Industrial internet data asset security management platform and method

By acquiring enterprise data assets and sensitive data feature databases, and using deep learning technology for feature extraction and correlation analysis, the problem of data security risks is solved, the recognition accuracy and data protection effect of sensitive data are improved, and the risk of data leakage is reduced.

CN119066682AInactive Publication Date: 2024-12-03YUNNAN YUNHUXUAN DIGITAL TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411261205.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-10
Publication Date
2024-12-03
Estimated Expiration
Not applicable · inactive patent

Smart Images

  • Figure CN119066682A_ABST
    Figure CN119066682A_ABST
Patent Text Reader

Abstract

The invention relates to the field of data asset security management, and particularly discloses an industrial internet data asset security management platform and method.The method comprises the steps that firstly, enterprise data assets and a sensitive data feature library are obtained, then feature extraction and correlation analysis are conducted on the enterprise data assets and the sensitive data feature library through the deep learning technology, and finally, the sensitive data feature library is obtained through a classifier; according to the method and the device, the enterprise data asset sensitive level classification is obtained, so that the sensitive data identification precision is improved, the data leakage risk is reduced, a targeted security policy is implemented, and the security and privacy of the enterprise data assets are protected.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of data asset security management, and more specifically, to an industrial Internet data asset security management platform and method. Background Art

[0002] Data assets refer to all the data sets owned by an enterprise or organization. These data can be structured, such as table information in a database, or unstructured, such as text, images, and videos. The value of data assets lies in their ability to provide insights, support decision-making, improve operational efficiency, and promote innovation. Effective management and utilization of data assets can create competitive advantages for enterprises and improve their business strategies and operational efficiency.

[0003] With the rapid development of the marketization and value creation of data elements, the value of data continues to be highlighted, while data security risks are also increasing day by day. Strengthening data security governance and security protection has become the most concerned issue in the digital economy era.

[0004] Therefore, an industrial Internet data asset security management platform and method are desired. Summary of the invention

[0005] In order to solve the above technical problems, this application is proposed. The embodiment of this application provides an industrial Internet data asset security management platform and method, which first obtains enterprise data assets and sensitive data feature libraries, then uses deep learning technology to extract features and perform correlation analysis on the two, and finally uses a classifier to obtain the sensitivity level classification of enterprise data assets, thereby improving the recognition accuracy of sensitive data and reducing the risk of data leakage, thereby implementing targeted security strategies to protect the security and privacy of enterprise data assets.

[0006] According to one aspect of the present application, an industrial Internet data asset security management platform is provided, which includes: Data asset security information acquisition module, used to obtain enterprise data assets and sensitive data feature libraries; A data asset security information extraction module, used to extract enterprise data asset global correlation feature vectors and sensitive data segment granularity feature vectors from the enterprise data assets and the sensitive data feature library; The asset sensitivity level classification module is used to obtain the enterprise data asset sensitivity level classification based on the enterprise data asset global association feature vector and the sensitive data segment granularity feature vector.

[0007] According to another aspect of the present application, a method for secure management of industrial Internet data assets is provided, comprising: Obtain enterprise data assets and sensitive data feature libraries; Extracting enterprise data asset global correlation feature vectors and sensitive data segment granularity feature vectors from the enterprise data assets and the sensitive data feature library; Based on the enterprise data asset global association feature vector and the sensitive data segment granularity feature vector, the enterprise data asset sensitivity level classification is obtained.

[0008] Compared with the existing technology, the industrial Internet data asset security management platform and method provided by this application first obtains the enterprise data assets and sensitive data feature library, then uses deep learning technology to perform feature extraction and correlation analysis on the two, and finally uses a classifier to obtain the sensitivity level classification of enterprise data assets, thereby improving the recognition accuracy of sensitive data and reducing the risk of data leakage, thereby implementing targeted security strategies to protect the security and privacy of enterprise data assets. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0010] Figure 1 It is a block diagram of the industrial Internet data asset security management platform according to an embodiment of the present application.

[0011] Figure 2 It is a block diagram of the data asset security information extraction module in the industrial Internet data asset security management platform according to the embodiment of the present application.

[0012] Figure 3 It is a block diagram of a feature extraction unit of a sensitive data feature library in an industrial Internet data asset security management platform according to an embodiment of the present application.

[0013] Figure 4 It is a block diagram of the asset sensitivity level classification module in the industrial Internet data asset security management platform according to the embodiment of the present application.

[0014] Figure 5 This is a flowchart of the industrial Internet data asset security management method according to an embodiment of the present application. DETAILED DESCRIPTION

[0015] Various exemplary embodiments, features and aspects of the present application will be described in detail below with reference to the accompanying drawings. The same reference numerals in the accompanying drawings represent elements with the same or similar functions. Although various aspects of the embodiments are shown in the accompanying drawings, the drawings are not necessarily drawn to scale unless otherwise specified.

