Data security classification method and data security management system based on deep learning

Abstract

The invention belongs to the technical field of data processing, and particularly provides a deep learning-based data security classification method, which comprises the following steps of: receiving data information from a user side in real time; preprocessing the image data and the text data; inputting the preprocessed data information into a trained deep learning model, performing data classification on the data information through the trained deep learning model, and outputting a classification result through the deep learning model; obtaining a security level corresponding to the data according to a classification result; and encrypting the classified data according to a preset encryption method of each security level. The deep learning model is adopted to perform data classification on the data information, and the encryption algorithm is adopted to encrypt the classified data according to the security level to which the classified data belong, so that the problems existing in traditional data classification are solved. The accuracy and efficiency of data classification are improved by combining a deep learning technology with encryption methods of different security levels, and the security of the data is effectively guaranteed.

Description

Technical Field

[0001] The present invention belongs to the field of data processing technology, and specifically relates to a data security classification method and a data security management system based on deep learning. Background Art

[0002] With the acceleration of digital transformation, data has become an important force driving social development. However, data security issues are becoming increasingly serious, especially in corporate environments, where risks such as data leakage, tampering, and abuse are increasing. In order to ensure data security, companies and individuals need to classify and encrypt data.

[0003] At present, most traditional data classification relies on manual work. With the diversification of data types, traditional manual data classification has problems such as time-consuming, inefficient, low accuracy, low data protection security, or inability to protect some confidential data well. These problems become more and more obvious when processing large amounts of data.

[0004] Based on this, the present invention provides a data security classification method and a data security management system based on deep learning to solve the above problems. Summary of the invention

[0005] In order to overcome the shortcomings of the prior art, the present invention provides a data security classification method and a data security management system based on deep learning, so as to solve the problems of manual data classification in the prior art being time-consuming, error-prone and low in security.

[0006] One embodiment of the present invention provides a data security classification method based on deep learning, comprising the following steps:

[0007] Receiving data information from a user terminal in real time, the data information including image data and text data;

[0008] Preprocessing the image data and text data;

[0009] Inputting the preprocessed data information into a trained deep learning model, classifying the data information through the trained deep learning model, and outputting the classification result through the deep learning model;

[0010] Obtaining a security level corresponding to the data according to the classification result, wherein the security level includes a public level, an internal level, and a confidential level;

[0011] The classified data is encrypted according to the encryption method preset for each security level, with the internal level using a first encryption algorithm and the confidential level using a second encryption algorithm.

[0012] In one embodiment, inputting the pre-processed data information into a trained deep learning model includes:

[0013] Acquire the data information, and set a target keyword group according to the content in the data information;

[0014] When a first target keyword group appears in the data information, dividing the content related to the first target keyword group into an internal level training set;

[0015] When a second target keyword group appears in the data information, the content related to the second target keyword group is divided into a confidential level training set;

[0016] The deep learning model is trained using the internal-level training set and the confidential-level training set respectively to obtain a trained deep learning model.

[0017] In one embodiment, the preprocessing of the image data and text data further includes:

[0018] Constructing a target word library, wherein the target word library contains key words related to all categories in the security level;

[0019] Performing target word detection on the preprocessed image data and the text data based on the target word library;

[0020] According to the target word detection result, feature annotation is performed on the data in the image data and the text data to obtain target annotation features.

[0021] In one embodiment, the deep learning model includes a convolutional neural network, a recurrent neural network and a fully connected layer; the preprocessed data information is input into the trained deep learning model, the data information is classified by the trained deep learning model, and the classification result is output by the deep learning model, including:

[0022] Acquire an image feature vector in the image data through the convolutional neural network, and acquire a text feature vector in the text data through the recurrent neural network;

[0023] The image feature vector is merged with the text feature vector to obtain a joint feature vector, wherein the joint feature vector contains target annotation features;

[0024] Inputting the joint feature vector into the fully connected layer, each output node of the fully connected layer outputs a classification result based on the input joint feature vector;

[0025] According to the classification results output by each output node, an activation function is used to convert the classification results into distribution values;

[0026] The category corresponding to the data information is output based on the distribution value to complete data classification.

