Financial data classification and grading method and device, program product and electronic equipment

Abstract

The invention discloses a financial data classification and grading method and device, a program product and electronic equipment, and relates to the field of financial science and technology or other related technical fields, and the method comprises the steps: carrying out the feature extraction of financial data to be classified and graded, and obtaining data features corresponding to the financial data; based on the data features corresponding to the financial data, rule matching is carried out in a preset knowledge base, preset rules matched with the financial data are obtained, and the preset rules in the preset knowledge base are obtained based on a preset safety specification file and historical record induction for classifying and grading historical financial data; and based on a preset rule matched with the financial data, classifying and grading the financial data from the dimensions of the data type, the data sensitivity and the business type to obtain a classification and grading result of the financial data. The technical problem that the effect of classifying and grading financial data based on the prior art is poor is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of financial technology or other related technical fields, in particular, to a financial data classification and grading method, device, program product and electronic equipment. BACKGROUND

[0002] With the globalization of financial services of financial institutions, the data assets of financial institutions show the characteristics of diversification and complexity, and financial institutions are facing increasingly severe data security management challenges. At present, when classifying and grading financial data, the financial industry mainly relies on manual auditing of data based on pre-prepared static classification rules. Although this method ensures the accuracy of data classification to a certain extent, with the explosive growth of the amount of data to be processed and the diversification of data language, this artificial intensive classification method not only is inefficient, but also introduces artificial detection errors, thereby causing the technical problem of poor effect of classifying and grading financial data based on the prior art.

[0003] For the above problems, no effective solution has been proposed so far. SUMMARY

[0004] The present application provides a financial data classification and grading method, device, program product and electronic equipment to at least solve the technical problem of poor effect of classifying and grading financial data based on the prior art.

[0005] According to one aspect of the present application, a financial data classification and grading method is provided, comprising: performing feature extraction on financial data to be classified and graded to obtain data features corresponding to the financial data, wherein the data features are used to represent semantic information, word frequency and context information of keywords in the financial data; performing rule matching in a preset knowledge base based on the data features corresponding to the financial data to obtain preset rules matched by the financial data, wherein the preset rules in the preset knowledge base are induced based on a preset security specification file and historical records of classifying and grading historical financial data; and classifying and grading the financial data from the dimensions of data type, data sensitivity and business type based on the preset rules matched by the financial data to obtain classification and grading results of the financial data.

[0006] Optionally, before feature extraction is performed on the financial data to be classified, the classification grading method of the financial data further comprises: performing data cleaning on original data in a preset database of the financial institution to obtain first data, wherein the original data is text data described in L preset languages, and L is a positive integer; dividing the first data based on the L preset languages to obtain L data segments, wherein the text data in each data segment is described in the same preset language; transmitting each data segment in the L data segments to a target translator corresponding to the data segment; translating, by the target translator, the preset language used to describe each data segment into a standard language to obtain L target segments corresponding to the L data segments, wherein the standard language is one of the L preset languages; and determining the financial data based on the L target segments.

[0007] Optionally, based on the data features corresponding to the financial data, rule matching is performed in the preset knowledge base to obtain a preset rule matched by the financial data, comprising: performing feature extraction on each preset rule in the preset knowledge base to obtain rule features corresponding to each preset rule; obtaining a correlation coefficient between the data features and the rule features corresponding to each preset rule; and in a case where the correlation coefficient is greater than or equal to a preset coefficient, taking the preset rule corresponding to the correlation coefficient as the preset rule matched by the financial data.

[0008] Optionally, after the rule matching is performed in the preset knowledge base based on the data features corresponding to the financial data, the classification grading method of the financial data further comprises: in a case where the preset rule matched by the financial data is not queried, creating a new rule based on the data features of the financial data by a target model, wherein the target model is a large language model pre-trained based on historical financial data and historical classification grading rules of P financial institutions, and P is a positive integer; performing a first test operation and a second test operation on the new rule, wherein the first test operation is used for conflict detection of the new rule and each preset rule in the preset knowledge base, and the second test operation is used for performance testing of the new rule; in a case where the first test operation and the second test operation are both passed, determining the classification grading result of the financial data based on the new rule, and storing the new rule as a new preset rule in the preset knowledge base.

[0009] Optionally, the training step of the target model further comprises: performing an initialization operation by the central server to obtain a global initial model, and transmitting model parameters of the global initial model to the P financial institutions, wherein the P financial institutions are participants of federated learning; controlling each of the P financial institutions to create a local initial model based on the model parameters of the global initial model; iteratively training the local initial model corresponding to each financial institution based on historical financial data and historical classification and grading rules of the financial institution to obtain a local model; encrypting and transmitting the gradient of the local model of each of the P financial institutions to the central server, and obtaining a target gradient by the central server by performing weighted summation on the gradient corresponding to each of the P financial institutions; iteratively updating the global initial model based on the target gradient to obtain a global federated model, and transmitting model parameters of the global federated model to the P financial institutions; and controlling each of the P financial institutions to update the local model of each financial institution based on the model parameters of the global federated model to obtain a target model of each financial institution.

[0010] Optionally, iteratively training the local initial model corresponding to each financial institution based on historical financial data and historical classification and grading rules of the financial institution to obtain a local model comprises: dividing the historical financial data and the historical classification and grading rules of each financial institution into a training set and a test set corresponding to each financial institution; iteratively training the local initial model of each financial institution based on the historical financial data and the historical classification and grading rules in the training set to obtain a training model corresponding to each financial institution; testing the training model corresponding to each financial institution based on the historical financial data and the historical classification and grading rules in the test set to obtain a model performance indicator of the training model corresponding to each financial institution, wherein the model performance indicator is at least one of: classification coverage, backflow rate, and data throughput; and in a case where the model performance indicator of the training model corresponding to each financial institution is within a preset performance indicator interval, taking the training model corresponding to each financial institution as the local model corresponding to each financial institution.

[0011] Optionally, in the process of classifying and grading the financial data from the dimensions of data type, data sensitivity, and business type based on the preset rules of financial data matching, the classification and grading method of the financial data further comprises: detecting a classification and grading execution duration corresponding to the financial data; and in a case where the classification and grading execution duration is greater than or equal to a preset duration, generating a warning information.