[0016] The word “exemplary” is used exclusively herein to mean “serving as an example, example, or illustration.” Any embodiment described herein as “exemplary” is not necessarily to be construed as preferred or advantageous over other embodiments.

[0017] In addition, in order to better illustrate the present application, numerous specific details are given in the following specific embodiments. It should be understood by those skilled in the art that the present application can also be implemented without certain specific details. In some examples, methods, means, components and circuits well known to those skilled in the art are not described in detail in order to highlight the subject matter of the present application.

[0018] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of this application, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0019] Figure 1 Schematic diagram of the industrial Internet data asset security management platform of the embodiment of the present application. Figure 1 As shown, the industrial Internet data asset security management platform 100 according to the embodiment of the present application includes: a data asset security information acquisition module 110, which is used to obtain enterprise data assets and sensitive data feature libraries; a data asset security information extraction module 120, which is used to extract enterprise data asset global correlation feature vectors and sensitive data segment granularity feature vectors from the enterprise data assets and the sensitive data feature library; an asset sensitivity level classification module 130, which is used to obtain enterprise data asset sensitivity level classification based on the enterprise data asset global correlation feature vector and the sensitive data segment granularity feature vector.

[0020] In the above-mentioned industrial Internet data asset security management platform 100, the data asset security information acquisition module 110 is used to obtain enterprise data assets and sensitive data feature libraries. It should be understood that data assets refer to all data sets held by an enterprise or organization, including structured data (such as tables in a database) and unstructured data (such as text). The value of these data is reflected in the ability to provide insights, support decision-making, improve operational efficiency and promote innovation. Proper management and use of these data can bring competitive advantages to enterprises and optimize business strategies and operational efficiency. With the rapid development of data marketization and value, the value of data is becoming increasingly prominent, and data security risks are also increasing. Therefore, in the technical solution of this application, enterprise data assets and sensitive data feature libraries are obtained, and combined with deep learning technology, enterprise data assets are classified by sensitivity level to improve the recognition accuracy of sensitive data and reduce the risk of data leakage. This process helps enterprises formulate targeted security strategies to more effectively protect the security and privacy of data assets.

[0021] Specifically, enterprise data assets refer to all data sets owned by an enterprise, including structured data (such as table information in a database) and unstructured data (such as text). These data assets play a core role in enterprise operations. They can provide in-depth business insights, support decision-making, improve operational efficiency, and promote innovation. Effective data asset management can help enterprises extract valuable information from large amounts of data, thereby optimizing business strategies, improving market competitiveness, and promoting business growth. The sensitive data feature library is a database specifically for classifying and storing sensitive data. Sensitive data includes personal identity information, financial data, etc., which may cause serious damage to individuals and enterprises if leaked or abused. The purpose of establishing a sensitive data feature library is to mark, classify and protect this data to ensure its security during storage and transmission. The feature library helps monitor the use and access of data and provide necessary security protection measures by recording the characteristics of sensitive data in detail. By establishing and maintaining a detailed sensitive data feature library, enterprises can more effectively implement data protection strategies such as data encryption, access control, and audit tracking, thereby reducing the risk of data leakage and abuse.

[0022] In summary, acquiring enterprise data assets and establishing a sensitive data feature library are key steps to ensure data security and maximize data utilization. They not only help protect sensitive information of enterprises and users, but also improve data management efficiency, giving enterprises an advantage in the fiercely competitive market.

[0023] In the above-mentioned industrial Internet data asset security management platform 100, the data asset security information extraction module 120 is used to extract the enterprise data asset global correlation feature vector and the sensitive data segment granularity feature vector from the enterprise data assets and the sensitive data feature library. It should be understood that extracting the global correlation feature vector and the granularity feature vector enables enterprises to fully grasp the structure of data assets and the security requirements of sensitive data. This not only optimizes the effect of data analysis and processing, but also enhances the ability to protect sensitive data, in line with data governance and compliance requirements. Through this refined data management, enterprises can maximize the utilization value of data while ensuring data security, and promote continuous growth and innovation of the business.

[0024] Figure 2 Schematic diagram of a data asset security information extraction module in the industrial Internet data asset security management platform according to an embodiment of the present application. Figure 2 As shown, in a specific embodiment of the present application, the data asset security information extraction module 120 includes: an enterprise data asset feature extraction unit 121, used to extract features from the enterprise data assets to obtain a global correlation feature vector of the enterprise data assets; a sensitive data feature library feature extraction unit 122, used to extract features from the sensitive data feature library to obtain a sensitive data segment granularity feature vector.

