An enterprise data management system based on cloud platform

Abstract

The present invention relates to the fields of cloud platform technology and data management technology, and provides an enterprise data management system based on a cloud platform. A data acquisition module collects enterprise data and transmits it to an encryption unit of a security module. The encryption unit encrypts the enterprise data and then uploads it to a main cloud platform of a cloud computing platform. The main cloud platform analyzes and calculates the enterprise data. At the same time, a data synchronization module synchronizes the data of the main cloud platform to a subsidiary cloud platform. After the main cloud platform completes the analysis, calculation and encryption, the data is transmitted to a verification unit. The verification unit completes verification and decryption and transmits the data to an interactive module for relevant personnel to review and operate. The data recovery module is started when the relevant personnel triggers a recovery operation, and extracts data from the subsidiary cloud platform for recovery. The present invention effectively reduces the dependence on a single cloud service provider through a cloud computing platform, and avoids the risks of additional costs, increased workload and business stagnation caused by dependence on a single provider.

Description

Technical Field

[0001] The present invention relates to the technical field of data management, and in particular to an enterprise data management system based on a cloud platform. Background Art

[0002] Enterprise data management system is a system used by enterprises to manage and store various types of data. With the development of cloud computing technology, more and more enterprises have begun to deploy data management systems on cloud platforms to achieve more efficient and flexible data management. At the same time, cloud platforms also provide flexible data backup and recovery functions to ensure data security and reliability.

[0003] The enterprise data management system based on the cloud platform can also realize cross-regional and cross-departmental data sharing and collaboration, improving the information flow and work efficiency within the enterprise. In addition, the cloud platform can also integrate various data analysis tools and artificial intelligence technologies to help enterprises better use data for business decision-making and innovation.

[0004] In general, the enterprise data management system based on the cloud platform has the advantages of low cost, high flexibility and strong security, and has become the preferred solution for data management of many enterprises.

[0005] However, the existing background technology still has the following disadvantages:

[0006] 1. Instability of a single cloud platform: When enterprises face unforeseen situations such as technical failures, service interruptions or business strategy adjustments by cloud service providers, they will face risks to their own business, which will not only bring additional costs and workload, but may also cause temporary stagnation of business. In addition, although the cloud platform provides backup functions, the storage location and recovery mechanism of backup data may have hidden dangers. For example, backup data may be stored in other data centers of the same cloud service provider. If the data center encounters natural disasters or other force majeure factors, the backup data may also be damaged.

[0007] 2. Most existing enterprise data management systems are not flexible enough in data analysis and decision-making calculations. Most of them cannot make dynamic adjustments based on real-time changes in enterprise data, cannot accurately generate decisions based on the actual situation of the enterprise, and cannot adapt to the complexity and changes of the market in a timely manner, which easily causes enterprises to lack accurate and effective data support when making decisions. Summary of the invention

[0008] In order to solve the above technical problems, the present invention provides an enterprise data management system based on a cloud platform.

[0009] The technical solution of the present invention is as follows:

[0010] An enterprise data management system based on a cloud platform, comprising:

[0011] A data collection module, which is used to collect enterprise data information from the enterprise intranet by means of web crawlers or system interface docking. The enterprise data includes but is not limited to financial data, customer data, sales data, production data, market data and operation data, and enterprise historical decision-making or historical strategy information data;

[0012] A cloud computing platform, comprising a main cloud platform and one or more subsidiary cloud platforms, for analyzing and computing enterprise data;

[0013] A data synchronization module, which is used to perform data synchronization and transmission tasks between the main cloud platform and the subsidiary cloud platform in real time;

[0014] A security module, which includes an encryption unit and a verification unit, and is used to encrypt the enterprise data before analysis and verify the enterprise data after analysis;

[0015] An interactive module, which is used to provide users with analysis and calculation results of the cloud computing platform and an operation port, which allows relevant users to perform system settings and adjust calculation parameters in the cloud computing platform;

[0016] A data recovery module is used to extract data from the subsidiary cloud platform for recovery.

[0017] The interaction relationship between the modules of the enterprise data management system based on the cloud platform is as follows:

[0018] The data acquisition module collects enterprise data and transmits it to the encryption unit of the security module. The encryption unit encrypts the enterprise data and uploads it to the main cloud platform of the cloud computing platform. The main cloud platform analyzes and calculates the enterprise data. At the same time, the data synchronization module synchronizes the data of the main cloud platform to the subsidiary cloud platform to realize data backup and cloud platform data sharing. After the main cloud platform completes the analysis and calculation, it transmits the data to the verification unit. The verification unit completes the verification and decryption and transmits the data to the interactive module for relevant personnel to review and operate. The data recovery module is connected to the subsidiary cloud platform and the interactive module data, and is started when the relevant personnel trigger the recovery operation, and extracts the data in the subsidiary cloud platform for recovery.

[0019] Preferably, when the data collection module automatically grabs data from various business systems, web pages, and document sources within the enterprise through a web crawler, each newly collected data will be compared with all the data already in the system. For each piece of data, the comparison is performed from multiple dimensions including the specific identifier of the data and key content fields. If it is found that the new data is completely consistent with an existing piece of data in these dimensions, then the new data is determined to be duplicate data and is filtered out. A real-time data push mechanism is used, so that when new data is generated by the data source, it can be immediately collected into the system to ensure the timeliness of the data.