[0012] According to another aspect of this application, a financial data classification and grading device is also provided, comprising: a data feature extraction unit, used to extract features from the financial data to be classified and graded to obtain data features corresponding to the financial data, wherein the data features are used to characterize the semantic information, word frequency, and contextual information of keywords in the financial data; a rule matching unit, used to perform rule matching in a preset knowledge base based on the data features corresponding to the financial data to obtain preset rules for matching the financial data, wherein the preset rules in the preset knowledge base are obtained based on a preset security specification document and historical records of classifying and grading historical financial data; and a classification and grading unit, used to classify and grade the financial data from the dimensions of data type, data sensitivity, and business type based on the preset rules for matching the financial data to obtain the classification and grading results of the financial data.

[0013] According to another aspect of this application, a computer program product is also provided, which stores a computer program, wherein, when the computer program is running, it controls the computer program product to execute the classification and grading method of financial data described above.

[0014] According to another aspect of this application, an electronic device is also provided, wherein the electronic device includes one or more processors and a memory for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors cause the one or more processors to implement the financial data classification and grading method described above.

[0015] In this application, feature extraction is first performed on the financial data to be classified and graded to obtain the corresponding data features. These data features are used to characterize the semantic information, word frequency, and contextual information of keywords in the financial data. Then, based on the data features corresponding to the financial data, rule matching is performed in a preset knowledge base to obtain preset rules for matching the financial data. The preset rules in the preset knowledge base are derived from preset security specification documents and historical records of classifying and grading historical financial data. Subsequently, based on the preset rules for matching the financial data, the financial data is classified and graded from the dimensions of data type, data sensitivity, and business type to obtain the classification and grading results of the financial data.

[0016] As described above, this application obtains data features, including semantic information, word frequency, and contextual information of keywords in financial data, through in-depth analysis of the data characteristics of financial data. This achieves the goal of accurately and quickly matching preset rules in a preset knowledge base based on data features. Subsequently, this application classifies and grades financial data according to data type, sensitivity, and business type based on the matched preset rules, ensuring the security and compliance of financial data management. This achieves the technical effect of efficient and accurate classification and grading of financial data, thereby solving the technical problem of poor performance in classifying and grading financial data based on existing technologies. Attached Figure Description

[0017] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0018] Figure 1 This is a hardware structure block diagram of an optional computer terminal (or mobile device) for implementing a classification and grading method for financial data, according to an embodiment of this application.

[0019] Figure 2 This is a flowchart of an optional method for classifying and grading financial data according to an embodiment of this application;

[0020] Figure 3 This is a flowchart of an optional data asset hierarchical classification method according to an embodiment of this application;

[0021] Figure 4 This is a schematic diagram of an optional financial data classification and grading device according to an embodiment of this application;

[0022] Figure 5 This is a structural block diagram of an electronic device according to an embodiment of this application. Detailed Implementation

[0023] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0024] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0025] It should also be noted that all relevant information (including but not limited to information used for display and analysis) and data (including but not limited to financial data requiring classification and grading, preset rule data in the preset knowledge base, and style data used for model training) involved in this application are information and data authorized by the user or fully authorized by all parties. For example, this system has an interface with the relevant user or institution. Before obtaining relevant information, it needs to send an acquisition request to the aforementioned user or institution through the interface, and obtain the relevant information after receiving consent from the aforementioned user or institution.

[0026] Furthermore, the collection, storage, use, processing, transmission, provision, disclosure, and application of relevant information and data involved in this application all comply with the relevant laws, regulations, and standards of the relevant regions, and necessary confidentiality measures have been taken. This application does not violate public order and good morals. In addition, this application provides a corresponding operation entry point for users to choose to agree to or refuse authorization. If the user chooses to refuse authorization, the corresponding expert decision-making process will be initiated.

[0027] The present invention will now be described in detail with reference to various embodiments.

[0028] Example 1

[0029] According to an embodiment of this application, an embodiment of a method for classifying and grading financial data is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0030] The methods and embodiments provided in this application can be executed on mobile terminals, computer terminals, or similar computing devices. Figure 1This is a hardware structure block diagram of a computer terminal (or mobile device) for implementing a classification and grading method for financial data, according to an embodiment of this application. Figure 1 As shown, the computer terminal 10 (or mobile device) may include one or more processors 102 (shown as 102a, 102b, ..., 102n in the figure) 102 (processor 102 may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA, etc.), a memory 104 for storing data, and a transmission device 106 for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of a BUS bus), a network interface, a power supply, and / or a camera. Those skilled in the art will understand that... Figure 1 The structure shown is for illustrative purposes only and does not limit the structure of the aforementioned electronic device. For example, computer terminal 10 may also include... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown.

[0031] It should be noted that the aforementioned one or more processors 102 and / or other data processing circuits are generally referred to herein as "data processing circuits". These data processing circuits may be embodied, in whole or in part, in software, hardware, firmware, or any other combination thereof. Furthermore, the data processing circuits may be a single, independent processing module, or may be integrated, in whole or in part, into any other element within the computer terminal 10 (or mobile device). As involved in the embodiments of this application, the data processing circuits serve as a processor control mechanism (e.g., selection of a variable resistor termination path connected to an interface).

[0032] The memory 104 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the financial data classification and grading method in this embodiment. The processor 102 executes various functional applications and data processing by running the software programs and modules stored in the memory 104, thereby realizing the aforementioned financial data classification and grading method. The memory 104 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include memory remotely located relative to the processor 102, and these remote memories can be connected to the computer terminal 10 via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0033] The transmission device 106 is used to receive or send data via a network. Specific examples of the network described above may include a wireless network provided by the communication provider of the computer terminal 10. In one example, the transmission device 106 includes a Network Interface Controller (NIC), which can connect to other network devices via a base station to communicate with the Internet. In another example, the transmission device 106 may be a Radio Frequency (RF) module, used for wireless communication with the Internet.

[0034] The display can be configured as a touchscreen liquid crystal display (LCD) that allows the user to interact with the user interface of the computer terminal 10 (or mobile device).

[0035] Under the aforementioned operating environment, this application provides a financial data classification and grading system (hereinafter referred to as the classification and grading system) for implementing the financial data classification and grading method in this application. Figure 2 This is a flowchart of an optional method for classifying and grading financial data according to an embodiment of this application, such as... Figure 2 As shown, the method includes the following steps:

[0036] Step S201: Extract features from the financial data to be classified and graded to obtain the data features corresponding to the financial data. The data features are used to characterize the semantic information, word frequency and contextual information of keywords in the financial data.

[0037] Optionally, data features include semantic information, word frequency, and contextual information of keywords in financial data.