[0027] In one embodiment, the first encryption algorithm is a symmetric encryption algorithm; the classified data is encrypted according to the encryption method preset for each security level, including:

[0028] When the received data information belongs to the internal level, the data information is encrypted by a symmetric encryption algorithm to obtain internal level encrypted data, which specifically includes the following steps:

[0029] When the received data information belongs to the internal level, a random encryption key is generated, and an encryption mode and a random initialization vector are determined;

[0030] encrypting the internal level data according to the encryption mode and the random encryption key, and completing encryption of the internal level data based on the encryption result;

[0031] The random encryption key and the random initialization vector are sent to a management terminal for storage to facilitate verification and decryption of internal level data.

[0032] In one embodiment, the second encryption algorithm includes a symmetric encryption algorithm, an asymmetric encryption algorithm and a hash algorithm; the encryption method preset for each security level, encrypting the classified data, includes:

[0033] When the received data information is classified as confidential, the data information is encrypted by a symmetric encryption algorithm, an asymmetric encryption algorithm, and a hash algorithm to obtain confidential encrypted data, which specifically includes the following steps:

[0034] When the received data information is classified as confidential, a random encryption key is generated, and an encryption mode and a random initialization vector are determined;

[0035] Generate a public key and a private key pair based on the random encryption key, and encrypt the random encryption key using the public key;

[0036] Encrypting the confidential-level data according to the encryption mode and the encrypted random encryption key, and completing the encryption of the confidential-level data based on the encryption result;

[0037] Use the hash function to calculate the encrypted confidential data to obtain the corresponding hash value;

[0038] The random encryption key, the random initialization vector, the private key and the hash value are sent to a management terminal for storage to facilitate verification and decryption of confidential data.

[0039] The present application also relates to a data security management system based on deep learning, comprising:

[0040] A data receiving module, used for receiving data information from a user terminal in real time;

[0041] A data preprocessing module, used for preprocessing image data and text data;

[0042] A data classification module, used to input the preprocessed data information into a trained deep learning model, classify the data information through the trained deep learning model, and output the classification result through the deep learning model;

[0043] A data level analysis module, used to obtain the security level corresponding to the data according to the classification result;

[0044] The data encryption module is used to encrypt the classified data according to the encryption method preset for each security level.

[0045] The present application also relates to a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned deep learning-based data security classification method when executing the computer program.

[0046] The present application also relates to a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the steps of the above-mentioned data security classification method based on deep learning are implemented.

[0047] The data security classification method and data security management system based on deep learning provided in the above embodiments have the following beneficial effects:

[0048] 1. By using a deep learning model to classify data information and using an encryption algorithm to encrypt the classified data according to the security level it belongs to, the problems of low efficiency and low accuracy of traditional data classification methods are solved. Through deep learning technology, the accuracy and efficiency of data classification are improved, and at the same time, combined with encryption methods of different security levels, the security of data is effectively guaranteed.

[0049] 2. In one of the embodiments, a target word library is constructed to detect target words in data information and perform feature annotation to obtain target annotation features, and the accuracy of the deep learning model outputting classification results is improved based on the target annotation features. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the structures shown in these drawings without paying creative work.

[0051] Figure 1 A flowchart of a data security classification method based on deep learning provided in an embodiment of the present invention.

[0052] Figure 2 A functional block diagram of a computer device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0053] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0054] It should be noted that if there are descriptions involving "first", "second", etc. in the embodiments of the present invention, the descriptions of "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of technical features indicated. Therefore, the features limited to "first" and "second" may explicitly or implicitly include at least one of the features. In addition, if "and / or" or "and / or" appears in the full text, its meaning includes three parallel schemes. Taking "A and / or B" as an example, it includes scheme A, or scheme B, or a scheme that satisfies both A and B. In addition, the technical solutions between the various embodiments can be combined with each other, but it must be based on the ability of ordinary technicians in this field to implement. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0055] Reference Figure 1 One embodiment of the present invention provides a data security classification method based on deep learning, comprising the following steps:

[0056] S100, receiving data information from a user terminal in real time, wherein the data information includes image data and text data;

[0057] In this embodiment, the data information at the user end is the internal data information of the enterprise, and the image data information and the text data information in the internal data information of the enterprise are received in real time.

[0058] S200, preprocessing the image data and text data;

[0059] In this embodiment, the image data and text data are preprocessed to optimize the data and remove useless information in the data so as to adapt them for input into a deep learning model and improve the accuracy of data classification.

