linear classification on the target field based on the word vector through the text classification model to obtain a preliminary classification result corresponding to the field text classification method.

[0179] In one embodiment, the classification method includes a neural network model classification method, and the target feature factor includes a field data factor and a field description factor; the field feature acquisition module is also used to combine the field data factor and the field description factor to obtain a third field feature corresponding to the neural network model classification method; the field preliminary classification module 1006 is also used to classify the target field based on the third field feature through the field classification network model corresponding to the neural network model classification method, and obtain a preliminary classification result corresponding to the neural network model classification method.

[0180] In one embodiment, it also includes a source field determination module and a field classification correction module; wherein: the source field determination module is used to determine the source field corresponding to the target field according to the field lineage relationship of the target field; the field classification correction module is used to classify and correct the field classification result based on the source field classification result corresponding to the source field, and obtain the field classification result after classification correction; the data level determination module 1010 is also used to map the data level of the data corresponding to the target field based on the field classification result after classification correction.

[0181] In one embodiment, the data level determination module 1010 includes a level mapping relationship acquisition module, a level mapping module and a data grading module; wherein: the level mapping relationship acquisition module is used to query the field level mapping relationship in the business system; the level mapping module is used to perform level mapping based on the field level mapping relationship and the field classification results to obtain the field level corresponding to the target field; the data grading module is used to determine the data level of the data corresponding to the target field according to the field level.

[0182] In one embodiment, it also includes a field creation request response module, a recommended classification result acquisition module and a recommended classification result sending module; wherein: the field creation request response module is used to send a field editing template to the terminal when receiving a field creation request sent by the terminal; the recommended classification result acquisition module is used to classify the field according to the attribute information of the field to be created when the attribute information of the field to be created is obtained through the field editing template, and obtain the recommended classification result corresponding to the field to be created; the recommended classification result sending module is used to send the recommended classification result to the terminal; the recommended classification result is used to instruct the terminal to determine the field category of the field to be created based on the recommended classification result; the field category of the field to be created is used to determine the data level of the data corresponding to the field to be created.

[0183] In one embodiment, an exception prompt module is also included, which is used to send field classification exception information for the field to be created to the terminal when the field to be created is used as the target field and the field classification result determined according to the target field does not match the field category of the field to be created.

[0184] In one embodiment, the characteristic factor determination module 1002 includes a target field determination module, a field information acquisition module and a field factor determination module; wherein: the target field determination module is used to determine the target field in the business system; the field information acquisition module is used to determine the target field in the business system; the field factor determination module is used to obtain more than one field characteristic factors corresponding to the target field based on the field description, field name and field data.

[0185] The specific definition of the data classification device can be found in the definition of the data classification method above and will not be repeated here. Each module in the above-mentioned data classification device can be implemented in whole or in part through software, hardware, or a combination thereof. Each of the above-mentioned modules can be embedded in or independent of the processor of the computer device in hardware form, or can be stored in the memory of the computer device in software form, so that the processor can call and execute the corresponding operations of each of the above modules.

[0186] 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 11As shown. The computer device includes a processor, a memory, and a network interface connected via a system bus. 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, 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 database of the computer device is used to store business data in the business system. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, a data classification method is implemented.

[0187] Those skilled in the art will understand that Figure 11 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0188] In one embodiment, a computer device is further provided, including a memory and a processor. The memory stores a computer program, and the processor implements the steps in the above method embodiments when executing the computer program.

[0189] In one embodiment, a computer-readable storage medium is provided, storing a computer program, which implements the steps in the above-mentioned method embodiments when executed by a processor.

[0190] In one embodiment, a computer program product or computer program is provided, the computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the steps of each of the above-described method embodiments.

[0191] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of relevant countries and regions.

[0192] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiment methods can be implemented 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, database or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory may include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, distributed databases based on blockchains. The processor involved in the various embodiments provided herein may be, but are not limited to, a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic unit, a data processing logic unit based on quantum computing, and the like.

[0193] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0194] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art could make various modifications and improvements without departing from the spirit of the present application, all of which fall within the scope of protection of the present application. Therefore, the scope of protection of the present patent application shall be determined by the appended claims.

Claims

1. A data classification method, characterized in that: The method comprises: Determining, based on attribute information and field data of a target field in a business system, more than one field characteristic factor corresponding to the target field, wherein the attribute information is used to describe the target field itself, and the field data includes data generated in the business system and belonging to the target field; Combining target feature factors in the field feature factors according to feature combination conditions corresponding to at least two classification methods to obtain field features corresponding to each classification method, wherein the target feature factor of each classification method is determined from the field feature factors based on the feature combination conditions corresponding to each classification method; According to each of the classification methods, the target field is classified according to the field characteristics corresponding to the corresponding classification method, and a preliminary classification result corresponding to each of the classification methods is obtained; Perform fusion based on the preliminary classification results to obtain a field classification result of the target field; The data level of the data corresponding to the target field is obtained based on the field classification result mapping.