[0038] Optionally, the classification and grading system can capture the deep semantics and keyword characteristics (e.g., word frequency) of financial data through feature extraction, thereby providing comprehensive information support for subsequent rule matching and classification. In particular, the semantic and contextual information of keywords in the data features can help the classification and grading system to more accurately understand the true meaning of financial data and reduce the probability of errors in classification and grading.

[0039] Step S202: Based on the data characteristics corresponding to the financial data, rule matching is performed in a preset knowledge base to obtain preset rules for matching financial data. The preset rules in the preset knowledge base are obtained based on preset security specification documents and historical records that classify and grade historical financial data.

[0040] Optionally, a pre-defined knowledge base is provided, which includes pre-defined rules derived from security specification documents and historical records that classify and categorize historical financial data. These pre-defined rules serve as the basis for decision-making regarding classification and grading operations.

[0041] Optionally, the classification and grading system obtains preset rules for matching financial data by performing rule matching in a preset knowledge base based on the data characteristics corresponding to the financial data, which has the following functions:

[0042] (1) Fast and accurate decision making: The use of preset rules can quickly match data features, reduce processing time, and improve classification efficiency.

[0043] (2) Improve the accuracy of classification and grading results: The preset rules used by the classification and grading system are automatically summarized based on security specification documents and historical records of classification and grading of historical financial data, which improves the accuracy and compliance of classification and grading results.

[0044] Step S203: Based on the preset rules for matching financial data, classify and grade the financial data from the dimensions of data type, data sensitivity, and business type to obtain the classification and grading results of the financial data.

[0045] Optionally, the classification and grading system classifies and grades financial data from the dimensions of data type, data sensitivity, and business type, which can comprehensively cover different analytical dimensions of financial data and ensure the comprehensiveness and accuracy of the classification and grading results. The automated matching of preset rules makes the classification and grading process more intelligent, reduces human intervention, and lowers the risk of operational errors. Grading financial data based on sensitivity and business type enables sensitive information to receive a higher level of protection, thereby improving the security of financial data after classification and grading.

[0046] As described above, this application obtains data features, including semantic information, word frequency, and contextual information of keywords in financial data, through in-depth analysis of the data characteristics of financial data. This achieves the goal of accurately and quickly matching preset rules in a preset knowledge base based on data features. Subsequently, this application classifies and grades financial data according to data type, sensitivity, and business type based on the matched preset rules, ensuring the security and compliance of financial data management. This achieves the technical effect of efficient and accurate classification and grading of financial data, thereby solving the technical problem of poor performance in classifying and grading financial data based on existing technologies.

[0047] In one optional embodiment, before extracting features from the financial data to be classified, the classification and grading system first cleans the raw data in the financial institution's preset database to obtain first data. The raw data consists of text data described using L preset languages, where L is a positive integer. Then, the classification and grading system divides the first data into L data segments based on the L preset languages. Each data segment contains text data described using the same preset language. Next, the classification and grading system transmits each of the L data segments to a corresponding target translator. Subsequently, the classification and grading system uses the target translator to translate the preset language describing each data segment into a standard language, resulting in L target segments corresponding to the L data segments. The standard language is one of the L preset languages. Finally, the classification and grading system determines the financial data based on the L target segments.

[0048] Optionally, the classification and grading system can clean the raw data in the financial institution's pre-set database to remove redundant and duplicate data, thereby improving the overall quality of the data processed by the subsequent classification and grading coefficients.

[0049] Optionally, the classification and grading system divides the first data obtained after data cleaning into data segments according to the text description language (i.e., L preset languages) used in the original data. This ensures that each data segment is described using the same preset language, thereby reducing the difficulty of subsequent standardization processing (i.e. translation processing) of the first data, avoiding the mixed use of languages, and improving the efficiency and accuracy of data translation processing.

[0050] Optionally, the classification and grading system controls the target translator corresponding to each data segment to translate the text data in each data segment, thereby obtaining data segments described in a preset standard language. Finally, the data is summarized to obtain financial data described in a unified standard language, thereby reducing data processing errors caused by language recognition barriers and improving the accuracy and efficiency of subsequent classification and grading of financial data.

[0051] In summary, through the implementation of the above steps, the classification and grading system can efficiently process raw data described in multiple languages ​​by financial institutions, ensuring that the data, after being cleaned, translated, and uniformly processed, can be accurately identified and managed by the classification and grading system. This improves the efficiency and accuracy of data classification and grading, and strengthens the security and compliance of financial data.

[0052] In one optional embodiment, the classification and grading system first extracts features from each preset rule in the preset knowledge base to obtain the rule features corresponding to each preset rule. Then, the classification and grading system obtains the correlation coefficient between the data features and the rule features corresponding to each preset rule. Subsequently, if the correlation coefficient is greater than or equal to the preset coefficient, the classification and grading system uses the preset rule corresponding to the correlation coefficient as the preset rule for financial data matching.

[0053] Optionally, by extracting rule features, the classification and grading system can deepen its understanding of the preset rules, thereby gaining a deeper understanding of the core elements of the rules and providing more accurate feature basis for subsequent rule matching. At the same time, by transforming abstract rules into concrete feature expressions, the classification and grading system can process and apply these preset rules more efficiently, improving the automation level and matching accuracy of rule matching.

[0054] Optionally, the correlation coefficient is used to measure the strength of the association between data features and rule features. The higher the correlation coefficient value, the higher the degree of matching between the two.

[0055] Optionally, the classification and grading system can quantify the matching degree between preset rules and specific data features by calculating the correlation coefficient, thereby achieving more accurate rule matching and reducing the probability of misjudgment and omission. At the same time, the introduction of the correlation coefficient avoids the introduction of human operation errors based on manual methods to determine the preset rules used for classification and grading, thus making the classification and grading decision-making process more accurate.

[0056] Optionally, the classification and grading system pre-sets a correlation coefficient threshold. When the actual calculated correlation coefficient is higher than or equal to the threshold, the financial data is considered to match a certain preset rule, thereby filtering out the matching preset rule to guide the classification and grading of financial data. This step provides an automatic matching mechanism for preset rules, reduces the labor cost in the classification and grading process, and also reduces errors introduced by human operation, thereby realizing the intelligentization of data classification and grading and improving the accuracy of classification and grading decisions.

