A method, apparatus, system, device, and medium for storing financial data.

Abstract

This application provides a method, apparatus, system, device, and medium for storing financial data, including: acquiring financial information; processing the financial information for detection; if the detection result is normal, extracting feature information from the financial information based on a preset feature model, and determining a target category from multiple preset classification profiles based on the feature information; extracting feature text and feature values ​​from the feature information, fusing the feature values ​​and feature text through a mapping function to obtain physical ciphertext, extending the physical ciphertext by calling a pseudo-random sub-cipher generation algorithm, and binding the extended ciphertext with a user ID to obtain an encryption factor; dividing the financial information into multiple information blocks, encrypting the multiple information blocks using the encryption factor to obtain encrypted information, and storing the encrypted information in a target database corresponding to the target category. This invention uses encryption factors to encrypt financial information, which can reduce the risk of information theft or tampering, thereby improving the security of information storage.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of device data security, in particular to a financial data storage method, device, system, equipment and medium. BACKGROUND

[0002] With the continuous expansion of the scale of enterprises, the financial activities between groups and member units are also increasing, and often require relevant personnel to jointly participate in the filling, checking, reporting and other work of financial information. In the process of handling financial information, relevant personnel inevitably cause information security problems due to frequent communication, which poses a major challenge to corporate data security. And the traditional encryption method still has security risks in the financial data storage link, which makes the financial data exist the risk of leakage, and affects the production and operation of enterprises.

[0003] Since financial data belongs to the private data of enterprises and users, in order to protect the rights and interests of enterprises and users, one of the commonly used encryption methods is to set a corresponding storage database, and use symmetric encryption or asymmetric encryption technology (such as DES, AES and RSA, etc.) to encrypt the entire financial database and the data in the database, in order to protect user privacy.

[0004] But the commonly used method has the following technical problems: the encryption key used is fixed, once the key is cracked, the financial database becomes an open source database, and the various data stored therein is easy to be tampered with by multiple persons, and the security of enterprise and user data cannot be guaranteed. SUMMARY

[0005] In view of the above problems, the present application is proposed to provide a financial data storage method, device, system, equipment and medium to overcome the above problems or at least partially solve the above problems, which comprises:

[0006] A financial data storage method, the method comprises:

[0007] Obtaining financial information to be processed, and detecting and processing the financial information to obtain a detection result;

[0008] If the detection result is normal, extracting feature information from the financial information based on a preset feature model, and determining a target category in a plurality of preset classification portraits according to the feature information;

[0009] Extracting feature text and feature numerical value from the feature information, fusing the feature numerical value and the feature text through a mapping function to obtain physical ciphertext, calling a pseudo-random sub-cipher generation algorithm to expand the physical ciphertext, and binding the expanded ciphertext with a user ID to obtain an encryption factor;

[0010] The financial information is divided into a plurality of information blocks, the plurality of information blocks are encrypted by using the encryption factor to obtain encrypted information, and the encrypted information is stored in a target database corresponding to the target category.

[0011] In a possible implementation, the method further includes:

[0012] The method further includes:

[0013] The method further includes:

[0014] The method further includes:

[0015] If the image similarity is greater than or equal to a preset first similarity, the processing result of the detection processing is determined to be normal.

[0016] If the image similarity is less than the preset first similarity, the processing result of the detection processing is determined to be abnormal.

[0017] In a possible implementation, the method further includes:

[0018] The method further includes:

[0019] The method further includes:

[0020] The method further includes:

[0021] The method further includes:

[0022] If the text similarity is greater than or equal to a preset second similarity, the processing result of the detection processing is determined to be normal.

[0023] If the text similarity is less than the preset second similarity, the processing result of the detection processing is determined to be abnormal.

[0024] In a possible implementation, the extracting feature information from the financial information based on a preset feature model and determining a target category in a plurality of preset classification portraits according to the feature information comprises:

[0025] The feature information is extracted from the financial information based on a preset feature model, and a portrait label set corresponding to each preset classification portrait is obtained, the portrait label set including a plurality of labels;

[0026] The semantic matching degrees are obtained by calculating semantic matching degrees between the feature information and labels of each portrait label set.

[0027] The semantic matching degree with the largest value is selected from the plurality of semantic matching degrees, and a category of a preset classification portrait corresponding to the semantic matching degree with the largest value is determined to obtain the target category.

[0028] In a possible implementation, the fusing the feature value and the feature text by using a mapping function to obtain a physical ciphertext comprises:

[0029] The feature value and the feature text are respectively converted into a matrix to obtain a text matrix and a value matrix;

[0030] The text matrix and the value matrix are spliced by using a mapping function to obtain a spliced matrix, and the spliced matrix is converted into numerical data to obtain the physical ciphertext.

[0031] In a possible implementation, the fusing the feature value and the feature text by using a mapping function to obtain a physical ciphertext comprises:

[0032] The feature text is encrypted by using secondary encoding to obtain a text ciphertext, and a parameter value is extracted from the feature value according to a preset numerical interval;

[0033] The parameter value is mapped to a character of the text ciphertext by using a mapping function to obtain a mapping ciphertext, and the mapping ciphertext is converted into a multi-dimensional feature to obtain the physical ciphertext.

[0034] A storage device of financial data, the device comprising:

[0035] An acquisition module configured to acquire financial information to be processed, and perform detection processing on the financial information to obtain a detection result;

[0036] A determination module configured to, if the detection result is normal, extract feature information from the financial information based on a preset feature model, and determine a target category in a plurality of preset classification portraits according to the feature information.

[0037] An encryption module is configured to extract feature text and feature values from the feature information, fuse the feature values and the feature text by a mapping function to obtain physical ciphertext, call a pseudo-random sub-cipher generation algorithm to expand the physical ciphertext, bind the expanded ciphertext with a user ID, and obtain an encryption factor;

[0038] A storage module is configured to divide the financial information into a plurality of information blocks, encrypt the plurality of information blocks by the encryption factor to obtain encrypted information, and store the encrypted information in a target database corresponding to the target category.

[0039] A financial data storage system, which comprises a processing terminal and a plurality of storage databases, each of the storage databases corresponding to a category of a classification portrait;

[0040] The processing terminal is in communication connection with the plurality of storage databases respectively;

[0041] The processing terminal executes the financial data storage method as described above.

[0042] An apparatus comprising a processor, a memory, and a computer program stored on the memory and executable on the processor, the computer program, when executed by the processor, implementing the steps of the financial data storage method as described above.

[0043] A computer-readable storage medium, which stores a computer program, the computer program, when executed by a processor, implementing the steps of the financial data storage method as described above.