[0060] S300, inputting the preprocessed data information into a trained deep learning model, classifying the data information through the trained deep learning model, and outputting the classification result through the deep learning model;

[0061] In this embodiment, the pre-processed image data and text data are input into the trained model. The trained model can classify the image data and text data according to preset classification standards, and the classified data is output as a classification result. According to the classification result, it can be known what category the data specifically belongs to, so as to facilitate the adoption of corresponding encryption measures to encrypt the data of that category and achieve the effect of data security classification.

[0062] S400, obtaining a security level corresponding to the data according to the classification result, wherein the security level includes a public level, an internal level and a confidential level;

[0063] In this embodiment, the security level is preset as public level, internal level and confidential level. After the deep learning model classifies the data information and outputs the classification results, the output classification results are divided into the corresponding security levels according to the preset security levels. For example, if the data information is information disclosed to the public, such as corporate publicity information, the data information is classified as public level; if the data information is information circulated within the enterprise, such as employee information and internal reports, the data information is classified as internal level; if the data information is financial data, research and development drawings and other highly confidential information within the enterprise, the data information is classified as confidential level.

[0064] S500. Encrypt the classified data according to the encryption method preset for each security level, wherein the internal level adopts a first encryption algorithm and the confidential level adopts a second encryption algorithm.

[0065] In this embodiment, different encryption methods are used to encrypt data information of different security levels, so that different data have different security, effectively ensuring the security of the data. For data information classified by the deep learning model and determining the security level, the data information at the internal level is encrypted using the first encryption algorithm, the data information at the confidential level is encrypted using the second encryption algorithm, and the data information at the public level does not need to be encrypted, thereby reducing unnecessary data encryption.

[0066] As described in the above steps, by using a deep learning model to classify data information, and using an encryption algorithm to encrypt the classified data according to the security level it belongs to, the problems of low efficiency and low accuracy of traditional data classification methods are solved. Through deep learning technology, the accuracy and efficiency of data classification are improved, and at the same time, combined with encryption methods of different security levels, the security of data is effectively guaranteed.

[0067] In one embodiment, the training method of the deep learning model in step S300 is:

[0068] S301, acquiring the data information, and setting a target keyword group according to the content in the data information;

[0069] S302, when a first target keyword group appears in the data information, dividing the content related to the first target keyword group into an internal level training set;

[0070] S303, when a second target keyword group appears in the data information, dividing the content related to the second target keyword group into a confidential level training set;

[0071] S304: Use the internal-level training set and the confidential-level training set to train the deep learning model respectively to obtain a trained deep learning model.

[0072] In this embodiment, as described in the above steps S301-S304, the pre-processed data information is obtained, and the target keyword group is set for the content in the data information, and the target keyword group includes a first target keyword group and a second target keyword group; for example, the first target keyword group includes relevant words such as "internal", and the second target keyword group includes relevant words such as "confidential", and the specific target keyword group is set according to the use requirements. When the first target keyword group is identified in the acquired data information, the data information is divided into an internal level training set, and when the second target keyword group is identified in the acquired data information, the data information is divided into a confidential level training set. The deep learning model is trained respectively by the obtained internal level training set and the confidential level training set, so as to obtain a trained deep learning model. The data information is classified by the trained deep learning model to improve the accuracy and efficiency of data classification.

[0073] In one embodiment, the step S200 further includes:

[0074] S210, constructing a target word library, wherein the target word library contains key words related to all categories in the security level;

[0075] S220, performing target word detection on the preprocessed image data and the text data based on the target word library;

[0076] S230 . Perform feature annotation on the image data and the data in the text data according to the target word detection result to obtain target annotation features.

[0077] In this embodiment, as described in the above steps S210-S230, a target word library is constructed, and the target word library contains key words related to all security levels (public level, internal level, confidential level). After preprocessing the image data and text data, the target detection algorithm is used to detect the target words of the preprocessed image data and text data, so that some keywords and key words in the data information can be feature-labeled according to the detection results to obtain a data information containing target labeling features. Since the data information has been preprocessed, target word detected, and feature-labeled, when it is input into the deep learning model for classification, the deep learning model can quickly determine whether the target labeling features in the data information belong to the internal level training set or the confidential level training set, thereby improving the accuracy and efficiency of the classification results output when the data information is input into the deep learning model for data classification.