[0057] In summary, the classification and grading system deepens the understanding of preset rules through feature extraction, achieves accurate diagnostic matching between data and rules by calculating correlation coefficients, and finally ensures the automation and accuracy of the matching process through the setting of preset coefficient conditions. This mechanism enables the classification and grading system to quickly respond to a large amount of diverse financial data, thereby achieving high-precision classification and grading.

[0058] In one optional embodiment, after performing rule matching in a preset knowledge base based on the data features corresponding to the financial data, if no preset rule matching the financial data is found, the classification and grading system creates a new rule based on the data features of the financial data using a target model. The target model is a large language model pre-trained based on historical financial data and historical classification and grading rules from P financial institutions, where P is a positive integer. Then, the classification and grading system performs a first test operation and a second test operation on the new rule. The first test operation detects conflicts between the new rule and each preset rule in the preset knowledge base, and the second test operation performs a performance test on the new rule. Subsequently, if both the first and second test operations pass, the classification and grading system determines the classification and grading result of the financial data based on the new rule and stores the new rule as a new preset rule in the preset knowledge base.

[0059] Optionally, the new rule refers to a new rule automatically generated by the target model based on the data characteristics of financial data, used to fill the gaps in the classification and grading of the preset rules in the preset knowledge base.

[0060] Optionally, if no preset rules matching the financial data are found, the classification and grading system can dynamically create new rules by calling the target model. This can dynamically supplement and optimize the rule content in the preset knowledge base, enabling the preset knowledge base to adapt to constantly changing business needs and the data types of financial data. At the same time, the target model is pre-trained based on the historical financial data and historical classification and grading rules of P financial institutions. It can intelligently generate new rules when the data features cannot match the preset rules, avoiding manual data classification and grading, thereby improving the accuracy of the classification and grading results.

[0061] Optionally, the classification and grading system eliminates potential logical conflicts in data classification and grading by performing conflict detection between the new rules and each preset rule in the preset knowledge base. This ensures that the new rules not only correctly match the data but also do not conflict with existing rules, thereby improving the accuracy and consistency of the rule-based classification and grading results.

[0062] Optionally, by conducting performance tests on the new rules, the classification and grading system avoids the data classification and grading bottlenecks that inefficient rules may introduce, thereby improving the overall operating efficiency of the classification and grading system.

[0063] Optionally, after the classification and grading system determines that the new rule has passed the conflict test and performance test, the classification and grading system will add the new rule to the preset knowledge base and use the new rule as a new preset rule, thereby continuously optimizing the content in the preset knowledge base. Subsequently, through the newly added new rule, the classification and grading system can immediately and accurately classify and grade the unmatched data, improving the timeliness and effectiveness of data processing.

[0064] In summary, through the implementation of the above steps, this technical solution not only overcomes the limitations of existing technologies in the classification and grading of financial data, but also realizes the dynamic generation and optimization of rules, improves the accuracy of classification and grading and system efficiency, and provides a classification and grading basis for the data security management of financial institutions.

[0065] In one optional embodiment, to train the target model, the classification and grading system first performs an initialization operation through a central server to obtain a global initial model. The model parameters of the global initial model are then transmitted to P financial institutions, which are participants in the federated learning process. Next, the classification and grading system controls each of the P financial institutions to create a local initial model based on the model parameters of the global initial model. Subsequently, the classification and grading system iteratively trains the local initial model corresponding to each financial institution based on its historical financial data and historical classification and grading rules to obtain a local model. Then, the classification and grading system encrypts and transmits the gradients of the local models of each of the P financial institutions to the central server. The central server performs a weighted summation of the gradients corresponding to each of the P financial institutions to obtain the target gradient. The classification and grading system then iteratively updates the global initial model based on the target gradient to obtain a global federated model. The model parameters of the global federated model are transmitted to the P financial institutions. Subsequently, the classification and grading system controls each of the P financial institutions to update its local model based on the model parameters of the global federated model, thus obtaining the target model for each financial institution.

[0066] Optionally, the central server refers to the core server in the federated learning system, which is responsible for the initialization of the global model, gradient aggregation, and iterative update control of the local model.

[0067] Optionally, the classification and grading system broadcasts the model parameters of the global initial model obtained from the initialization operation to P financial institutions, ensuring that all participating financial institutions start training from a unified model version. The efficient transmission of model parameters enables the P financial institutions to quickly and synchronously start local model training, thereby enhancing the overall efficiency of the model training process.

[0068] Alternatively, the classification and grading system allows each financial institution to train its model in its own data environment by controlling the creation of a local initial model, thereby obtaining a local model that is better adapted to the characteristics of its own financial data and classification and grading needs.

[0069] Optionally, the classification and grading system iteratively trains the local model of each financial institution, allowing each institution to optimize its local model in a personalized way to better adapt to its specific data classification and grading needs. Since the iterative training is carried out locally by each financial institution, the data of the financial institution does not need to leave its control, thereby ensuring the privacy and security of the local data of each financial institution.

[0070] Optionally, the classification and grading system ensures data security and prevents the leakage of sensitive information by encrypting the gradients used by the local models of P financial institutions when transmitting model training information. Then, the classification and grading system controls the central server to perform gradient aggregation, using a weighted summation method to aggregate gradient information, taking into account the differences in data volume and model training quality among financial institutions, making the update of the global model more scientific and reasonable.

[0071] Optionally, the classification and grading system iteratively updates the global initial model based on the target gradient obtained by gradient aggregation to obtain a global federated model. This allows the iteratively updated global federated model to absorb the training experience of all financial institutions, thereby improving the generalization ability and classification and grading accuracy of the updated federated model. At the same time, the iterative update of the model parameters realizes the continuous optimization of the model, ensuring that the classification and grading rules can adapt to the ever-changing financial data environment.

[0072] As described above, the classification and grading system uses federated learning technology to perform federated training on the historical financial data and historical classification and grading rules of P financial institutions to obtain a global federated model. Then, the model parameters of the global federated model are sent to each financial institution. Each financial institution updates the model parameters of its local model based on the latest global federated model parameters, using the most recently updated local model as the target model to achieve classification and grading of the financial data of each financial institution. This achieves synchronous optimization of the local models corresponding to the P financial institutions. The target model of each financial institution remains synchronously updated after each iteration, realizing the consistency improvement and optimization of the model in federated learning. Furthermore, each financial institution can immediately apply the updated target model to new data classification and grading, improving the timeliness and effectiveness of model application.