[0044] The present application has the following advantages:

[0045] In the embodiments of the present application, the financial information to be processed can be acquired, and the financial information can be detected and processed to obtain a detection result. If the detection result is normal, feature information is extracted from the financial information based on a preset feature model, and a target category is determined among a plurality of preset classification portraits according to the feature information. Feature text and feature values are extracted from the feature information, physical ciphertext is obtained by fusing the feature values and the feature text by a mapping function, the physical ciphertext is expanded by calling a pseudo-random sub-cipher generation algorithm, and the expanded ciphertext is bound with a user ID to obtain an encryption factor. The financial information is divided into a plurality of information blocks, the plurality of information blocks are encrypted by the encryption factor to obtain encrypted information, and the encrypted information is stored in a target database corresponding to the target category. The financial information can be encrypted by the encryption factor, so as to reduce the risk of information being stolen or tampered with, and improve the security of information storage. Moreover, the encrypted information is classified and stored, so that the same type of data can be stored together, the data gap is broken, business personnel can analyze and process the data by business type, and the efficiency of analysis and research is improved. Attached Figure Description

[0046] To more clearly illustrate the technical solution of this application, the drawings used in the description of this application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0047] Figure 1 This is a flowchart illustrating the steps of a method for storing financial data according to an embodiment of this application;

[0048] Figure 2 This is a structural block diagram of a financial data storage device provided in one embodiment of this application;

[0049] Figure 3 This is a structural block diagram of a financial data storage system provided in one embodiment of this application;

[0050] Figure 4 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0051] To make the objectives, features, and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0052] As the company continues to expand, financial activities among the group and its member units also increase, frequently requiring stakeholders to participate in the completion, verification, and submission of financial information. During the processing of financial information, frequent communication inevitably leads to information security issues, posing a significant challenge to the company's data security. Traditional encryption methods also have security vulnerabilities in the financial data storage process, creating a risk of data leakage and impacting the company's production and operations.

[0053] Since financial data is considered private data for both businesses and users, a common encryption method to protect their rights is to set up a corresponding storage database and use symmetric or asymmetric encryption technologies (such as DES, AES, and RSA) to encrypt the entire financial database and the data within it, thereby protecting user privacy.

[0054] However, the current commonly used method has the following technical problems: the encryption keys used are fixed, once the key is cracked, the financial database becomes an open source database, and various data stored therein is easily tampered with by multiple persons, and the security of enterprise and user data cannot be guaranteed.

[0055] To solve the above technical problems, with reference to Figure 1 , a step flow chart of a financial data storage method provided by an embodiment of the present application is shown;

[0056] In an embodiment, the financial data storage method is applicable to a data platform of a terminal or an enterprise. The data platform can be connected with multiple different storage databases, each of which corresponds to a data storage category. For example, the enterprise is a financial service company, a storage database is used to store financial data of an insurance transaction category; a storage database is used to store financial data of an insurance transaction category; a storage database is used to store financial data of an employee salary category; and a storage database is used to store internal financial data of the enterprise.

[0057] To further refine the classification type, each storage database can be divided into multiple different storage areas, and each storage area can store different characteristic data. For example, the a storage database used to store financial data of the insurance transaction category can be divided into three areas, the first area stores financial data of life insurance, the second area stores financial data of vehicle insurance, and the third area stores financial data of business accident.

[0058] Optionally, the data platform can also be connected with multiple intelligent terminals for communication, and the intelligent terminals can be intelligent terminals of business service personnel or intelligent terminals of transaction users. The user can upload transaction-related financial data through the intelligent terminal, and the business service personnel can view various financial information of the transaction user through the intelligent terminal, thereby providing corresponding business services.

[0059] For example, the financial data storage method can include:

[0060] S11, obtaining financial information to be processed, and detecting the financial information to obtain a detection result.

[0061] In an embodiment, the financial information to be processed can be obtained. The financial information to be processed can include various transaction data. For example, the user needs to report an accident for vehicle repair, and the enterprise provides the user with an accident reporting service, and the financial information can include: user reservation data, accident transaction data, accident detection image data, accident audit result data, and expense details data, which can include transaction-related data on the user side, and can also include transaction-related data on the enterprise side.

[0062] In an embodiment, the detection process is a process of determining whether the financial information is abnormal based on text or image of the financial information.

[0063] After the detection process is completed, the financial information can be classified according to the content of the financial information, and after the classification, the financial information can be encrypted, and the encrypted financial information can be stored in a corresponding storage database according to the classification result, so as to reduce the risk of tampering or stealing and improve the security of the stored data.

[0064] Before classification and storage, the financial information can be subjected to a detection process to determine whether the financial information is abnormal, for example, whether the risky transaction is abnormal, whether the payment data is abnormal, whether the audit result data is abnormal, and whether the detection image data is abnormal. When it is determined that each item of data is not abnormal, subsequent classification and storage processing is performed to avoid using incorrect data for classification and storage.

[0065] In an optional embodiment, the financial information can include text and numerical values related to transactions, and can also include transaction vouchers, invoices or recorded images, and can also include related reports, etc.

[0066] The numerical values and images of the financial information can be compared to view the process and related content of the business transaction. In order to perform abnormal detection processing on the images of the financial information, it is determined whether the transaction is incorrect to avoid subsequent financial audit errors, and also to avoid storing incorrect financial data. As an example, the financial information to be processed is obtained, and the financial information is subjected to a detection process to obtain a detection result, which can include the following sub-steps:

[0067] S21, obtaining financial information to be processed, and extracting a first financial image and a first financial text from the financial information.

[0068] S22, extracting a financial key value from the first financial text, and constructing a comparative financial report using the financial key value and a preset report template, and obtaining a comparative financial image by intercepting an image corresponding to the comparative financial report.

[0069] S23, calculating an image similarity between the first financial image and the comparative financial image.

[0070] S24, if the image similarity is greater than or equal to a preset first similarity, determining that the processing result of the detection process is normal.

[0071] S25, if the image similarity is less than the preset first similarity, determining that the processing result of the detection process is abnormal.

[0072] In an embodiment, after obtaining the financial information to be stored, the first financial image and the first financial text can be extracted from the financial information respectively. The first financial image can be a report of the current business, and specifically can be a report constructed by a business auditor of an enterprise according to relevant data of a transaction record. The first financial text can be financial data uploaded by a user about the current transaction.

[0073] For example, the current business service is a user life insurance business, and the first financial image can be data of an xx life insurance recorded by a business personnel after the user purchases the xx life insurance, and a report constructed by an auditor according to the data uploaded by the business personnel, and an image of the report, and the first financial image.

[0074] The first financial text is relevant data uploaded by the user on a smart terminal after the user purchases the xx life insurance. Then the first financial text is read, and a preset keyword extraction model is used to identify the first financial text, and financial keywords of the current business are extracted, and corresponding numerical values of the financial keywords are extracted, and financial key numerical values are obtained. There can be multiple financial keywords. For example, the business is life insurance, and the financial keywords can include the insurance period, the insurance amount, and the insurance type. After multiple financial keywords are extracted, corresponding numerical values of the financial keywords can be extracted, and financial key numerical values can be obtained.