2. The method according to claim 1, characterized in that The target feature factors in the field feature factors are combined according to the feature combination conditions corresponding to the at least two classification methods to obtain the field features corresponding to each classification method, including: Identify at least two ways to classify; Determining a target characteristic factor corresponding to each classification method from the field characteristic factors according to the characteristic combination conditions corresponding to each of the at least two classification methods; Based on the target characteristic factors corresponding to each of the classification methods, the field characteristics corresponding to each of the classification methods are obtained.

3. The method according to claim 2, characterized in that The classification method includes a field data matching method, and the target feature factor includes a field data factor; the field feature corresponding to each classification method is obtained based on the target feature factor corresponding to each classification method, including: Determining the field data factor as a first field feature corresponding to the field data matching method; The target field is classified according to each classification method and the field characteristics corresponding to the corresponding classification method to obtain the preliminary classification results corresponding to each classification method, including: The target field is classified based on the first field feature through the field data matching method to obtain a preliminary classification result corresponding to the field data matching method.

4. The method according to claim 3, characterized in that The classifying the target field based on the first field feature by using the field data matching method to obtain a preliminary classification result corresponding to the field data matching method includes: Obtaining the field data matching condition corresponding to the field data matching method; Performing regular expression matching on the first field feature and the field data matching condition to obtain a regular expression matching result; When the regular matching result passes the matching verification, a preliminary classification result corresponding to the field data matching mode is obtained according to the regular matching result.

5. The method according to claim 2, characterized in that The classification method includes a field text classification method, and the target feature factor includes a field description factor and a field name factor. The field feature corresponding to each classification method is obtained based on the target feature factor corresponding to each classification method, including: Combining the field description factor and the field name factor to obtain a second field feature corresponding to the field text classification method; The target field is classified according to each classification method and the field characteristics corresponding to the corresponding classification method to obtain the preliminary classification results corresponding to each classification method, including: The target field is classified based on the second field feature using a text classification model corresponding to the field text classification method to obtain a preliminary classification result corresponding to the field text classification method.

6. The method according to claim 5, characterized in that The text classification model corresponding to the field text classification method is used to classify the target field based on the second field feature to obtain a preliminary classification result corresponding to the field text classification method, including: Performing vector conversion on the second field feature using a text classification model corresponding to the field text classification method to obtain a word vector; The target field is linearly classified based on the word vector through the text classification model to obtain a preliminary classification result corresponding to the field text classification method.

7. The method according to claim 2, characterized in that The classification method includes a neural network model classification method, and the target feature factor includes a field data factor and a field description factor; the field feature corresponding to each classification method is obtained based on the target feature factor corresponding to each classification method, including: Combining the field data factor and the field description factor to obtain a third field feature corresponding to the neural network model classification method; The target field is classified according to each classification method and the field characteristics corresponding to the corresponding classification method to obtain the preliminary classification results corresponding to each classification method, including: The target field is classified based on the third field feature through the field classification network model corresponding to the neural network model classification method to obtain a preliminary classification result corresponding to the neural network model classification method.

8. The method according to claim 1, characterized in that The method further comprises: Determine the source field corresponding to the target field according to the field lineage relationship of the target field; Based on the source field classification result corresponding to the source field, the field classification result is classified and corrected to obtain a field classification result after classification correction; The obtaining of the data level of the data corresponding to the target field based on the field classification result mapping includes: Based on the field classification result after the classification correction, the data level of the data corresponding to the target field is mapped.

9. The method according to claim 1, characterized in that The obtaining of the data level of the data corresponding to the target field based on the field classification result mapping includes: Query the field level mapping relationship in the business system; Performing level mapping based on the field level mapping relationship and the field classification result to obtain the field level corresponding to the target field; The data level of the data corresponding to the target field is determined according to the field level.

10. The method according to claim 1, characterized in that The method further comprises: When receiving a field creation request sent by a terminal, sending a field editing template to the terminal; When the attribute information of the field to be created is obtained through the field editing template, the field is classified according to the attribute information of the field to be created to obtain a recommended classification result corresponding to the field to be created; The recommended classification result is sent to the terminal; the recommended classification result is used to instruct the terminal to determine the field category of the field to be created based on the recommended classification result; the field category of the field to be created is used to determine the data level of the data corresponding to the field to be created.

11. The method according to claim 10, characterized in that The method further comprises: When the field to be created is used as the target field and the field classification result determined according to the target field does not match the field category of the field to be created, field classification exception information for the field to be created is sent to the terminal.

12. The method according to any one of claims 1 to 11, characterized in that The determining of more than one field characteristic factors corresponding to the target field based on the attribute information and field data of the target field in the business system includes: Determine the target fields in the business system; Obtain the field description, field name, and field data of the target field; More than one field characteristic factor corresponding to the target field is obtained according to the field description, the field name and the field data.