[0025] It should be understood that enterprise data assets usually include a variety of structured and unstructured data, such as sales records, customer feedback, market trends, etc. These data may be fragmented and difficult to use directly in their original form, and there may be complex and non-obvious relationships between data points. For example, there may be implicit correlations between customer behavior data and sales performance. By extracting global correlation feature vectors, enterprises can identify these hidden patterns and conduct deeper data analysis. This in-depth analysis helps to reveal bottlenecks in business processes, discover market opportunities, and optimize operational strategies.

[0026] Furthermore, feature extraction is performed on the sensitive data feature library to obtain granular feature vectors of sensitive data segments, aiming to achieve more accurate data protection and management. This extraction process helps to subdivide sensitive data into smaller segments, forming detailed granular feature vectors, so that specific security measures can be applied to each data segment. Fine-grained feature vectors can not only effectively identify and classify the sensitivity of each data segment, ensuring the security of data during storage, processing and transmission, but also improve the flexibility and accuracy of data protection. With this method, enterprises can achieve dynamic management and monitoring of sensitive data, prevent data leakage or misuse, and ensure that compliance requirements are met. In general, this feature extraction method provides enterprises with enhanced data security protection and optimized management means, effectively reducing the risks in sensitive data processing.

[0027] In a specific embodiment of the present application, the enterprise data asset feature extraction unit 121 includes: passing the enterprise data assets through an enterprise data asset information semantic encoder to obtain multiple enterprise data asset feature vectors; constructing the multiple enterprise data asset feature vectors into enterprise data asset association feature vectors; passing the enterprise data asset association feature vectors through an enterprise data asset convolutional neural network based on a multi-scale neighborhood feature extraction module to obtain the enterprise data asset global association feature vector.

[0028] It should be understood that enterprise data assets usually contain a large amount of heterogeneous data, covering different formats, sources and contents. By converting this data into feature vectors, the information semantic encoder can map complex, unstructured raw data into a structured, processable form. This conversion process allows enterprises to better understand and utilize data, because feature vectors provide a unified and concise representation that can clearly reveal the key features and potential relationships of the data. Among them, the feature vectors processed by the semantic encoder can reveal the implicit patterns and structures in the data. For example, by encoding customer behavior data, sales records and market trend data into feature vectors, complex associations and patterns between data can be identified. This helps enterprises discover business opportunities, optimize operational processes and make data-driven decisions. Specifically, the enterprise data assets are segmented to obtain an enterprise data asset word sequence; the embedding layer of the enterprise data asset information semantic encoder is used to map each enterprise data asset word in the enterprise data asset word sequence into a word embedding vector to obtain a sequence of enterprise data asset word embedding vectors; the sequence of enterprise data asset word embedding vectors is semantically encoded based on global context using the converter-based Bert model of the enterprise data asset information semantic encoder to obtain multiple enterprise data asset feature vectors.

[0029] Furthermore, the main purpose of constructing multiple enterprise data asset feature vectors into enterprise data asset associated feature vectors is to reveal and utilize the complex relationships between data, thereby providing more comprehensive and in-depth data insights. Specifically, considering that a single feature vector can usually only represent a part of the characteristics of the data, by constructing multiple feature vectors into associated feature vectors, enterprises can comprehensively consider the interactions and relationships between different data sources. This construction can reveal implicit patterns and associations between data, such as the relationship between customer purchasing behavior and product sales. By integrating different data features, associated feature vectors can help enterprises identify interaction effects and trends in data, thereby optimizing decisions and strategies, avoiding the limitations of a single feature perspective.

[0030] Furthermore, enterprise data assets often have multi-level and multi-dimensional characteristics, which requires that both local and global features of the data be effectively captured. That is, enterprise data assets usually involve complex relationships and patterns, and relying on a single feature extraction method alone may not be able to fully capture these complex relationships. The multi-scale neighborhood feature extraction module can analyze data from different scales and perspectives to extract multi-level feature information. Among them, convolutional neural networks (CNNs) are good at automatically learning local features in data, and can efficiently extract and combine local features through convolution operations. CNN based on the multi-scale neighborhood feature extraction module can integrate feature information at different scales, automatically extract important features in enterprise data assets, and integrate these features into high-dimensional feature representations through convolutional layers. The resulting global correlation feature vector can accurately represent the overall characteristics and complex relationships of the data. Specifically, the enterprise data asset convolutional neural network based on the multi-scale neighborhood feature extraction module includes: a first convolutional layer, a second convolutional layer parallel to the first convolutional layer, and a cascade layer connected to the first convolutional layer and the second convolutional layer, wherein the first convolutional layer uses a one-dimensional convolutional kernel with a first scale, and the second convolutional layer uses a one-dimensional convolutional kernel with a second scale. More specifically, the first convolutional layer of the enterprise data asset convolutional neural network based on the multi-scale neighborhood feature extraction module is used to perform one-dimensional convolution encoding on the enterprise data asset associated feature vector to obtain a first-scale feature vector; the second convolutional layer of the enterprise data asset convolutional neural network based on the multi-scale neighborhood feature extraction module is used to perform one-dimensional convolution encoding on the enterprise data asset associated feature vector to obtain a second-scale feature vector; the first-scale feature vector and the second-scale feature vector are cascaded to obtain the enterprise data asset global associated feature vector.