[0075] Then, the financial key numerical values and a preset report template are used to construct a comparative financial report, an image corresponding to the comparative financial report is intercepted, and a comparative financial image is obtained. The preset report template can be a template used by the auditor to construct the report, or can be a template used by the first financial image.

[0076] Finally, the image similarity between the target image and the comparative image can be calculated, and if the image similarity is greater than or equal to a preset first similarity, the processing result of the detection processing is determined to be normal.

[0077] On the contrary, if the image similarity is less than the preset first similarity, the processing result of the detection processing is determined to be abnormal.

[0078] Through the above detection processing method, the transaction-related data uploaded by the user can be determined, and the transaction-related data uploaded by the user is used to construct a report, and an image of the report is intercepted to obtain a comparative financial image. Then, an image of a report constructed by a business personnel and an auditor is obtained, and a first financial image is obtained.

[0079] Since the templates of the two images correspond to the report, when detecting, the similarity of the two images can be calculated, and the values of the two report images are determined based on the similarity. If the similarity is large, it means that the values of the two images are the same, and the detection result is normal. On the contrary, if the similarity is small, it means that the values of the two images deviate greatly, which may be due to the data error of the auditor or the data error of the user uploading, and the detection result is abnormal.

[0080] In order to further investigate, if the detection result is abnormal, the detection result can be transmitted to the auditor first, so that the auditor re-audits the data of the business personnel to determine whether the related financial data of the transaction or the constructed report is deviated. If there is no deviation, the detection result can be transmitted to the user, and the user detects whether the uploaded data is deviated.

[0081] When any party has a modification, the detection process can be performed again to ensure that the data of the two parties is consistent and can be matched, and then subsequent storage processing is performed.

[0082] The data uploaded by the user and the report recorded by the enterprise can be obtained, and the data of the two parties is detected to ensure whether the data is abnormal, so as to avoid storing abnormal data.

[0083] In another optional embodiment, the transaction of the business involves multiple operations, and each operation involves multiple values. In order to directly compare whether each value is wrong, as an example, the obtaining of the financial information to be processed and the detection processing of the financial information to obtain a detection result can include the following sub-steps:

[0084] S31, obtaining the financial information to be processed, and extracting a second financial image and a second financial text from the financial information.

[0085] S32, identifying a transaction mark area of the second financial image and extracting a mark report text corresponding to the transaction mark area.

[0086] S33, splitting the data of the mark report text to obtain a mark information set, and splitting the data of the second financial text to obtain a financial information set.

[0087] S34, calculating a text similarity according to the mark information set and the financial information set.

[0088] S35, if the text similarity is greater than or equal to a preset second similarity, determining that the processing result of the detection processing is normal.

[0089] S36, if the text similarity is less than the preset second similarity, determining that the processing result of the detection processing is abnormal.

[0090] In an operation mode, after obtaining the financial information to be stored, the second financial image and the second financial text can be extracted from the financial information. The second financial image can be a screenshot or image related to a transaction voucher uploaded by the user, including an invoice, a payment record, etc. The second financial text can be a text of transaction information recorded by a business staff.

[0091] For example, the business is a vehicle accident business, and the second financial image can be an image of a 4S shop accident payment or an image of a received accident compensation receipt. The second financial text is the information of this accident recorded by the business staff of the business service enterprise.

[0092] Then, the second financial image is recognized, and the transaction mark area in the second financial image is recognized and the mark report text corresponding to the transaction mark area is extracted. The transaction mark area can be a mark added by the user in the image, such as a circle, a mark of the transaction amount and the transaction time, etc. Then the mark report text in the transaction mark area can be extracted.

[0093] Then, the mark report text can be read, and the mark report text can be recognized by the preset keyword extraction model, and the keywords of the mark report text can be extracted, and the values corresponding to the keywords can be obtained.

[0094] Then each keyword and value can be converted into a vector, and multiple vectors can be combined into a set to obtain a mark information set.

[0095] Similarly, the second financial text can also be recognized and split by the preset keyword extraction model, and the keywords of the second financial text can be extracted, and the values corresponding to the keywords can be obtained. At the same time, each keyword and value can be converted into a vector, and multiple vectors can be combined into a set to obtain a financial information set.

[0096] The vector distance of the mark information set and the vector distance of the financial information set are calculated to obtain a text similarity. If the text similarity is greater than or equal to a preset second similarity, it can be determined that the processing result of the detection processing is normal.

[0097] On the contrary, if the text similarity is less than the preset second similarity, it can be determined that the processing result of the detection processing is abnormal.

[0098] Through the above-mentioned manner, the image uploaded by the user and the data recorded by the enterprise can be obtained, and the data of both parties can be detected to determine whether the transaction data of the user in this business is the same as the data recorded by the enterprise in this business, and whether the data is abnormal can be ensured to avoid storing abnormal data.

[0099] Optionally, when calculating the vector distance between the word vectors of the two sets, the cosine value (cosine similarity), Euclidean distance, minimum edit distance, Hamming distance, etc. distance algorithm can be implemented, which is not limited here, and can be flexibly selected according to the actual application scene.

[0100] Similarly, when the text similarity is less than the preset second similarity, it is determined that the processing result of the detection processing is abnormal. It can be determined that the transaction data of the user in this business and the data recorded by the enterprise in this business do not match, and any one is wrong, which may be the image data uploaded by the user or the data recorded by the enterprise.

[0101] In order to prompt the user in time, when the processing result of the detection processing is abnormal, an alarm prompt can be sent to the user and the business personnel of the enterprise to prompt the relevant personnel to perform audit processing.

[0102] In addition, the registration detail data, test result data, medication data, and cost detail data can also be subjected to detection processing.

[0103] For business transaction data, the amount of the user's collection data and the amount of the enterprise's payment data can be compared. If they are the same, the detection result is normal, otherwise it is abnormal.

[0104] For users who perform the same business, their business data is generally the same as the previous business data. In order to simultaneously perform abnormal detection processing on multiple data, in an optional embodiment, when it is determined that the business is the same, the isolation forest algorithm can be used to calculate the abnormal score value of the above-mentioned user uploaded business data and historical data. The historical data can include data provided to the same user for the same business in the past.

[0105] The isolation forest algorithm is an anomaly detection method based on tree structure. In the analysis of anomaly detection processing, the isolation forest algorithm mainly calculates the path length of the business data and the historical data as a reference. The data with too short path is determined as an outlier, and vice versa. By counting multiple outliers, the abnormal score value can be obtained. If the abnormal score value is greater than the preset score value, it means that the detection result is abnormal, otherwise it means that the detection result is normal.

[0106] In yet another optional embodiment, each item of data can be detected, and when the result of each detection is normal, it is determined that the processing result of the detection processing is normal; if any one detection result is abnormal, the processing result of the detection processing is also abnormal.