13. A data classification device, characterized in that: The device comprises: a characteristic factor determination module, configured to determine one or more field characteristic factors corresponding to a target field based on attribute information and field data of the target field in the business system, wherein the attribute information is used to describe the target field itself, and the field data includes data generated in the business system and belonging to the target field; a feature factor combination module, configured to combine target feature factors in the field feature factors according to feature combination conditions corresponding to at least two classification methods, to obtain field features corresponding to each classification method, wherein the target feature factor of each classification method is determined from the field feature factors based on the feature combination conditions corresponding to each classification method; A field preliminary classification module is used to classify the target field according to each classification method and the field characteristics corresponding to the corresponding classification method, and obtain preliminary classification results corresponding to each classification method; A classification result fusion module is used to fuse the preliminary classification results to obtain the field classification result of the target field; A data level determination module is used to obtain the data level of the data corresponding to the target field based on the field classification result mapping.

14. The device according to claim 13, characterized in that The feature factor combination module is also used to determine at least two classification methods; according to the feature combination conditions corresponding to the at least two classification methods, determine the target feature factor corresponding to each classification method from the field feature factors; based on the target feature factor corresponding to each classification method, obtain the field feature corresponding to each classification method.

15. The device according to claim 14, characterized in that The classification method includes a field data matching method, and the target characteristic factor includes a field data factor; The feature factor combination module is further configured to determine that the field data factor is the first field feature corresponding to the field data matching method; The field preliminary classification module is further configured to classify the target field based on the first field feature through the field data matching method to obtain a preliminary classification result corresponding to the field data matching method.

16. The device according to claim 15, characterized in that The field preliminary classification module is also used to obtain the field data matching conditions corresponding to the field data matching method; perform regular matching on the first field feature and the field data matching conditions to obtain a regular matching result; when the regular matching result passes the matching verification, obtain the preliminary classification result corresponding to the field data matching method based on the regular matching result.

17. The device according to claim 14, characterized in that The classification method includes a field text classification method, and the target feature factors include a field description factor and a field name factor; The feature factor combination module is further configured to combine the field description factor and the field name factor to obtain a second field feature corresponding to the field text classification method; The field preliminary classification module is further used to classify the target field based on the second field feature using a text classification model corresponding to the field text classification method, and obtain a preliminary classification result corresponding to the field text classification method.

18. The device according to claim 17, characterized in that The field preliminary classification module is also used to vectorize the second field features through the text classification model corresponding to the field text classification method to obtain a word vector; through the text classification model, linearly classify the target field based on the word vector to obtain a preliminary classification result corresponding to the field text classification method.

19. The device according to claim 14, characterized in that The classification method includes a neural network model classification method, and the target feature factors include a field data factor and a field description factor; The feature factor combination module is further used to combine the field data factor and the field description factor to obtain a third field feature corresponding to the neural network model classification method; The field preliminary classification module is also used to classify the target field based on the third field feature through the field classification network model corresponding to the neural network model classification method, and obtain a preliminary classification result corresponding to the neural network model classification method.

20. The device according to claim 13, wherein The device further comprises: A source field determination module, configured to determine a source field corresponding to the target field based on the field lineage relationship of the target field; A field classification correction module is used to perform classification correction on the field classification result based on the source field classification result corresponding to the source field, to obtain a field classification result after classification correction; The data level determination module is further configured to map and obtain the data level of the data corresponding to the target field based on the field classification result after the classification correction.

21. The device according to claim 13, characterized in that The data level determination module is also used to query the field level mapping relationship in the business system; perform level mapping based on the field level mapping relationship and the field classification result to obtain the field level corresponding to the target field; and determine the data level of the data corresponding to the target field according to the field level.

22. The device according to claim 13, characterized in that The device further comprises: A field creation request response module is used to send a field editing template to the terminal when receiving a field creation request sent by the terminal; A recommended classification result obtaining module is used to, when obtaining attribute information of a field to be created through the field editing template, perform field classification according to the attribute information of the field to be created, and obtain a recommended classification result corresponding to the field to be created; The recommended classification result sending module is used to send the recommended classification result to the terminal; the recommended classification result is used to instruct the terminal to determine the field category of the field to be created based on the recommended classification result; the field category of the field to be created is used to determine the data level of the data corresponding to the field to be created.

23. The device according to claim 22, characterized in that The device further comprises: The exception prompt module is used to send field classification exception information for the field to be created to the terminal when the field to be created is used as the target field and the field classification result determined according to the target field does not match the field category of the field to be created.

24. The device according to any one of claims 13 to 23, characterized in that The characteristic factor determination module is also used to determine the target field in the business system; obtain the field description, field name and field data of the target field; and obtain more than one field characteristic factors corresponding to the target field based on the field description, the field name and the field data.

25. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 12 are implemented.

26. 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 method according to any one of claims 1 to 12 are implemented.

27. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 12 are implemented.

Images