[0031] Figure 3 Schematic diagram of a block diagram of a sensitive data feature library feature extraction unit in an industrial Internet data asset security management platform according to an embodiment of the present application. Figure 3As shown, in a specific embodiment of the present application, the sensitive data feature library feature extraction unit 122 includes: a sensitive data feature library data preprocessing subunit 1221, used to perform data preprocessing on the sensitive data feature library to obtain a sensitive data segment embedding matrix; a sensitive data feature library feature encoding subunit 1222, used to perform feature encoding on the sensitive data segment embedding matrix to obtain a granular feature vector of the sensitive data segment.

[0032] It should be understood that sensitive data feature libraries usually contain data from different sources and formats, which may differ in structure, type, and representation. The first step of data preprocessing is to convert these heterogeneous data into a unified format to ensure data compatibility and consistency. By normalizing the data format and structure, preprocessing can eliminate noise and inconsistency and provide clean and structured input data for subsequent embedding generation. For example, data can be converted into numerical features, categorical features, or embedding vectors to improve the operability of the data. The embedding matrix represents the data segment as a vector of fixed dimension, so that the data can be further analyzed and utilized. The embedding matrix can map the sensitive data segment to a low-dimensional vector space, retaining important information and relationships in the data. This representation can reflect the context and semantics of the data, helping to better understand and process the data.

[0033] Furthermore, the sensitive data segment embedding matrix converts the original data into a vector representation in a high-dimensional space, capturing the basic features and semantic information of the data. However, these vectors may contain redundant information in their original form or be inconvenient to directly apply to analysis and modeling. The feature encoding process can further refine and abstract the information in these embedding matrices and convert them into feature vectors with specific granularity and structure. This granular feature vector can more accurately express the key information of the data, which helps to improve the effectiveness of subsequent analysis and models.

[0034] In a specific embodiment of the present application, the sensitive data feature library data preprocessing subunit 1221 includes: segmenting the sensitive data feature library to obtain multiple sensitive data segment data items; arranging the multiple sensitive data segment data items into the sensitive data segment embedding matrix.

[0035] It should be understood that processing large-scale data sets may face computing and storage challenges. Segmenting data can reduce the amount of data for each operation, thereby improving the efficiency of data processing and computing. Specifically, segmenting the data splits the entire feature library into several small, easy-to-manage data segments, each of which focuses on a specific data type or feature. This fine-grained processing method can make analysis and processing more precise and help identify and process patterns and trends in specific data segments. In this way, the internal structure of the data can be better understood and potential details and anomalies in the data can be discovered.

[0036] Furthermore, arranging multiple sensitive data segment data items into a sensitive data segment embedding matrix is ​​to optimize the data representation and processing process to improve the effect and efficiency of data analysis. Among them, the sensitive data segment embedding matrix provides a structured data representation method by integrating different data segment data items into a unified matrix. This matrix form can arrange and encode the characteristics of the data items in an orderly manner, ensure the consistency and standardization of the data, and enable subsequent analysis and processing to be based on a unified format, thereby reducing the complexity of data conversion and processing. That is, in many machine learning and deep learning models, the input of data usually needs to be in the form of a matrix. Arranging multiple sensitive data segment data items into an embedding matrix allows these data to be directly input into the model for training and prediction. For example, the embedding matrix in deep learning can be used as part of the input layer to extract the potential features of the data through matrix operations, thereby improving the performance and prediction ability of the model.

[0037] In a specific embodiment of the present application, the sensitive data feature library feature encoding subunit 1222 includes: embedding the sensitive data segment into a matrix through a sensitive data segment convolutional neural network based on a convolutional neural network to obtain a sensitive data segment granularity feature map; pooling the sensitive data segment granularity feature map into the sensitive data segment granularity feature vector.