[0107] With reference to the above embodiments, it can be known that any one of the data errors leads to the detection processing result being abnormal. In order to timely inform the business, the error data can be corrected. As an example, after the step of obtaining the financial information to be processed and performing detection processing on the financial information to obtain a detection result, the method further comprises:

[0108] S41, determining that the detection processing result is abnormal, and generating an alarm prompt information.

[0109] S42, in response to the replacement parameter, performing adjustment processing on the financial information by using the replacement parameter. The replacement parameter is a modification parameter added by the business personnel according to the alarm prompt information after the alarm prompt information is displayed to the business.

[0110] If the detection processing result is abnormal, the alarm prompt information is generated, and the alarm prompt information can be displayed to the user and the business personnel, so as to inform the user and the business personnel that the financial information is incorrect.

[0111] In an operation mode, the alarm prompt information can be generated by using the incorrect data or comparing the two incorrect data.

[0112] After the business personnel views the incorrect data or the two incorrect data in the alarm prompt information, the business personnel can determine the error cause and input the replacement parameter. After receiving the replacement parameter input by the user or the business personnel, the replacement parameter is used to perform adjustment processing on the financial information. Specifically, the replacement parameter can be used to modify the data corresponding to the financial information.

[0113] In another operation mode, in order to keep the modification trace for subsequent auditing and analysis, the incorrect data can be marked and kept in the financial information, and the replacement parameter can be added to the financial information.

[0114] S12, if the detection result is normal, extracting feature information from the financial information based on a preset feature model, and determining a target category in a plurality of preset classification portraits according to the feature information.

[0115] When the detection processing result is normal, the feature information needs to be extracted from the financial information, the category of the financial information is determined by using the feature information, and the feature information can be used for encryption, so as to improve the security of the financial information and reduce the risk of being stolen or tampered.

[0116] In an optional embodiment, the preset feature model can be called to extract the corresponding feature information from the financial information.

[0117] Since the financial information includes the user's appointment data, the transaction data, the loss detection image data, the loss audit result data, the expense detail data, and the like, different data have different characteristics.

[0118] In order to be targeted for different data, in an operation mode, for the text and words of the above-mentioned data, a preset BERT model can be called for extraction. For the images or videos of the above-mentioned data, a preset DeepLab model, CNN model or FCN model can be called for extraction. For the numerical values of the above-mentioned data, a preset Unet deep learning model can be called for extraction.

[0119] For text, transaction type and business location characteristics can be extracted. For numerical values, specific numerical values and parameters of transactions can be extracted. For videos or images, labeled regions of transactions can be obtained.

[0120] In an embodiment, after obtaining the characteristic information, the category to which the financial information belongs can be determined according to the characteristic information, so as to obtain the target category corresponding to the stored data.

[0121] In order to clearly and meticulously distinguish the categories corresponding to different businesses, different labels can be added to different businesses, so that each business can correspond to multiple labels. By comprehensively considering the multiple labels corresponding to each business, a classification portrait corresponding to each business can be obtained, and each business can correspond to a category, and each classification portrait corresponds to a category.

[0122] A plurality of portraits can be constructed in advance according to different businesses, so as to obtain a plurality of preset classification portraits, and then the target category can be determined according to the characteristic information in one of the plurality of preset classification portraits.

[0123] In an optional embodiment, there are a plurality of preset classification portraits, and each preset classification portrait includes a plurality of labels. In order to fit the labels of each preset classification portrait, as an example, the preset characteristic model is used to extract the characteristic information from the financial information, and the target category is determined according to the characteristic information in the plurality of preset classification portraits, including:

[0124] S121, based on a preset characteristic model, extracting characteristic information from the financial information, and obtaining a portrait label set corresponding to each preset classification portrait, the portrait label set including a plurality of labels.

[0125] S122, using the characteristic information and the labels of each portrait label set to calculate semantic matching degrees, obtaining a plurality of semantic matching degrees.

[0126] S123, selecting a semantic matching degree with the largest value from a plurality of semantic matching degrees, and determining the category of the preset classification portrait corresponding to the semantic matching degree with the largest value, to obtain the target category.

[0127] In one embodiment, multiple image tags contained in each preset category image can be obtained and combined to form multiple image tags, thereby obtaining an image tag set, which contains multiple tags;

[0128] Next, the semantic matching degree can be calculated using the feature information and the labels of the first image label set to obtain the first semantic matching degree. The semantic matching degree can then be calculated using the feature information and the labels of the second image label set to obtain the second semantic matching degree. This process can be repeated to calculate the semantic matching degree using the feature information and the labels of each image label set, resulting in multiple semantic matches.

[0129] In one operation mode, there are multiple feature information items. Each feature information item can be converted into a vector. Simultaneously, each image label of the preset classification image can also be converted into a vector. During calculation, the matching degree of the multiple vectors of the calculated feature information is calculated with the matching degree of multiple vectors of the first preset classification image. Then, the matching degrees are summed and averaged to obtain the semantic matching degree between the feature information and the first preset classification image. Next, the matching degree of the multiple vectors of the calculated feature information is calculated with the multiple vectors of the second preset classification image, and the matching degrees are summed and averaged to obtain the semantic matching degree between the feature information and the second preset classification image. This process continues.

[0130] In another approach, each feature information is converted into a vector, and each image label of the preset classification image is also converted into a vector. During calculation, the vector distances between multiple vectors of the feature information and multiple vectors of the first preset classification image can be calculated, and then the multiple vector distances are summed and averaged to obtain the semantic matching degree between the feature information and the first preset classification image.

[0131] In another approach, there are multiple feature information items, and the preset category portrait contains multiple portrait tags. The semantic matching degree can be obtained by counting the number of portrait tags in the preset category portrait that are identical to the feature information, and then calculating the ratio of the number of identical portrait tags to the total number of portrait tags. For example, if there are 10 feature information items, and the first preset category portrait contains 100 portrait tags, and 5 of them are identical to the feature information, then the semantic matching degree = 5 / 100 = 0.05.

[0132] The specific operating method can be adjusted according to actual needs, and is not limited here.

[0133] After calculating multiple semantic matching scores, the semantic matching score with the largest value can be selected from the multiple semantic matching scores, and the category corresponding to the preset classification profile with the largest semantic matching score can be determined, and that category can be used as the target category.

[0134] S13. Extract feature text and feature values ​​from the feature information, fuse the feature values ​​and feature text through a mapping function to obtain physical ciphertext, call a pseudo-random sub-cipher generation algorithm to expand the physical ciphertext, and bind the expanded ciphertext with the user ID to obtain the encryption factor.

[0135] After obtaining the target category, specific encryption factors can be constructed using feature information, and then these encryption factors can be used to encrypt financial information. Since encryption factors can be constructed using feature information, they can be tailored to the user's specific situation, forming unique encryption factors. Even if the stored information is stolen, hackers will find it difficult to crack the encrypted information, thus ensuring that the information is not tampered with.

[0136] Furthermore, once bound to a user ID, the corresponding encryption factor can be found based on the user ID each time, and then the encryption factor can be used to decrypt the encrypted information.