[0073] In summary, through the implementation of the above steps, this technical solution utilizes federated learning technology to solve the problems faced by financial institutions in data classification and grading, such as data silos in model training, data privacy protection, and model performance improvement, achieving a dual improvement in model training efficiency and classification and grading accuracy.

[0074] In one optional embodiment, the classification and grading system first divides the historical financial data and historical classification and grading rules of each financial institution into a training set and a test set corresponding to each financial institution. Then, based on the historical financial data and historical classification and grading rules in the training set, the classification and grading system iteratively trains the local initial model of each financial institution to obtain the training model corresponding to each financial institution. Then, based on the historical financial data and historical classification and grading rules in the test set, the classification and grading system tests the training model corresponding to each financial institution to obtain the model performance index of the training model corresponding to each financial institution. The model performance index is at least one of the following: classification coverage, backflow rate, and data throughput. Subsequently, if the model performance index of the training model corresponding to each financial institution is within a preset performance index range, the classification and grading system uses the training model corresponding to each financial institution as the local model corresponding to each financial institution.

[0075] Optionally, the classification and grading system divides the historical financial data and historical classification and grading rules of each financial institution into a training set and a test set corresponding to each financial institution. Then, the classification and grading system trains the model based on the data in the training set and independently tests the trained model based on the data in the separately retained test set. This can evaluate the model's performance on unseen data, thereby improving the generalization ability of the trained model. By separating the training set and the test set, the problem of the model performing too well on the training data and performing poorly on new data (i.e., overfitting) can be avoided, thus improving the practicality of the trained model.

[0076] Optionally, the classification and grading system can iteratively train the local initial model of each financial institution based on historical financial data and historical classification and grading rules in the training set. This can make full use of the local computing resources of financial institutions, accelerate the model optimization process, and improve the overall training efficiency.

[0077] Optionally, the classification and grading system tests the training model corresponding to each financial institution based on historical financial data and historical classification and grading rules in the test set, obtaining model performance indicators such as classification coverage, return rate, and data throughput. Classification coverage reflects the proportion of financial data that the model can accurately classify; return rate indicates the ratio of model classification results that are manually corrected or adjusted and then reclassified into the same category, indirectly reflecting the accuracy and stability of the model's classification; data throughput is a performance indicator used to measure the model's data processing speed, and this step has the following functions:

[0078] (1) Performance evaluation and verification: Through test set verification, the performance of the model on classification and grading tasks can be quantitatively evaluated to ensure the effectiveness and reliability of the model.

[0079] (2) Model transparency and interpretability: The clarification of performance indicators helps financial institutions understand and explain the behavior of the model, increasing the trust and acceptability of the model.

[0080] In one optional embodiment, during the process of classifying and grading financial data based on preset rules for matching financial data, from the dimensions of data type, data sensitivity, and business type, the classification and grading system first detects the classification and grading execution time corresponding to the financial data. Then, if the classification and grading execution time is greater than or equal to the preset time, the classification and grading system generates an early warning message.

[0081] Optionally, the classification and grading execution time refers to the time consumed by financial institutions when applying the data classification and grading model to process specific financial data, that is, the time required for the entire process from financial data input and processing to the output of classification and grading results.

[0082] Optionally, by monitoring the execution time of the classification and grading of financial data, the classification and grading system can monitor the operational efficiency of the system in real time. Long execution wait times indicate that there may be data processing bottlenecks in the system, which helps financial institutions to identify the bottlenecks in a timely manner and take corresponding measures to optimize them.

[0083] Optionally, if the classification and grading execution time is greater than or equal to the preset time, the classification and grading system generates an early warning message. The early warning mechanism can provide immediate feedback on the performance abnormalities of the classification and grading system, enabling financial institutions to respond quickly and prevent the decline in the efficiency of the classification and grading system from affecting classification and grading business operations. Through early warning, service interruptions caused by long processing times can be avoided, enhancing the stability and reliability of the system.

[0084] As described above, this application obtains data features, including semantic information, word frequency, and contextual information of keywords in financial data, through in-depth analysis of the data characteristics of financial data. This achieves the goal of accurately and quickly matching preset rules in a preset knowledge base based on data features. Subsequently, this application classifies and grades financial data according to data type, sensitivity, and business type based on the matched preset rules, ensuring the security and compliance of financial data management. This achieves the technical effect of efficient and accurate classification and grading of financial data, thereby solving the technical problem of poor performance in classifying and grading financial data based on existing technologies.

[0085] In one alternative embodiment, Figure 3 This is a flowchart of an optional data asset hierarchical classification method according to an embodiment of this application, such as... Figure 3 As shown, the method includes the following steps:

[0086] First, a comprehensive analysis is conducted on various security rules, data security classification standards issued by local regulators of different financial institutions, and historical cases. The analysis is carried out using a combination of "comprehensive literature analysis + classification and classification historical record backtracking + case studies" to obtain preset rules. Based on the summarized preset rules, a preset knowledge base is built that can be used by the classification and classification system.

[0087] Subsequently, when it is necessary to classify and grade financial data, the financial data to be classified is obtained in batches (e.g., 5000 data points per batch). Then, it is checked whether non-Chinese characters are used in the financial data to be classified (i.e.,...). Figure 5 The text data described in the foreign language is translated into Chinese using a pre-trained agent (i.e., a target translator) when foreign language data is available. Then, based on the preset rules in the preset knowledge base, the text data described in Chinese is matched to obtain classification and grading results.

[0088] For example, for the database table field KEY (password), the output will be a multi-level classification result of business data (Category 1) --> account information (Category 2) --> media information (Category 3) --> basic media information (Category 4), and the corresponding data security level will be defined as "Level 1" (high sensitivity) level.

[0089] Optionally, if the preset rules in the preset knowledge base cannot successfully classify and grade the financial data, a new type of rule can be generated using a pre-trained large language model to classify and grade the aforementioned text data in Chinese description mode.

[0090] Optionally, by using a large language model, the amount of manual calls during the data analysis process is gradually reduced. Ultimately, technicians only need to process long-tail data, which accounts for less than 10% of financial data, thus reducing labor costs. Furthermore, this technical solution enables the accumulation of pre-defined rule knowledge, forming a reusable pre-defined knowledge base.

[0091] Example 2

[0092] This application embodiment can also provide a financial data classification and grading device. It should be noted that the financial data classification and grading device of this application embodiment can be used to execute the financial data classification and grading method provided in this application embodiment. The following is a description of the financial data classification and grading device provided in this application embodiment.