[0038] It should be understood that convolutional neural networks are good at automatically extracting local features from data. Through convolution operations, CNNs can identify local patterns in the matrix, such as edges, corners, or other important features. This is particularly important for sensitive data segments, because these data segments may contain subtle patterns or structures that may be overlooked in data preprocessing or feature engineering. Convolution operations can reveal subtle differences and important information in the data, helping to gain a deeper understanding of the characteristics of the data. Through multiple layers of convolution operations and pooling processes, CNNs are able to automatically learn and build high-level features of the data. This ability to automatically learn features reduces the reliance on manual feature engineering, allowing the model to discover and extract the most useful features from the raw data. For sensitive data segments, in this way, CNNs are able to effectively identify and represent important information in the data and improve the effect of feature extraction. The granular feature maps generated by convolutional neural networks can more clearly represent the key features in the data, making the analysis and interpretation of the data more intuitive. These feature maps can help analysts understand the underlying patterns and structures in the data, so as to make more targeted analysis and decisions. Specifically, each layer of the sensitive data segment convolutional neural network based on the convolutional neural network is used to perform convolution processing, mean pooling processing based on the local feature matrix and nonlinear activation processing on the input data in the forward pass of the layer, so that the last layer of the sensitive data segment convolutional neural network based on the convolutional neural network outputs the sensitive data segment granularity feature map, wherein the input of the sensitive data segment convolutional neural network based on the convolutional neural network is the sensitive data segment embedding matrix.

[0039] Furthermore, pooling operations (such as max pooling or average pooling) can significantly reduce the dimensionality of feature maps, thereby converting high-dimensional feature maps into low-dimensional feature vectors. This dimensionality reduction process not only reduces computational and storage requirements, but also simplifies the data structure, making subsequent processing and analysis more efficient and manageable. The core function of the pooling operation is to extract the main information in the feature map and ignore unimportant details. For example, max pooling retains the most significant features by selecting the maximum value in each pooling region, while average pooling retains the overall information of the region by calculating the average value within the region. This method ensures that the feature vector contains the key information in the feature map, which helps improve the model's ability to understand and analyze the data.

[0040] In the above-mentioned industrial Internet data asset security management platform 100, the asset sensitivity level classification module 130 is used to obtain the enterprise data asset sensitivity level classification based on the enterprise data asset global association feature vector and the sensitive data segment granularity feature vector. It should be understood that the global association feature vector represents the overall structure and global relationship of the enterprise data asset, and can reveal the position and role of the data in the entire system. The sensitive data segment granularity feature vector focuses on the detailed features of a specific data segment, providing a deep insight into individual data fragments. Combining these two features, data assets can be comprehensively evaluated from both macro and micro levels to ensure a comprehensive analysis of the sensitivity of the data. By analyzing global and local features, it is possible to accurately identify which data has a higher sensitivity. Based on the classification results, enterprises can formulate more targeted protection strategies. Data with high sensitivity levels requires more stringent protection measures, such as encryption and access control, while data with low sensitivity levels can adopt relatively loose management measures. This classification method makes data protection strategies more reasonable and effective, and helps optimize resource allocation and risk management.

[0041] Figure 4 Schematic diagram of the asset sensitivity level classification module in the industrial Internet data asset security management platform according to the embodiment of the present application. Figure 4 As shown, in a specific embodiment of the present application, the asset sensitivity level classification module 130 includes: a data asset security feature fusion unit 131, which is used to perform fine-grained correspondence correction on the global correlation feature vector of the enterprise data asset and the sensitive data segment granularity feature vector based on the high-dimensional manifold foreground-background structure to obtain an asset sensitivity level judgment classification feature vector; an enterprise data asset sensitivity level classification judgment unit 132, which is used to pass the asset sensitivity level judgment classification feature vector through a classifier to obtain a classification result, and the classification result is used to represent the enterprise data asset sensitivity level classification.

[0042] It should be understood that by integrating global and granular features, the sensitivity level of data can be more accurately assessed, thereby formulating more appropriate data protection strategies. Data with high sensitivity levels may require stricter protection measures, such as encryption and access control, while data with low sensitivity levels can adopt relatively loose management measures. This classification method helps enterprises consider both the overall importance of data and the specific sensitivity details when formulating data protection strategies. This fusion method makes data sensitivity assessment more comprehensive and accurate, helping enterprises to effectively protect and manage data in complex data environments.