[0137] Since the feature information can include numerical values, text, and images, one or more feature information can be selected to construct an encryption factor. In one embodiment, the step of fusing the feature numerical values ​​and the feature text through a mapping function to obtain the physical ciphertext may include the following sub-steps:

[0138] S51. Convert the feature values ​​and feature text into matrices respectively to obtain a text matrix and a value matrix.

[0139] S52. By concatenating the text matrix and the numerical matrix using a mapping function, a concatenated matrix is ​​obtained, and the concatenated matrix is ​​converted into numerical data to obtain physical ciphertext.

[0140] In one operation mode, feature text and feature values ​​can be extracted from feature information.

[0141] Next, the feature text can be converted into a matrix to obtain a text matrix. Alternatively, the feature values ​​can be converted into a matrix to obtain a numerical matrix.

[0142] In one approach, the feature text is input into a language model, which performs initial text encoding and outputs encoded word vectors. These encoded word vectors are then input into a stylized encoding network, which performs stylized transformations on them, outputting stylized feature vectors. Finally, the stylized feature vectors are processed using head and tail semantic feature extraction and a dual affine attention mechanism to obtain the text matrix.

[0143] The language model can be a pre-trained language model based on modern text, such as using a BERT-based CASED pre-trained model to encode the text. Further, a stylized encoding network is used to perform stylized feature transformation on the encoded word vectors, enabling the model to better recognize the semantics of finance and avoid incorrect sentence segmentation due to misunderstandings of the financial meaning. The stylized encoding network can employ a Transformer network. Each layer of the Transformer network structure can include a bidirectional self-attention network with rotational position encoding and a multilayer perceptron, where rotational position encoding better incorporates relative positions. A stylized feature matrix is ​​obtained by learning from financial language text based on a multilayer Transformer structure.

[0144] There may be multiple values ​​for the feature values. If all numbers are used for conversion, it is difficult to operate. Instead, the corresponding characters can be extracted according to the preset characters, such as "1" and "3". For example, if there are 3 instances of the character "1" and 2 instances of the character "3", the values ​​3 and 2 can be obtained respectively.

[0145] Then, a matrix is ​​generated using the two values ​​to obtain the numerical matrix.

[0146] Alternatively, a matrix can be generated using one of the feature values ​​to obtain a numerical matrix.

[0147] In practice, we can determine the number of values ​​contained in the feature values. If the number is large, we can extract a few characters to generate a value matrix. If the number is small, we can use all the values ​​of the feature values ​​to generate a value matrix.

[0148] In one operation, the text matrix might be a 5x5 matrix, and the numerical matrix might be a 4x4 matrix. The matrices can be manipulated to make them have the same number of bits. Then, a mapping function can be used to concatenate the text matrix and the numerical matrix, resulting in a concatenated matrix. After obtaining the concatenated matrix, it is converted into numerical data to obtain the encryption factor.

[0149] In one mode of operation, the mapping function can be a mapping function of homography transformation.

[0150] In one operation mode, a certain character length can be set to convert the concatenated matrix into numerical data. Specifically, the characters can be converted into numerical data to obtain the encryption factor.

[0151] Since the feature information can include both numerical values ​​and text, in order to fuse the numerical values ​​and text to construct an encryption factor, in another optional embodiment, the step of fusing the feature numerical values ​​and the feature text through a mapping function to obtain physical ciphertext may include the following sub-steps:

[0152] S61. Encode and encrypt the feature text to obtain ciphertext, and extract parameter values ​​from the feature values ​​according to a preset numerical interval.

[0153] S62. The parameter values ​​are mapped to characters in the text ciphertext using a mapping function to obtain the mapped ciphertext, and the mapped ciphertext is subjected to multi-dimensional feature transformation to obtain the physical ciphertext.

[0154] In one operation mode, feature text and feature values ​​can be extracted from feature information. The feature text can be the text of the feature information, and the feature values ​​can be relevant numerical values ​​of the image within the feature information.

[0155] Physical parameters can be extracted from an image containing features. In one approach, the image containing feature information can be preprocessed, including noise reduction and grayscale conversion. Then, physical parameters can be extracted from the processed image to obtain parameter values. These physical parameters may include the image's area, length, width, etc.

[0156] The ciphertext can be obtained by encoding and encrypting the feature text. In one operation, the feature text can be split into 16 blocks by bytes, and then encrypted on a 4x4 matrix. Through 10 rounds of transformation functions—including round key addition transformation, byte substitution, row shift transformation, and column hybrid transformation—multiple iterations are performed, and a subkey matrix is ​​generated by the extended key module, thus obtaining the ciphertext.

[0157] After obtaining the parameter values, it can be determined whether the values ​​are reasonable. Specifically, threshold analysis can be performed to determine whether the physical parameters exceed the threshold. The threshold can be set and divided according to the target category. If the parameter values ​​are reasonable, the physical parameters can be combined with the text ciphertext to obtain the physical ciphertext.

[0158] In one operation mode, physical parameters can be concatenated with text ciphertext to obtain physical ciphertext, or physical parameters can be added to text ciphertext to obtain physical ciphertext. The specific operation mode can be adjusted according to actual needs and is not limited here.

[0159] In one specific operation, the parameter values ​​can be mapped to characters in the ciphertext using a mapping function to obtain the mapped ciphertext. In one embodiment, the mapping function can be a combination of linear mappings. For linear mappings T and S, the matrix corresponding to their linear combination αT+βS is αA+βB.

[0160] Where A and B are the parameter value and the ciphertext, respectively, α is the composite mapping, and β corresponds to matrix multiplication, used for the concatenation of multi-step transformations.

[0161] Next, multidimensional feature transformations can be performed on the mapped ciphertext to obtain the physical ciphertext. These multidimensional feature transformations can be linear transformations and matrix operations, nonlinear feature mappings, column encryption, or double column encryption.

[0162] Linear transformations and matrix operations can be used to spatially reconstruct ciphertext features through matrix operations in linear algebra (such as rotation, scaling, and shearing). For example, the affine transformation y = Ax + b can be combined with translation and linear transformation through dimensionality increase. The orthogonality of eigenvectors of symmetric matrices can also be used for feature dimensionality reduction (such as PCA).

[0163] Nonlinear feature mapping can be exponential / logarithmic. Specifically, the feature scale can be expanded or compressed using an exponential function. For example, the probability value (0.3, 0.3, 0.4) can be mapped to (1.35, 1.35, 1.49) to enhance the difference.

[0164] Columnar encryption involves rearranging the plaintext by rows and columns and then rereading the columns in order to generate ciphertext. Double columnar encryption involves applying column transposition encryption twice consecutively to enhance security.

[0165] Finally, a pseudo-random subcipher generation algorithm can be used to automatically expand the physical ciphertext, thereby generating encryption factors.