[0078] In one embodiment, the deep learning model includes a convolutional neural network, a recurrent neural network and a fully connected layer; the step S300 specifically includes:

[0079] S310, obtaining an image feature vector in the image data through the convolutional neural network, and obtaining a text feature vector in the text data through the recurrent neural network;

[0080] S320, fusing the image feature vector with the text feature vector to obtain a joint feature vector, wherein the joint feature vector contains target annotation features;

[0081] S330, inputting the joint feature vector into the fully connected layer, and each output node of the fully connected layer outputs a classification result based on the input joint feature vector;

[0082] S340, according to the classification results output by each output node, respectively converting the classification results into distribution values ​​using an activation function;

[0083] S350: Output the category corresponding to the data information based on the distribution value to complete data classification.

[0084] In this embodiment, the deep learning model includes a convolutional neural network, a recurrent neural network and a fully connected layer. As described in the above steps S310-S350, when the pre-processed image data and text data are input into the deep learning model, the image feature vector in the image data is extracted by the convolutional neural network, and the text feature vector in the text data is extracted by the recurrent neural network. Then, the extracted image feature vector and the text feature vector are fused to obtain a joint feature vector, and the obtained joint feature vector is connected to a fully connected layer, and the fully connected layer contains the same number of output nodes as the security level. The joint features are connected and input into the fully connected layer, so that each output node of the fully connected layer outputs the classification result; because the joint feature vector contains the target annotation feature, the accuracy of the data category to which the output classification result corresponds is improved. After the fully connected layer outputs the classification result, the softmax activation function is applied to convert these output results into distribution values ​​to determine the final classification. When determining the final classification of the data, the category corresponding to the maximum value of the distribution value converted by the output node of the fully connected layer is selected as the classification result of the deep learning model. For example, A is taken as the public level, B as the internal level, and C as the confidential level, and the output vector is [A0.1, B0.6, C0.3]. The data information predicted by the deep learning model belongs to the internal level. Since B has the highest probability, the data information is classified as internal-level data information, thereby completing the data classification.

[0085] In one embodiment, the first encryption algorithm is a symmetric encryption algorithm; the classified data is encrypted according to the encryption method preset for each security level, including:

[0086] S510: When the received data information belongs to the internal level, the data information is encrypted by a symmetric encryption algorithm to obtain internal level encrypted data. Step S510 specifically includes the following steps:

[0087] S511, when the received data information belongs to the internal level, generate a random encryption key, and determine the encryption mode and the random initialization vector;

[0088] S512, encrypting the internal-level data according to the encryption mode and the random encryption key, and completing encryption of the internal-level data based on the encryption result;

[0089] S513: Send the random encryption key and the random initialization vector to a management terminal for storage to facilitate verification and decryption of internal-level data.

[0090] In this embodiment, the first encryption algorithm uses the AES-256 symmetric encryption algorithm. When the data information is determined to be internal-level data information in the classification result output in the deep learning model, the AES-256 symmetric encryption algorithm is used to encrypt the data information to obtain internal-level encrypted data. The specific encryption method is as described in the above steps S510-S513. When it is determined that the received data information belongs to the internal level, a 256-bit symmetric encryption key is randomly generated, and the CBC (Cipher Block Chaining) encryption mode and the corresponding initial vector (IV) are selected to ensure that even if the same data is encrypted multiple times, different ciphertexts will be generated. The IV is randomly generated and is unique for each encryption. The internal-level data is encrypted according to the determined encryption mode, the generated 256-bit random encryption key and the corresponding initial vector (IV), and the plaintext data is converted into ciphertext data to complete the encryption of the internal-level data. The encrypted ciphertext data, encryption key, and initial vector are sent to the management end for storage to facilitate the verification and decryption of the internal-level data.

[0091] In one embodiment, the second encryption algorithm includes a symmetric encryption algorithm, an asymmetric encryption algorithm and a hash algorithm; the encryption method preset for each security level, encrypting the classified data, includes:

[0092] S520: When the received data information is of a confidential level, the data information is encrypted by using a symmetric encryption algorithm, an asymmetric encryption algorithm, and a hash algorithm to obtain confidential level encrypted data.