[0043] In particular, in the technical solution of the present application, the global correlation feature vector of enterprise data assets is processed by a semantic encoder and a multi-scale neighborhood feature extraction module, aiming to capture the overall information and global relationship of data assets, reflecting the correlation and structural pattern of data assets at the macro level, and emphasizing the global feature representation and complex contextual relationship. Relatively speaking, the processing of sensitive data segments pays more attention to local details and paragraph-level features. The feature vector of sensitive data segments is extracted by a convolutional neural network to capture local granular information, such as the specific content and pattern of the data segment. Due to the different processing focuses, there are significant differences in the dimensions and detail levels between enterprise data asset features and sensitive data segment features. This difference leads to the offset of the fine-grained dimension of the features, making it difficult for the two features to effectively integrate information when fused. Specifically, the global features of enterprise data assets may not fully reflect the local details of sensitive data segments, and the granular features of sensitive data segments cannot fully capture the global information of enterprise data assets. The loss or distortion of local structure will cause the final fused feature vector to fail to accurately represent the sensitivity level of the data. For example, the subtle information in the sensitive data segment may be diluted, and the global information of the enterprise data assets may not be able to effectively make up for the lack of these details, thereby affecting the judgment accuracy of the classifier. Therefore, in the technical solution of the present application, a fine-grained correspondence correction is performed on the global correlation feature vector of enterprise data assets and the granular feature vector of sensitive data segments based on the high-dimensional manifold foreground-background structure to obtain a classification feature vector for judging the asset sensitivity level.

[0044] Among them, the fine-grained correspondence correction based on the high-dimensional manifold foreground-background structure is performed on the global correlation feature vector of the enterprise data asset and the granularity feature vector of the sensitive data segment to obtain the asset sensitivity level judgment classification feature vector, including: calculating the position difference between the global correlation feature vector of the enterprise data asset and the granularity feature vector of the sensitive data segment to obtain the fine-grained correspondence correction difference feature vector; calculating the position point multiplication between the global correlation feature vector of the enterprise data asset and the granularity feature vector of the sensitive data segment to obtain the fine-grained correspondence correction position point multiplication feature vector; calculating the position addition between the global correlation feature vector of the enterprise data asset and the granularity feature vector of the sensitive data segment to obtain the fine-grained correspondence correction position addition feature vector; adding the fine-grained correspondence correction difference feature vector, the fine-grained correspondence correction position point multiplication feature vector and The fine-grained corresponding correction position-added feature vectors are cascaded and input into a one-dimensional convolutional layer and maximum pooling is performed to obtain a fine-grained corresponding correction convolutional coding pooling feature vector; the square root of the two norms of the fine-grained corresponding correction convolutional coding pooling feature vector is weighted and added with the one norm of the fine-grained corresponding correction convolutional coding pooling feature vector to obtain a fine-grained corresponding correction factor; the square root of the length of the fine-grained corresponding correction positional dot product feature vector and the global association feature vector of the enterprise data asset are added positionally to obtain a fine-grained corresponding correction intermediate vector; the fine-grained corresponding correction factor is used as the first weighting coefficient and the second weighted hyperparameter is used as the second weighting coefficient to calculate the position-weighted sum of the fine-grained corresponding correction position-added feature vector and the fine-grained corresponding correction intermediate vector to obtain an asset sensitivity level judgment classification feature vector.

[0045] Among them, the global correlation feature vector of enterprise data assets and the granular feature vector of sensitive data segments are corrected in fine-grained correspondence based on the high-dimensional manifold foreground-background structure to obtain the asset sensitivity level judgment classification feature vector, including: ; ; ;in, Represents the global correlation feature vector of enterprise data assets, Represents the sensitive data segment granularity feature vector, It means subtracting by position. It means point multiplication by position. It means adding by position. represents a one-dimensional convolutional layer, represents maximum pooling, represents the fine-grained corresponding rectified convolutional coding pooling feature vector, represents a norm, represents the two-norm, Indicates cascade, represents the first weighted hyperparameter, represents the correction factor corresponding to fine granularity, Represents the length of the global correlation feature vector of enterprise data assets, represents the second weighted hyperparameter, Represents the classification feature vector for judging asset sensitivity level.

[0046] In the technical solution of the present application, in the process of fusing the enterprise data asset global correlation feature vector and the sensitive data segment granular feature vector, the feature fine-grained dimension offset between the enterprise data asset global correlation feature vector and the sensitive data segment granular feature vector is taken into consideration, which may cause the fused asset sensitivity level judgment classification feature vector to have local structural collapse. Based on this, in the technical solution of the present application, the enterprise data asset global correlation feature vector and the sensitive data segment granular feature vector are subjected to fine-grained correspondence correction based on the high-dimensional manifold foreground-background structure.