[0166] In an optional embodiment, since users perform the same business operations, their feature information is similar or very similar, and the generated encryption factors may also be similar or the same.

[0167] To further differentiate between different encryption factors, they can be stored after generation. The newly generated encryption factor is then compared with previously generated encryption factors to determine if they are identical.

[0168] If the newly generated encryption factor is the same as the previously generated encryption factor, the user's current business processing time can be obtained. Then, the value of the current business processing time can be combined with the encryption factor to obtain a new encryption factor, thereby distinguishing the current encryption factor from the previously generated encryption factor.

[0169] Optionally, if the newly generated encryption factor is the same as a previously generated encryption factor, the generation interval of the previously generated encryption factor can also be obtained. This generation interval is the time interval between the time node that generated the encryption factor and the current time node.

[0170] If the generation interval is longer than the preset interval, it means that the interval of the previously generated encryption factors is too long. The previously generated encryption factors can be deleted. At the same time, the newly generated encryption factors are also easy to crack. They can also be regenerated. Specifically, feature information can be extracted again from the financial information. The newly extracted feature information is the one that was not extracted before. Then, the encryption factors can be constructed using the new feature information according to the above steps.

[0171] S14. Divide the financial information into multiple information blocks, encrypt the multiple information blocks using the encryption factor to obtain encrypted information, and store the encrypted information in the target database corresponding to the target category.

[0172] After obtaining the encryption factor, the financial information can be encrypted using the encryption factor to obtain encrypted information. Next, a target storage database is selected from several databases based on the target category. As explained earlier, each storage database corresponds to a type, and each type can correspond to a business type. Finally, the encrypted information can be transmitted to the target storage database for storage.

[0173] Because the stored financial information can be quite large, transmitting a large amount of data at once places high demands on communication. To improve transmission efficiency, in one operational method, the financial information can be divided into multiple information blocks during encryption, say 10 blocks. The first information block is then encrypted using an encryption factor. At this point, the target storage database can be selected from several databases based on the target category. The first encrypted information block is then transmitted to the target database. During transmission, the second information block is encrypted using the same encryption factor, and then transmitted. This process is repeated until all 10 information blocks are encrypted and transmitted.

[0174] When partitioning data, it can be done according to a preset data capacity, assuming each transmission is 10 MB. Financial information can be partitioned into 10 MB units.

[0175] After acquiring a user's financial information, this invention can detect any anomalies to avoid storing abnormal financial information. When the financial information is normal, features are extracted from the information, and encryption factors are constructed based on these features to determine the storage category. Subsequently, the encryption factors are used to encrypt the information, and the encrypted information is allocated to the corresponding storage database for storage. Encryption reduces the risk of information theft or tampering, thereby improving the security of information storage. Furthermore, by classifying and storing data, similar types of data can be stored together, breaking down data barriers and facilitating subsequent analysis and processing by different personnel.

[0176] In subsequent analysis and processing, when business personnel need to extract data from the storage database for processing, the data can only provide simple numerical values ​​and report images, which are difficult to reflect the actual situation. If a large amount of data needs to be viewed for each analysis and processing, it is not only cumbersome to operate, but also has poor data display effect. In order to facilitate business personnel to perform financial analysis more intuitively based on the data stored in the database, in an optional embodiment, after the step of storing the encrypted information in the target database corresponding to the target category, the method may further include the following steps:

[0177] S15. In response to the data display request from the business personnel, extract the data to be displayed from the target storage database according to the data display request.

[0178] S16. Use the data to be displayed and the preset digital human to generate a dynamic display animation and dynamically display the dynamic display animation.

[0179] In one embodiment, a data display request from a business user can be obtained. This request can be a request from the business user to extract and display data. Then, the data to be displayed can be retrieved from the target storage database based on the data display request.

[0180] The data to be displayed can include user transaction records, reports, and screenshots. To provide a more intuitive viewing experience for business personnel, the data can be added to a pre-defined digital persona, which can then be labeled with the data and display effects to generate a dynamic animation. Finally, the dynamic animation can be transmitted to the business personnel's smart terminals for display.

[0181] In this embodiment, the present application provides a method for storing financial data, the advantages of which are as follows: the present application can acquire financial information to be processed and perform detection processing on the financial information to obtain detection results; if the detection result is normal, feature information is extracted from the financial information based on a preset feature model, and the target category is determined in multiple preset classification profiles according to the feature information; feature text and feature values ​​are extracted from the feature information, and the feature values ​​and feature text are fused through a mapping function to obtain physical ciphertext; a pseudo-random sub-cipher generation algorithm is called to expand the physical ciphertext and bind the expanded ciphertext with the user ID to obtain an encryption factor; the financial information is divided into multiple information blocks, and the encryption factor is used to encrypt multiple information blocks to obtain encrypted information, and the encrypted information is stored in the target database corresponding to the target category. The present invention encrypts financial information through encryption factors, which can reduce the risk of information being stolen or tampered with, thereby improving the security of information storage; moreover, the present invention can store encrypted information in categories, allowing data of the same type to be stored together, breaking down data barriers, and facilitating business personnel to analyze and process data according to business types, thereby improving the efficiency of analysis and research.

[0182] As the device embodiment is basically similar to the method embodiment, the description is relatively simple, and relevant parts can be found in the description of the method embodiment.

[0183] Reference Figure 2 The diagram shows a structural block diagram of a financial data storage device according to an embodiment of this application.

[0184] Specifically, it includes:

[0185] The acquisition module 201 is used to acquire the financial information to be processed, and to perform detection processing on the financial information to obtain the detection result;

[0186] The determination module 202 is used to extract feature information from the financial information based on a preset feature model if the detection result is normal, and determine the target category in multiple preset classification profiles based on the feature information;

[0187] Encryption module 203 is used to extract feature text and feature value from the feature information, fuse the feature value and feature text through a mapping function to obtain physical ciphertext, call a pseudo-random sub-cipher generation algorithm to expand the physical ciphertext, and bind the expanded ciphertext with the user ID to obtain an encryption factor;

[0188] The storage module 204 is used to divide the financial information into multiple information blocks, encrypt the multiple information blocks using the encryption factor to obtain encrypted information, and store the encrypted information in the target database corresponding to the target category.

[0189] Optionally, the step of acquiring the financial information to be processed and performing detection processing on the financial information to obtain the detection result includes:

[0190] Obtain the financial information to be processed, and extract the first financial image and the first financial text from the financial information respectively;

[0191] Extract key financial values ​​from the first financial text, and construct an analog financial statement using the key financial values ​​and a preset report template. Extract the image corresponding to the analog financial statement to obtain an analog financial image.

[0192] Calculate the image similarity between the first financial image and the analogous financial image;

[0193] If the image similarity is greater than or equal to a preset first similarity, then the processing result of the detection process is determined to be normal.

[0194] If the image similarity is less than a preset first similarity, then the processing result of the detection process is determined to be abnormal.