[0093] Step S520 specifically includes the following steps:

[0094] S521. When the received data information is classified as confidential, a random encryption key is generated, and an encryption mode and a random initialization vector are determined;

[0095] S522, generating a public key and a private key pair based on the random encryption key, and encrypting the random encryption key using the public key;

[0096] S523, encrypting the confidential-level data according to the encryption mode and the encrypted random encryption key, and completing the encryption of the confidential-level data based on the encryption result;

[0097] S524, using a hash function to calculate the encrypted confidential data to obtain a corresponding hash value;

[0098] S525: Send the random encryption key, the random initialization vector, the private key, and the hash value to a management terminal for storage to facilitate verification and decryption of confidential data.

[0099] In this embodiment, the second encryption algorithm uses the AES-512 symmetric encryption algorithm, the RAS-4096 asymmetric encryption algorithm, and the SHA-512 hash algorithm. When the data information is determined to be confidential data information in the classification result output from the deep learning model, the AES-512 symmetric encryption algorithm, the RAS-4096 asymmetric encryption algorithm, and the SHA-512 hash algorithm are combined to encrypt the data information to obtain confidential encrypted data. The specific encryption method is as described in the above steps S520-S525. When it is determined that the received data information belongs to the confidential level, a 512-bit symmetric encryption key is randomly generated, and the CBC (CipherBlock Chaining) encryption mode and the corresponding initial vector (IV) are selected; based on the randomly generated 512-bit symmetric encryption key, an RSA-4096-bit public key and private key pair are generated, and the public key is used to encrypt the randomly generated 512-bit symmetric encryption key, and the private key is used for decryption. After the symmetric encryption key completes the encryption, the confidential data is encrypted based on the selected encryption mode and the encrypted symmetric encryption key, the plaintext data is converted into ciphertext data, and the encryption of the confidential data is completed based on the encryption result. After the encryption of the confidential data is completed, the encrypted confidential data is calculated using a hash function to obtain the corresponding hash value to ensure the integrity of the encrypted data. The encrypted ciphertext data, encryption key, initial vector, private key, and hash value are sent to the management end for storage to facilitate the verification and decryption of the confidential data. Since the randomly generated 512-bit symmetric encryption key itself has been encrypted by the RSA-4096-bit public key, even if the ciphertext data is intercepted, the data cannot be decrypted without the RSA-4096-bit private key. The generated hash value is used to verify whether the encrypted data has been tampered with after decryption.

[0100] In one embodiment, a data security management system based on deep learning is provided, and the data security management system based on deep learning corresponds to the data security classification method based on deep learning in the above embodiment. The data security management system based on deep learning includes:

[0101] A data receiving module, used for receiving data information from a user terminal in real time;

[0102] A data preprocessing module, used for preprocessing image data and text data;

[0103] A data classification module, used to input the preprocessed data information into a trained deep learning model, classify the data information through the trained deep learning model, and output the classification result through the deep learning model;

[0104] A data level analysis module, used to obtain the security level corresponding to the data according to the classification result;

[0105] The data encryption module is used to encrypt the classified data according to the encryption method preset for each security level.

[0106] Optionally, also include:

[0107] The target word detection module is used to perform target word detection on the preprocessed image data and text data based on the target word library.

[0108] Optionally, also include:

[0109] The data storage module is used to store the encrypted ciphertext data, encryption keys, etc.

[0110] For the specific definition of a data security management system based on deep learning, please refer to the definition of a data security classification method based on deep learning above, which will not be repeated here. Each module in the above-mentioned data security management system based on deep learning can be implemented in whole or in part through software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0111] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 2As shown. The computer device includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store data information, encryption keys, encryption information, etc. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a data security classification method based on deep learning is implemented.

[0112] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, a data security classification method based on deep learning is implemented.

[0113] In one of the embodiments, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, a data security classification method based on deep learning is implemented.

[0114] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0115] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0116] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A data security classification method based on deep learning, characterized in that: The steps include: Receiving data information from a user terminal in real time, the data information including image data and text data; Preprocessing the image data and text data; Inputting the preprocessed data information into a trained deep learning model, classifying the data information through the trained deep learning model, and outputting the classification result through the deep learning model; Obtaining a security level corresponding to the data according to the classification result, wherein the security level includes a public level, an internal level, and a confidential level; The classified data is encrypted according to the encryption method preset for each security level, with the internal level using a first encryption algorithm and the confidential level using a second encryption algorithm.