[0047] Specifically, based on the restricted expression of the foreground-background structure distinction of the high-dimensional manifold of the global association feature vector of the enterprise data asset and the granular feature vector of the sensitive data segment by the amplitude of the semantic difference vector between the global association feature vector of the enterprise data asset and the granular feature vector of the sensitive data segment, the fine-grained correspondence of key features between the global association feature vector of the enterprise data asset and the granular feature vector of the sensitive data segment is modeled, and then active fine-grained correspondence adjustment propagation is performed through the back-propagation displacement compensation information between the corresponding eigenvalues ​​of the global association feature vector of the enterprise data asset and the granular feature vector of the sensitive data segment, so as to avoid the imbalance at the feature vector level of the global association feature vector of the enterprise data asset and the granular feature vector of the sensitive data segment in a focusing manner, thereby synchronizing the micro-dimensions of the two feature vectors, ensuring that in the fusion process, the information between the feature vectors can be effectively aligned and combined, thereby avoiding the abnormal accumulation of local structures.

[0048] Furthermore, the asset sensitivity level judgment classification feature vector provides a comprehensive perspective and details of the data. The classifier uses this information to more accurately judge the sensitivity of the data, thereby improving the accuracy of the classification. By outputting the classification results of "highly sensitive data", "generally sensitive data" and "lowly sensitive data", it can provide enterprises with a clear division of data sensitivity levels, thereby supporting the formulation of specific data protection strategies. Specifically, data with a high sensitivity level may require more stringent protection measures, such as encryption and access control, while data with a low sensitivity level can adopt relatively loose management measures. This classification result helps enterprises formulate targeted data protection strategies, effectively allocate resources, and reduce risks.

[0049] In summary, the embodiment of the present application first obtains the enterprise data assets and sensitive data feature library, then uses deep learning technology to perform feature extraction and correlation analysis on the two, and finally uses a classifier to obtain the sensitivity level classification of the enterprise data assets, thereby improving the recognition accuracy of sensitive data and reducing the risk of data leakage, thereby implementing targeted security strategies to protect the security and privacy of enterprise data assets.

[0050] As described above, the industrial Internet data asset security management platform 100 according to the embodiment of the present application can be implemented in various terminal devices. In one example, the industrial Internet data asset security management platform 100 can be integrated into the terminal device as a software module and / or a hardware module. For example, the industrial Internet data asset security management platform 100 can be a software module in the operating system of the terminal device, or it can be an application developed for the terminal device; of course, the industrial Internet data asset security management platform 100 can also be one of the many hardware modules of the terminal device.

[0051] Alternatively, in another example, the industrial Internet data asset security management platform 100 and the terminal device may also be separate devices, and the industrial Internet data asset security management platform 100 may be connected to the terminal device via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0052] Figure 5 Flowchart of the industrial Internet data asset security management method according to the embodiment of the present application. Figure 5 As shown, according to the industrial Internet data asset security management method of the embodiment of the present application, it includes: S110, obtaining enterprise data assets and sensitive data feature libraries; S120, extracting enterprise data asset global correlation feature vectors and sensitive data segment granularity feature vectors from the enterprise data assets and the sensitive data feature libraries; S130, obtaining enterprise data asset sensitivity level classification based on the enterprise data asset global correlation feature vectors and the sensitive data segment granularity feature vectors.

[0053] Here, those skilled in the art can understand that the specific operations of each step in the above-mentioned industrial Internet data asset security management method have been referenced above. Figures 1 to 4 It has been introduced in detail in the description of the Industrial Internet Data Asset Security Management Platform, and therefore, its repeated description will be omitted.

[0054] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of the modules is only a logical function division, and there may be other division methods in actual implementation.

[0055] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0056] In addition, each functional module in each embodiment of the present invention may be integrated into one processing unit, each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0057] It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0058] Therefore, no matter from which point of view, the embodiments should be regarded as illustrative and non-restrictive, and the scope of the present invention is limited by the appended claims rather than the above description, so it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims are included in the present invention. Any attached figure mark in the claims should not be regarded as limiting the claims involved.

[0059] In addition, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices stated in the system claim can also be implemented by one unit or device through software or hardware. The second and other words are used to indicate names, but not to indicate any particular order.

[0060] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit of the technical solution of the present invention.

Claims

1. An industrial Internet data asset security management platform, characterized in that: include: Data asset security information acquisition module, used to obtain enterprise data assets and sensitive data feature libraries; A data asset security information extraction module, used to extract enterprise data asset global correlation feature vectors and sensitive data segment granularity feature vectors from the enterprise data assets and the sensitive data feature library; The asset sensitivity level classification module is used to obtain the enterprise data asset sensitivity level classification based on the enterprise data asset global association feature vector and the sensitive data segment granularity feature vector.

2. The industrial Internet data asset security management platform according to claim 1 is characterized in that: The data asset security information extraction module includes: An enterprise data asset feature extraction unit, used to extract features from the enterprise data assets to obtain a global correlation feature vector of the enterprise data assets; The sensitive data feature library feature extraction unit is used to extract features from the sensitive data feature library to obtain the sensitive data segment granularity feature vector.