[0093] According to an embodiment of this application, an apparatus for implementing the above-described method for classifying and grading financial data is also provided. Figure 4 This is a schematic diagram of an optional financial data classification and grading device according to an embodiment of this application, such as... Figure 4 As shown, the device includes: a data feature extraction unit 401, a rule matching unit 402, and a classification and grading unit 403.

[0094] Optionally, the data feature extraction unit 401 is used to extract features from the financial data to be classified and graded, and obtain the data features corresponding to the financial data. The data features are used to characterize the semantic information, word frequency, and contextual information of keywords in the financial data. The rule matching unit 402 is used to perform rule matching in a preset knowledge base based on the data features corresponding to the financial data, and obtain preset rules for matching the financial data. The preset rules in the preset knowledge base are obtained based on preset security specification documents and historical records of classifying and grading historical financial data. The classification and grading unit 403 is used to classify and grade the financial data from the dimensions of data type, data sensitivity, and business type based on the preset rules for matching the financial data, and obtain the classification and grading results of the financial data.

[0095] In one optional embodiment, the financial data classification and grading device further includes: a data cleaning unit, a data segmentation unit, a data transmission unit, a data translation unit, and a financial data determination unit.

[0096] Optionally, a data cleaning unit is used to clean the raw data in a preset database of a financial institution to obtain first data, wherein the raw data is text data described using L preset languages, where L is a positive integer; a data partitioning unit is used to partition the first data based on the L preset languages ​​to obtain L data segments, wherein the text data in each data segment is described using the same preset language; a data transmission unit is used to transmit each of the L data segments to a target translator corresponding to each data segment; a data translation unit is used to translate the preset language used to describe each data segment into a standard language through the target translator to obtain L target segments corresponding to the L data segments, wherein the standard language is one of the L preset languages; and a financial data determination unit is used to determine financial data based on the L target segments.

[0097] In one optional embodiment, the rule matching unit 402 includes: a rule feature extraction subunit, a correlation coefficient acquisition subunit, and a preset rule matching subunit.

[0098] Optionally, the rule feature extraction subunit is used to extract features from each preset rule in the preset knowledge base to obtain the rule features corresponding to each preset rule; the correlation coefficient acquisition subunit is used to obtain the correlation coefficient between the data features and the rule features corresponding to each preset rule; and the preset rule matching subunit is used to use the preset rule corresponding to the correlation coefficient as the preset rule for financial data matching when the correlation coefficient is greater than or equal to the preset coefficient.

[0099] In one optional embodiment, the financial data classification and grading device further includes: a rule creation unit, a testing unit, and a rule storage unit.

[0100] Optionally, the rule creation unit is used to create new rules based on the data characteristics of financial data through a target model when no preset rules matching the financial data are found. The target model is a large language model pre-trained based on historical financial data and historical classification and grading rules of P financial institutions, where P is a positive integer. The testing unit is used to perform a first test operation and a second test operation on the new rules. The first test operation is used to detect conflicts between the new rules and each preset rule in the preset knowledge base, and the second test operation is used to perform performance testing on the new rules. The rule storage unit is used to determine the classification and grading results of the financial data based on the new rules and store the new rules as new preset rules in the preset knowledge base when both the first and second test operations pass.

[0101] In one optional embodiment, the financial data classification and grading device further includes: a global initial model creation unit, a local model creation unit, a local model training unit, a gradient aggregation unit, a global initial model update unit, and a target model determination unit.

[0102] Optionally, the global initial model creation unit is used to perform initialization operations through the central server to obtain a global initial model, and transmit the model parameters of the global initial model to P financial institutions, where the P financial institutions are participants in federated learning; the local model creation unit is used to control each of the P financial institutions to create a local initial model based on the model parameters of the global initial model; the local model training unit is used to iteratively train the local initial model corresponding to each financial institution based on the historical financial data and historical classification and grading rules of each financial institution to obtain a local model; the gradient aggregation unit is used to encrypt and transmit the gradients of the local models of each of the P financial institutions to the central server, and through the central server, perform weighted summation of the gradients corresponding to each of the P financial institutions to obtain the target gradient; the global initial model update unit is used to iteratively update the global initial model based on the target gradient to obtain a global federated model, and transmit the model parameters of the global federated model to the P financial institutions; the target model determination unit is used to control each of the P financial institutions to update the local model of each financial institution based on the model parameters of the global federated model to obtain the target model of each financial institution.

[0103] In one optional embodiment, the local model training unit includes: a training sample partitioning subunit, an iterative training subunit, a training model testing subunit, and a local model determination subunit.

[0104] Optionally, the training sample partitioning subunit is used to divide the historical financial data and historical classification and grading rules of each financial institution into a training set and a test set corresponding to each financial institution; the iterative training subunit is used to iteratively train the local initial model of each financial institution based on the historical financial data and historical classification and grading rules in the training set to obtain the training model corresponding to each financial institution; the training model testing subunit is used to test the training model corresponding to each financial institution based on the historical financial data and historical classification and grading rules in the test set to obtain the model performance index of the training model corresponding to each financial institution, wherein the model performance index is at least one of the following: classification coverage, backflow rate, and data throughput; the local model determination subunit is used to determine the training model corresponding to each financial institution as the local model corresponding to each financial institution if the model performance index of the training model corresponding to each financial institution is within a preset performance index range.

[0105] In one optional embodiment, the financial data classification and grading device further includes an execution duration detection unit and an early warning information generation unit.

[0106] Optionally, the execution duration detection unit is used to detect the classification and grading execution duration corresponding to the financial data; the early warning information generation unit is used to generate early warning information when the classification and grading execution duration is greater than or equal to the preset duration.

[0107] As described above, this device obtains data features, including semantic information, word frequency, and contextual information of keywords in financial data, through in-depth analysis of the data characteristics of financial data. This achieves the goal of accurately and quickly matching preset rules in a preset knowledge base based on data features. Subsequently, this device classifies and grades financial data according to the matched preset rules from dimensions such as data type, sensitivity, and business type, ensuring the security and compliance of financial data management. This achieves the technical effect of efficient and accurate classification and grading of financial data, thereby solving the technical problem of poor performance in classifying and grading financial data based on existing technologies.