2. The data security classification method based on deep learning according to claim 1, characterized in that: The step of inputting the pre-processed data information into a trained deep learning model comprises: Acquire the data information, and set a target keyword group according to the content in the data information; When a first target keyword group appears in the data information, dividing the content related to the first target keyword group into an internal level training set; When a second target keyword group appears in the data information, the content related to the second target keyword group is divided into a confidential level training set; The deep learning model is trained using the internal-level training set and the confidential-level training set respectively to obtain a trained deep learning model.

3. The data security classification method based on deep learning as claimed in claim 2, characterized in that: The preprocessing of the image data and the text data further includes: Constructing a target word library, wherein the target word library contains key words related to all categories in the security level; Performing target word detection on the preprocessed image data and the text data based on the target word library; According to the target word detection result, feature annotation is performed on the data in the image data and the text data to obtain target annotation features.

4. The data security classification method based on deep learning as claimed in claim 3, characterized in that: The deep learning model includes a convolutional neural network, a recurrent neural network and a fully connected layer; the pre-processed data information is input into the trained deep learning model, the data information is classified by the trained deep learning model, and the classification result is output by the deep learning model, including: Acquire an image feature vector in the image data through the convolutional neural network, and acquire a text feature vector in the text data through the recurrent neural network; The image feature vector is merged with the text feature vector to obtain a joint feature vector, wherein the joint feature vector contains target annotation features; Inputting the joint feature vector into the fully connected layer, each output node of the fully connected layer outputs a classification result based on the input joint feature vector; According to the classification results output by each output node, an activation function is used to convert the classification results into distribution values; The category corresponding to the data information is output based on the distribution value to complete data classification.

5. The data security classification method based on deep learning as claimed in claim 4, characterized in that: The first encryption algorithm is a symmetric encryption algorithm; The encrypting of the classified data according to the encryption method preset for each security level includes: When the received data information belongs to the internal level, the data information is encrypted by a symmetric encryption algorithm to obtain internal level encrypted data, which specifically includes the following steps: When the received data information belongs to the internal level, a random encryption key is generated, and an encryption mode and a random initialization vector are determined; encrypting the internal level data according to the encryption mode and the random encryption key, and completing encryption of the internal level data based on the encryption result; The random encryption key and the random initialization vector are sent to a management terminal for storage to facilitate verification and decryption of internal level data.

6. The data security classification method based on deep learning as claimed in claim 4, characterized in that: The second encryption algorithm includes a symmetric encryption algorithm, an asymmetric encryption algorithm and a hash algorithm; The encrypting of the classified data according to the encryption method preset for each security level includes: When the received data information is classified as confidential, the data information is encrypted by a symmetric encryption algorithm, an asymmetric encryption algorithm, and a hash algorithm to obtain confidential encrypted data, which specifically includes the following steps: When the received data information is classified as confidential, a random encryption key is generated, and an encryption mode and a random initialization vector are determined; Generate a public key and a private key pair based on the random encryption key, and encrypt the random encryption key using the public key; Encrypting the confidential-level data according to the encryption mode and the encrypted random encryption key, and completing the encryption of the confidential-level data based on the encryption result; Use the hash function to calculate the encrypted confidential data to obtain the corresponding hash value; The random encryption key, the random initialization vector, the private key and the hash value are sent to a management terminal for storage to facilitate verification and decryption of confidential data.

7. A data security management system based on deep learning, used to implement the steps of a data security classification method based on deep learning as described in any one of claims 1 to 6, characterized in that: include: A data receiving module is used to receive data information from a user terminal in real time; A data preprocessing module, used for preprocessing image data and text data; A data classification module, used to input the preprocessed data information into a trained deep learning model, classify the data information through the trained deep learning model, and output the classification result through the deep learning model; A data level analysis module, used to obtain the security level corresponding to the data according to the classification result; The data encryption module is used to encrypt the classified data according to the encryption method preset for each security level.

8. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of a data security classification method based on deep learning as described in any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of a data security classification method based on deep learning as described in any one of claims 1 to 6 are implemented.

Images