3. The industrial Internet data asset security management platform according to claim 2 is characterized in that: The enterprise data asset feature extraction unit includes: Passing the enterprise data assets through an enterprise data asset information semantic encoder to obtain a plurality of enterprise data asset feature vectors; constructing the plurality of enterprise data asset feature vectors into an enterprise data asset associated feature vector; The enterprise data asset association feature vector is passed through an enterprise data asset convolutional neural network based on a multi-scale neighborhood feature extraction module to obtain the enterprise data asset global association feature vector.

4. The industrial Internet data asset security management platform according to claim 3 is characterized in that: The sensitive data feature library feature extraction unit includes: A sensitive data feature library data preprocessing subunit, used for performing data preprocessing on the sensitive data feature library to obtain a sensitive data segment embedding matrix; The sensitive data feature library feature encoding subunit is used to perform feature encoding on the sensitive data segment embedding matrix to obtain the sensitive data segment granularity feature vector.

5. The industrial Internet data asset security management platform according to claim 4 is characterized in that: The sensitive data feature library data preprocessing subunit includes: Segmenting the sensitive data feature library to obtain a plurality of sensitive data segment data items; The multiple sensitive data segment data items are arranged into the sensitive data segment embedding matrix.

6. The industrial Internet data asset security management platform according to claim 5 is characterized in that: The sensitive data feature library feature encoding subunit includes: The sensitive data segment embedding matrix is ​​passed through a sensitive data segment convolutional neural network based on a convolutional neural network to obtain a sensitive data segment granularity feature map; The sensitive data segment granularity feature map is pooled into the sensitive data segment granularity feature vector.

7. The industrial Internet data asset security management platform according to claim 6, characterized in that: The asset sensitivity level classification module includes: The data asset security feature fusion unit is used to perform fine-grained correspondence correction on the enterprise data asset global correlation feature vector and the sensitive data segment granularity feature vector based on the high-dimensional manifold foreground-background structure to obtain the asset sensitivity level judgment classification feature vector; The enterprise data asset sensitivity level classification judgment unit is used to pass the asset sensitivity level judgment classification feature vector through a classifier to obtain a classification result, and the classification result is used to represent the enterprise data asset sensitivity level classification.

8. The industrial Internet data asset security management platform according to claim 7, characterized in that: The data asset security feature fusion unit includes: Calculating the positional difference between the enterprise data asset global association feature vector and the sensitive data segment granularity feature vector to obtain a fine-grained corresponding correction differential feature vector; Calculate the position point multiplication between the enterprise data asset global association feature vector and the sensitive data segment granularity feature vector to obtain a fine-grained corresponding corrected position point multiplication feature vector; Calculate the positional addition between the enterprise data asset global association feature vector and the sensitive data segment granularity feature vector to obtain a fine-grained corresponding corrected positional addition feature vector; The fine-grained corresponding correction difference feature vector, the fine-grained corresponding correction position point product feature vector and the fine-grained corresponding correction position addition feature vector are cascaded and input into a one-dimensional convolution layer and maximum pooling is performed to obtain a fine-grained corresponding correction convolutional coding pooling feature vector; The square root value of the two-norm of the fine-grained corresponding correction convolutional coding pooling feature vector is weighted and added to the one-norm of the fine-grained corresponding correction convolutional coding pooling feature vector to obtain a fine-grained corresponding correction factor; Add the square root of the length of the point-by-point feature vector of the fine-grained corresponding correction and the global associated feature vector of the enterprise data asset by position to obtain a fine-grained corresponding correction intermediate vector; Using the fine-grained corresponding correction factor as the first weighting coefficient and the second weighted hyperparameter as the second weighting coefficient, the position-weighted sum of the fine-grained corresponding correction position-added feature vector and the fine-grained corresponding correction intermediate vector is calculated to obtain the asset sensitivity level judgment classification feature vector.

9. A method for secure management of industrial Internet data assets, characterized in that: include: Obtain enterprise data assets and sensitive data feature libraries; Extracting enterprise data asset global correlation feature vectors and sensitive data segment granularity feature vectors from the enterprise data assets and the sensitive data feature library; Based on the enterprise data asset global association feature vector and the sensitive data segment granularity feature vector, the enterprise data asset sensitivity level classification is obtained.

10. The industrial Internet data asset security management method according to claim 9 is characterized in that: Extracting the enterprise data asset global correlation feature vector and the sensitive data segment granularity feature vector from the enterprise data asset and the sensitive data feature library includes: Extracting features from the enterprise data assets to obtain a global correlation feature vector of the enterprise data assets; Feature extraction is performed on the sensitive data feature library to obtain the sensitive data segment granularity feature vector.