[0108] It should be noted that the data feature extraction unit 401, rule matching unit 402, and classification and grading unit 403 correspond to steps S201 to S203 in the method embodiment. The instances and application scenarios implemented by the above units and corresponding steps are the same, but are not limited to the content disclosed in the above embodiments. It should be noted that the above modules or units can be hardware or software components stored in memory (e.g., memory 104) and processed by one or more processors (e.g., processors 102a, 102b, ..., 102n). The above modules can also be part of the device and run in the computer terminal 10 provided in the embodiment.

[0109] Example 3

[0110] Embodiments of this application can also provide an electronic device. Figure 5 This is a structural block diagram of an electronic device according to an embodiment of this application, such as... Figure 5 As shown, the electronic device includes: one or more ( Figure 5 (Only one is shown) processor 502, memory 504, memory controller, and peripheral interface, wherein the peripheral interface is connected to the radio frequency module, audio module and display.

[0111] The memory can be used to store software programs and modules, such as the program instructions / modules corresponding to the methods and devices in the embodiments of this application. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, thereby realizing the above-mentioned classification and grading method for financial data.

[0112] The memory may include high-speed random access memory (RAM), and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include memory remotely located relative to the processor, which can be connected to the terminal via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks (LANs), mobile communication networks, and combinations thereof.

[0113] The processor can access information and applications stored in the memory via a transmission device to perform the following steps: extracting features from the financial data to be classified and graded to obtain data features corresponding to the financial data, wherein the data features are used to characterize the semantic information, word frequency, and contextual information of keywords in the financial data; performing rule matching in a preset knowledge base based on the data features corresponding to the financial data to obtain preset rules for matching the financial data, wherein the preset rules in the preset knowledge base are derived from preset security specification documents and historical records of classifying and grading historical financial data; and classifying and grading the financial data from the dimensions of data type, data sensitivity, and business type based on the preset rules for matching the financial data to obtain the classification and grading results of the financial data.

[0114] The processor can access information and applications stored in memory via a transmission device to perform the following steps: Cleaning the raw data in a financial institution's pre-defined database to obtain first data, where the raw data is text data described using L pre-defined languages, where L is a positive integer; dividing the first data into L data segments based on the L pre-defined languages, where the text data in each data segment is described using the same pre-defined language; transmitting each of the L data segments to a target translator corresponding to each data segment; translating the pre-defined language describing each data segment into a standard language using the target translator to obtain L target segments corresponding to the L data segments, where the standard language is one of the L pre-defined languages; and determining financial data based on the L target segments.

[0115] The processor can call the information and application stored in the memory through the transmission device to perform the following steps: extract features from each preset rule in the preset knowledge base to obtain the rule features corresponding to each preset rule; obtain the correlation coefficient between the data features and the rule features corresponding to each preset rule; if the correlation coefficient is greater than or equal to the preset coefficient, use the preset rule corresponding to the correlation coefficient as the preset rule for financial data matching.

[0116] The processor can access information and applications stored in memory via a transmission device to perform the following steps: In the absence of a pre-defined rule matching the financial data, a new rule is created based on the data characteristics of the financial data using a target model, where the target model is a large language model pre-trained based on historical financial data and historical classification and grading rules of P financial institutions, where P is a positive integer; a first test operation and a second test operation are performed on the new rule, where the first test operation is used to detect conflicts between the new rule and each pre-defined rule in the pre-defined knowledge base, and the second test operation is used to perform performance testing on the new rule; if both the first and second test operations pass, the classification and grading results of the financial data are determined based on the new rule, and the new rule is stored as a new pre-defined rule in the pre-defined knowledge base.

[0117] The processor can access information and applications stored in memory via a transmission device to execute the following steps: First, an initialization operation is performed through a central server to obtain a global initial model. The model parameters of the global initial model are then transmitted to P financial institutions, which are participants in the federated learning process. Next, each of the P financial institutions is controlled to create a local initial model based on the model parameters of the global initial model. Then, based on the historical financial data and historical classification and grading rules of each financial institution, its corresponding local initial model is iteratively trained to obtain a local model. The gradients of the local models of each of the P financial institutions are encrypted and transmitted to the central server. The central server then performs a weighted summation of the gradients corresponding to each of the P financial institutions to obtain the target gradient. Finally, the global initial model is iteratively updated based on the target gradient to obtain a global federated model. The model parameters of the global federated model are then transmitted to the P financial institutions. Finally, each of the P financial institutions is controlled to update its local model based on the model parameters of the global federated model to obtain the target model for each financial institution.

[0118] The processor can access information and applications stored in the memory via a transmission device to perform the following steps: dividing the historical financial data and historical classification and grading rules of each financial institution into a training set and a test set corresponding to each financial institution; iteratively training the local initial model of each financial institution based on the historical financial data and historical classification and grading rules in the training set to obtain the training model corresponding to each financial institution; testing the training model corresponding to each financial institution based on the historical financial data and historical classification and grading rules in the test set to obtain the model performance index of the training model corresponding to each financial institution, wherein the model performance index is at least one of the following: classification coverage, backflow rate, and data throughput; if the model performance index of the training model corresponding to each financial institution is within a preset performance index range, the training model corresponding to each financial institution is used as the local model corresponding to each financial institution.

[0119] The processor can access the information and application programs stored in the memory via the transmission device to perform the following steps: detect the classification and grading execution time corresponding to the financial data; and generate early warning information if the classification and grading execution time is greater than or equal to the preset time.

[0120] This application provides a classification and grading scheme for financial data. By deeply analyzing the data features of financial data, this application obtains data features including semantic information, word frequency, and contextual information of keywords within the financial data. This achieves accurate and rapid matching of preset rules in a pre-defined knowledge base based on data features. Subsequently, this application classifies and grades the financial data according to the matched preset rules from dimensions such as data type, sensitivity, and business type, ensuring the security and compliance of financial data management. This achieves efficient and accurate classification and grading of financial data, thereby solving the technical problem of poor performance in classification and grading of financial data based on existing technologies.

[0121] Those skilled in the art will understand that Figure 5 The structure shown is for illustrative purposes only. Electronic devices can also be smartphones, tablets, PDAs, mobile internet devices, PADs, and other terminal devices. Figure 5 This does not limit the structure of the aforementioned electronic device. For example, electronic devices may also include components that are more... Figure 5 The more or fewer components shown (such as network interfaces, display devices, etc.), or having the same ​ The different configurations shown.

[0122] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing the hardware related to the terminal device. The program can be stored in a computer-readable storage medium, which may include: flash drive, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0123] Example 4

[0124] Embodiments of this application may also provide a storage medium.