[0195] Optionally, the step of acquiring the financial information to be processed and performing detection processing on the financial information to obtain the detection result includes:

[0196] Obtain the financial information to be processed, and extract the second financial image and the second financial text from the financial information respectively;

[0197] Identify the transaction marker region of the second financial image and extract the marker report text corresponding to the transaction marker region;

[0198] The data of the marked report text is split to obtain a set of marked information, and the data of the second financial text is split to obtain a set of financial information;

[0199] Calculate text similarity based on the set of labeled information and the set of financial information;

[0200] If the text similarity is greater than or equal to a preset second similarity, then the processing result of the detection process is determined to be normal.

[0201] If the text similarity is less than a preset second similarity, then the processing result of the detection process is determined to be abnormal.

[0202] Optionally, the step of extracting feature information from the financial information based on a preset feature model and determining the target category from multiple preset classification profiles based on the feature information includes:

[0203] Feature information is extracted from the financial information based on a preset feature model, and a set of profile tags corresponding to each preset category profile is obtained, wherein the set of profile tags contains multiple tags;

[0204] The semantic matching degree is calculated by using the feature information and the tags of each of the portrait tag sets to obtain multiple semantic matching degrees;

[0205] The maximum semantic matching score is selected from multiple semantic matching scores, and the category of the preset classification profile corresponding to the maximum semantic matching score is determined to obtain the target category.

[0206] Optionally, the step of fusing the feature values ​​and the feature text through a mapping function to obtain the physical ciphertext includes:

[0207] The feature values ​​and feature text are converted into matrices respectively to obtain a text matrix and a value matrix;

[0208] The concatenated matrix is ​​obtained by concatenating the text matrix and the numerical matrix using a mapping function, and then the concatenated matrix is ​​converted into numerical data to obtain the physical ciphertext.

[0209] Optionally, the step of fusing the feature values ​​and the feature text through a mapping function to obtain the physical ciphertext includes:

[0210] The feature text is then encoded and encrypted to obtain ciphertext, and parameter values ​​are extracted from the feature text at preset numerical intervals.

[0211] The parameter values ​​are mapped to characters in the ciphertext using a mapping function to obtain the mapped ciphertext. The mapped ciphertext is then subjected to multidimensional feature transformation to obtain the physical ciphertext.

[0212] As the system implementation is basically similar to the method implementation, it is described in a relatively simple way. For relevant details, please refer to the description of the method implementation.

[0213] Reference Figure 3 The diagram shows a structural block diagram of a financial data storage system provided in an embodiment of this application.

[0214] Specifically, it includes: a processing terminal and multiple storage databases, each of which corresponds to a category of a categorized profile;

[0215] The processing terminal is communicatively connected to multiple storage databases;

[0216] The processing terminal executes the financial data storage method as described in the above embodiments.

[0217] Reference Figure 4 The computer device illustrating a method for storing financial data according to this application may specifically include the following:

[0218] The computer device 12 described above is in the form of a general-purpose computing device. The components of the computer device 12 may include, but are not limited to: one or more processors or processing units 16, system memory 28, and bus 18 connecting different system components (including system memory 28 and processing unit 16).

[0219] Bus 18 refers to one or more of several types of bus 18 architectures, including memory bus 18 or memory controller, peripheral bus 18, graphics acceleration port, processor, or local bus 18 using any of the various bus 18 architectures. For example, these architectures include, but are not limited to, Industry Standard Architecture (ISA) bus 18, Micro Channel Architecture (MAC) bus 18, Enhanced ISA bus 18, Audio / Video Electronics Standards Association (VESA) local bus 18, and Peripheral Component Interconnect (PCI) bus 18.

[0220] Computer device 12 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by computer device 12, including volatile and non-volatile media, removable and non-removable media.

[0221] System memory 28 may include computer system readable media in the form of volatile memory, such as random access memory (RAM) 30 and / or cache memory 32. Computer device 12 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, storage system 34 may be used to read and write non-removable, non-volatile magnetic media (commonly referred to as a "hard disk drive"). Figure 4 As not shown, a disk drive for reading and writing to a removable non-volatile disk (such as a "floppy disk") and an optical disk drive for reading and writing to a removable non-volatile optical disk (such as a CD-ROM, DVD-ROM, or other optical media) may be provided. In these cases, each drive may be connected to bus 18 via one or more data media interfaces. The memory may include at least one program product having a set (e.g., at least one) of program modules 42 configured to perform the functions of the embodiments of this application.

[0222] A program / utility 40 having a set (at least one) of program modules 42 may be stored, for example, in memory. Such program modules 42 include—but are not limited to—an operating system, one or more application programs, other program modules 42, and program data. Each or some combination of these examples may include an implementation of a network environment. Program modules 42 typically perform the functions and / or methods described in the embodiments of this application.

[0223] Computer device 12 can also communicate with one or more external devices 14 (e.g., keyboard, pointing device, display 24, camera, etc.), and with one or more devices that enable a user to interact with the computer device 12, and / or with any device that enables the computer device 12 to communicate with one or more other computing devices (e.g., network card, modem, etc.). This communication can be performed through input / output (I / O) interface 22. Furthermore, computer device 12 can also communicate with one or more networks (e.g., local area network (LAN)), wide area network (WAN), and / or public networks (e.g., the Internet) via network adapter 20. As shown, network adapter 20 communicates with other modules of computer device 12 via bus 18. It should be understood that, although... Figure 4 Not shown, it can be combined with computer device 12 to use other hardware and / or software modules, including but not limited to: microcode, device drivers, redundant processing unit 16, external disk drive array, RAID system, tape drive and data backup storage system 34, etc.

[0224] The processing unit 16 executes various functional applications and data processing by running programs stored in the system memory 28, such as implementing the financial data storage method provided in the embodiments of this application.

[0225] That is, when the processing unit 16 executes the above program, it achieves the following:

[0226] Obtain the financial information to be processed, and perform detection processing on the financial information to obtain the detection result;

[0227] If the detection result is normal, feature information is extracted from the financial information based on the preset feature model, and the target category is determined in multiple preset classification profiles based on the feature information;

[0228] Feature text and feature values ​​are extracted from the feature information. The feature values ​​and feature text are fused through a mapping function to obtain physical ciphertext. A pseudo-random sub-cipher generation algorithm is called to expand the physical ciphertext and bind the expanded ciphertext with the user ID to obtain the encryption factor.

[0229] The financial information is divided into multiple information blocks, and the multiple information blocks are encrypted using the encryption factor to obtain encrypted information. The encrypted information is then stored in the target database corresponding to the target category.

[0230] In this application embodiment, the application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method for storing financial data as provided in all embodiments of the application.