[0125] Optionally, in this embodiment of the application, the storage medium can be used to store the program code executed by the financial data classification and grading method provided in the above method embodiment.

[0126] Optionally, in this embodiment, the storage medium may be located in any computer terminal in a group of computer terminals in a computer network, or in any mobile terminal in a group of mobile terminals.

[0127] This application also provides a computer program product that, when executed on a data processing device, is suitable for performing steps of a classification and grading method for financial data.

[0128] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0129] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0130] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0131] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0132] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0133] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard drive, magnetic disk, or optical disk.

[0134] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for classifying and grading financial data, characterized in that, include: Feature extraction is performed on the financial data to be classified and graded to obtain the data features corresponding to the financial data, wherein the data features are used to characterize the semantic information, word frequency and contextual information of the keywords in the financial data; Based on the data characteristics corresponding to the financial data, rule matching is performed in a preset knowledge base to obtain preset rules for matching the financial data. The preset rules in the preset knowledge base are obtained based on preset security specification documents and historical records that classify and grade historical financial data. Based on the preset rules for matching the financial data, the financial data is classified and graded from the dimensions of data type, data sensitivity, and business type to obtain the classification and grading results of the financial data.

2. The method for classifying and grading financial data according to claim 1, characterized in that, Before extracting features from the financial data to be classified, the classification and grading method for the financial data also includes: Data cleaning is performed on the raw data in the preset database of financial institutions to obtain the first data, wherein the raw data is text data described in L preset languages, where L is a positive integer; Based on the L preset languages, the first data is divided into L data segments, wherein the text data in each data segment is described using the same preset language; Each of the L data segments is transmitted to the target translator corresponding to each data segment; The target translator translates the preset language used to describe each data segment into a standard language to obtain L target segments corresponding to the L data segments, wherein the standard language is one of the L preset languages; The financial data is determined based on the L target segments.

3. The method for classifying and grading financial data according to claim 1, characterized in that, Based on the data characteristics corresponding to the financial data, rule matching is performed in a preset knowledge base to obtain preset rules for matching the financial data, including: Feature extraction is performed on each preset rule in the preset knowledge base to obtain the rule features corresponding to each preset rule; Obtain the correlation coefficient between the data features and the rule features corresponding to each preset rule; If the correlation coefficient is greater than or equal to a preset coefficient, the preset rule corresponding to the correlation coefficient shall be used as the preset rule for matching the financial data.

4. The method for classifying and grading financial data according to claim 1, characterized in that, After performing rule matching in a preset knowledge base based on the data characteristics corresponding to the financial data, the classification and grading method for the financial data further includes: In the absence of a pre-defined rule for matching the financial data, a new rule is created based on the data characteristics of the financial data through a target model. The target model is a large language model pre-trained based on historical financial data and historical classification and grading rules of P financial institutions, where P is a positive integer. The novel rule is subjected to a first test operation and a second test operation, wherein the first test operation is used to detect conflicts between the novel rule and each preset rule in the preset knowledge base, and the second test operation is used to perform performance testing on the novel rule. If both the first and second test operations pass, the classification and grading results of the financial data are determined based on the new rules, and the new rules are stored as new preset rules in the preset knowledge base.

5. The method for classifying and grading financial data according to claim 4, characterized in that, The training steps for the target model also include: An initialization operation is performed by a central server to obtain a global initial model. The model parameters of the global initial model are then transmitted to the P financial institutions, which are the participants in the federated learning process. Control each of the P financial institutions to create a local initial model based on the model parameters of the global initial model; Based on the historical financial data and historical classification and grading rules of each financial institution, the local initial model corresponding to the financial institution is iteratively trained to obtain a local model. The gradient of the local model of each of the P financial institutions is encrypted and transmitted to the central server. The central server then performs a weighted summation of the gradients corresponding to each of the P financial institutions to obtain the target gradient. The global initial model is iteratively updated based on the target gradient to obtain a global federated model, and the model parameters of the global federated model are transmitted to the P financial institutions. By controlling each of the P financial institutions, and updating the local model of each financial institution based on the model parameters of the global federated model, the target model of each financial institution is obtained.

6. The method for classifying and grading financial data according to claim 5, characterized in that, Based on the historical financial data and historical classification and grading rules of each financial institution, the local initial model corresponding to that financial institution is iteratively trained to obtain a local model, including: The historical financial data and historical classification and grading rules of each financial institution are divided into training sets and test sets corresponding to each financial institution. Based on the historical financial data and historical classification and grading rules in the training set, the local initial model of each financial institution is iteratively trained to obtain the training model corresponding to each financial institution. Based on the historical financial data and historical classification and grading rules in the test set, the training model corresponding to each financial institution is tested to obtain the model performance index of the training model corresponding to each financial institution, wherein the model performance index is at least one of the following: Category coverage, return rate, and data throughput; If the model performance index of the training model corresponding to each financial institution is within a preset performance index range, the training model corresponding to each financial institution will be used as the local model corresponding to each financial institution.

7. The method for classifying and grading financial data according to claim 1, characterized in that, In the process of classifying and grading the financial data based on preset rules for matching the financial data, from the dimensions of data type, data sensitivity, and business type, the classification and grading method for financial data further includes: Detect the execution time of the classification and grading corresponding to the financial data; If the execution time of the classification and grading is greater than or equal to the preset time, an early warning message is generated.

8. A classification and grading device for financial data, characterized in that, include: The data feature extraction unit is used to extract features from the financial data to be classified and graded, and obtain the data features corresponding to the financial data. The data features are used to characterize the semantic information, word frequency and contextual information of the keywords in the financial data. The rule matching unit is used to perform rule matching in a preset knowledge base based on the data characteristics corresponding to the financial data to obtain preset rules for matching the financial data. The preset rules in the preset knowledge base are obtained based on preset security specification documents and historical records that classify and classify historical financial data. The classification and grading unit is used to classify and grade the financial data based on preset rules for matching the financial data, from the dimensions of data type, data sensitivity, and business type, and obtain the classification and grading results of the financial data.

9. A computer program product, characterized in that, The computer program product includes a computer program, wherein, when the computer program is executed, it controls the computer program product to perform the classification and grading method for financial data as described in any one of claims 1 to 7.

10. An electronic device, characterized in that, It includes one or more processors and a memory, the memory being used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors cause the one or more processors to implement the financial data classification and grading method according to any one of claims 1 to 7.

Images