[0231] That is, to implement the following when the program is executed by the processor:

[0232] Obtain the financial information to be processed, and perform detection processing on the financial information to obtain the detection result;

[0233] If the detection result is normal, feature information is extracted from the financial information based on the preset feature model, and the target category is determined in multiple preset classification profiles based on the feature information;

[0234] Feature text and feature values ​​are extracted from the feature information. The feature values ​​and feature text are fused through a mapping function to obtain physical ciphertext. A pseudo-random sub-cipher generation algorithm is called to expand the physical ciphertext and bind the expanded ciphertext with the user ID to obtain the encryption factor.

[0235] The financial information is divided into multiple information blocks, and the multiple information blocks are encrypted using the encryption factor to obtain encrypted information. The encrypted information is then stored in the target database corresponding to the target category.

[0236] Any combination of one or more computer-readable media may be used. A computer-readable medium may be a computer-to-signal medium or a computer-readable storage medium. A computer-readable storage medium may be, for example—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium may be any tangible medium that contains or stores a program that may be used by or in connection with an instruction execution system, apparatus, or device.

[0237] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including—but not limited to—electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, capable of transmitting, propagating, or transmitting programs for use by or in connection with an instruction execution system, apparatus, or device.

[0238] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof. These programming languages ​​include object-oriented programming languages—such as Java, Smalltalk, and C++—and conventional procedural programming languages—such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider). The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably.

[0239] Although preferred embodiments of the present application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of the present application.

[0240] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes said element.

[0241] The foregoing has provided a detailed description of a method, apparatus, system, device, and medium for storing financial data provided in this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method of storing financial data, characterized by, The method includes: Obtain the financial information to be processed, and perform detection processing on the financial information to obtain the detection result; If the detection result is normal, feature information is extracted from the financial information based on the preset feature model, and the target category is determined in multiple preset classification profiles based on the feature information; Feature text and feature values ​​are extracted from the feature information. The feature values ​​and feature text are fused through a mapping function to obtain physical ciphertext. A pseudo-random sub-cipher generation algorithm is called to expand the physical ciphertext and bind the expanded ciphertext with the user ID to obtain the encryption factor. The financial information is divided into multiple information blocks, and the multiple information blocks are encrypted using the encryption factor to obtain encrypted information, which is then stored in the target database corresponding to the target category. The process of fusing the feature values ​​and the feature text using a mapping function to obtain the physical ciphertext includes: The feature values ​​and feature text are converted into matrices respectively to obtain a text matrix and a value matrix; The concatenated matrix is ​​obtained by concatenating the text matrix and the numerical matrix using a mapping function, and then the concatenated matrix is ​​converted into numerical data to obtain the physical ciphertext. or; The process of fusing the feature values ​​and the feature text using a mapping function to obtain the physical ciphertext includes: The feature text is then encoded and encrypted to obtain ciphertext, and parameter values ​​are extracted from the feature text at preset numerical intervals. The parameter values ​​are mapped to characters in the ciphertext using a mapping function to obtain the mapped ciphertext. The mapped ciphertext is then subjected to multidimensional feature transformation to obtain the physical ciphertext.

2. The method of claim 1, wherein, The process of acquiring the financial information to be processed and performing detection processing on the financial information to obtain the detection result includes: Obtain the financial information to be processed, and extract the first financial image and the first financial text from the financial information respectively; Extract key financial values ​​from the first financial text, and construct an analog financial statement using the key financial values ​​and a preset report template. Extract the image corresponding to the analog financial statement to obtain an analog financial image. Calculate the image similarity between the first financial image and the analogous financial image; If the image similarity is greater than or equal to a preset first similarity, then the processing result of the detection process is determined to be normal. If the image similarity is less than a preset first similarity, then the processing result of the detection process is determined to be abnormal.

3. The method of claim 1, wherein, The process of acquiring the financial information to be processed and performing detection processing on the financial information to obtain the detection result includes: Obtain the financial information to be processed, and extract the second financial image and the second financial text from the financial information respectively; Identify the transaction marker region of the second financial image and extract the marker report text corresponding to the transaction marker region; The data of the marked report text is split to obtain a set of marked information, and the data of the second financial text is split to obtain a set of financial information; Calculate text similarity based on the set of labeled information and the set of financial information; If the text similarity is greater than or equal to a preset second similarity, then the processing result of the detection process is determined to be normal. If the text similarity is less than a preset second similarity, it is determined that the processing result of the detection processing is abnormal.

4. The method of claim 1, wherein, The feature information is extracted from the financial information based on a preset feature model, and a target category is determined in a plurality of preset classification portraits according to the feature information, including: The feature information is extracted from the financial information based on a preset feature model, and a portrait label set corresponding to each preset classification portrait is obtained, the portrait label set containing a plurality of labels; The semantic matching degrees are calculated by using the feature information and the labels of each portrait label set, and a plurality of semantic matching degrees are obtained. The maximum semantic matching degree is selected from the plurality of semantic matching degrees, and the category of the preset classification portrait corresponding to the maximum semantic matching degree is determined to obtain the target category.

5. A storage device for financial data, characterized by The device comprises: An acquisition module is configured to acquire financial information to be processed, and perform detection processing on the financial information to obtain a detection result. A determination module is configured to, if the detection result is normal, extract feature information from the financial information based on a preset feature model, and determine a target category in a plurality of preset classification portraits according to the feature information. An encryption module is configured to extract feature text and feature values from the feature information, fuse the feature values and the feature text by a mapping function to obtain physical ciphertext, call a pseudo-random sub-cipher generation algorithm to expand the physical ciphertext, bind the expanded ciphertext with a user ID, and obtain an encryption factor. A storage module is configured to divide the financial information into a plurality of information blocks, encrypt the plurality of information blocks by using the encryption factor to obtain encrypted information, and store the encrypted information in a target database corresponding to the target category. The fusion of the feature values and the feature text by the mapping function to obtain the physical ciphertext comprises: The feature values and the feature text are respectively converted into matrices to obtain a text matrix and a value matrix; The text matrix and the value matrix are spliced by a mapping function to obtain a spliced matrix, and the spliced matrix is converted into numerical data to obtain the physical ciphertext. Alternatively; The fusion of the feature values and the feature text by the mapping function to obtain the physical ciphertext comprises: The feature text is subjected to secondary encoding encryption to obtain text ciphertext, and a parameter value is extracted from the feature values according to a preset value interval; The parameter value is mapped to a character of the text ciphertext by a mapping function to obtain mapping ciphertext, and the mapping ciphertext is subjected to multi-dimensional feature conversion to obtain the physical ciphertext.

6. A system for storing financial data, characterized by The system comprises a processing terminal and a plurality of storage databases, each of the storage databases corresponding to a category of a classification portrait. The processing terminal is in communication connection with the plurality of storage databases. The processing terminal executes the storage method of the financial data according to any one of claims 1-4.

7. An electronic device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the storage method of the financial data according to any one of claims 1-4 when executing the computer program. A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the storage method of the financial data according to any one of claims 1-4 when executing the computer program.

8. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a computer executable program for causing a computer to execute the financial data storage method according to any one of claims 1-4.

Images