[0020] Preferably, the main cloud platform includes:

[0021] A data storage unit, wherein the data storage unit stores the information collected by the data collection module by adopting a data partition storage strategy, calculates the call frequency of the enterprise data, obtains the median of the call frequency, stores the data whose enterprise call frequency is greater than the median frequency in a high-speed storage medium, and stores the data whose enterprise call frequency is less than the median frequency in a low-cost large-capacity storage device;

[0022] A data marking unit, which is used to perform anomaly detection on the stored data, mark and record the detected abnormal data, and then transmit it to the interaction module;

[0023] A data analysis unit, which is used to analyze enterprise data, analyze the degree of connection between data and decision-making based on the acquired enterprise data, calculate the strategy to be adopted after the enterprise data changes, and provide support for enterprise decision-making;

[0024] The user rights management unit implements decentralized management of user rights based on blockchain technology, ensures that the rights information cannot be tampered with and has high security, and realizes automatic allocation and control of rights by setting and managing the access and operation rights of different users to the main cloud platform data.

[0025] Preferably, the data marking unit marking process is as follows:

[0026] First, suppose there are d items of data in the enterprise data, and construct k isolated data trees based on the classification of the time series of the data; C represents the specific data, and its path length in the zth tree is ; The formula for calculating the average path length is as follows:

[0027]

[0028] Calculate whether data C is abnormal, and judge by calculating the abnormal score. First, calculate the correction value ;

[0029]

[0030] Among them, R is the harmonic value. The number of terms affects the correction value through the harmonic number. When the number of terms increases, the harmonic number will also increase. The smaller the average path length or the larger the number of terms, the larger the correction value will be. is a harmonic number, R is equal to ;

[0031] Then the anomaly score of data item C is;

[0032]

[0033] like The closer it is to 1, the more abnormal the data item C is; compare the size of the abnormality, or sort according to the size of the abnormality, mark and record the data item with the largest abnormality.

[0034] Preferably, the data analysis unit uses a deep belief network to perform complex data analysis;

[0035] The deep belief network is composed of multiple restricted Boltzmann machines (RBMs) stacked together. For a single RBM, it has its own energy function , the energy function is expressed as:

[0036]

[0037] in, represents the amount of change in enterprise data, including but not limited to financial data, customer data, sales data, production data, market data, and operational data, Indicates Types of enterprise data, Represents the total number of categories of enterprise data collected; It means that different types of enterprise decisions or strategies are generated under the influence of different types of enterprise data. Each decision or strategy is manually quantitatively evaluated based on factors including importance, priority, implementation difficulty, expected benefits, resource investment and risk level, and converted into numerical values ​​for quantification; Indicates A decision or strategy based on the influence of enterprise data. Express the total number of decision or strategy types; is the bias of v; is the bias of u; is the connection weight between v and u, representing the strength of the association between them. The bias and connection weight are uniformly initialized and set, and the parameters are optimized through subsequent training;

[0038] Trained with a large amount of enterprise data to better optimize the deep belief network model; for each combination Calculate its energy function , constantly adjust the parameters , and , in order to minimize the energy function to achieve the training purpose; the parameter update formula of the training process is as follows:

[0039]

[0040]

[0041]

[0042] Where t represents the number of iterations. Represents the learning rate, which determines the step size of parameter updates. It can be selected and generated through grid search or adaptive adjustment. Grid search is an exhaustive parameter search method, and adaptive adjustment means adjusting according to the model training process.

[0043] Compute partial derivatives:

[0044]

[0045] represents the partial derivative, Represents enterprise data and decision making The degree of influence of , substitute the data after partial derivative calculation into the parameter update formula:

[0046]

[0047] After calculating all the training data, the parameters that ensure the minimization of the energy function are determined as the final parameters. The deep belief network after the final training is able to change according to the input of new enterprise data. , calculated with The energy function results ,for , choose the energy function Smallest The data analysis unit outputs corporate decisions, thereby providing decision-making basis for corporate managers.

[0048] Preferably, the subsidiary cloud platform includes:

[0049] A data backup unit, which receives data from the main cloud platform through a data synchronization module in real time and performs a complete backup, records the backup time point and version information, and uses a distributed database storage architecture to store different types of enterprise data in a distributed manner;

[0050] A data recovery preparation unit, which continuously monitors the integrity and availability of backup data when the main cloud platform is operating normally, to ensure rapid response when recovery is required;

[0051] A monitoring unit, which monitors the status of the main cloud platform in real time. When the main cloud platform is detected to be unavailable, the monitoring unit immediately issues an alarm and activates an emergency takeover unit;

[0052] The emergency takeover unit is activated after the monitoring unit issues an alarm, and then takes over the data acceptance task of the main cloud platform and performs storage and backup, and maintains communication with the main cloud platform. When the main cloud platform returns to normal, the emergency takeover unit will synchronize the data received during the emergency takeover back to the main cloud platform through the data synchronization module.

[0053] Preferably, the encryption process of the encryption unit in the security module is as follows:

[0054] Assume that the original data of the enterprise is G. The encryption unit processes the data G collected and uploaded by the data collection module: First, a pair of public and private keys are generated using a random number generator. The public key PH is made public to the main cloud platform, and the private key QH is saved by the enterprise itself and uploaded to the verification module. AES is used to generate a random symmetric key H. The symmetric key H is then used to encrypt the data G to obtain the ciphertext F1=EnFrypt(G,H). Next, the encryption unit uses the public key PH to asymmetrically encrypt the symmetric key H to obtain F2=EnFrypt(H,PH).

[0055] Upload the ciphertext F1 and the encrypted symmetric key F2 to the main cloud platform;

[0056] The main cloud platform decrypts the symmetric key F2 according to the public key PH to obtain the symmetric key H, and uses the symmetric key H to decrypt the ciphertext F1 to obtain the original data G for data analysis and calculation; after completion, the calculated data is encrypted using the public key PH to obtain the ciphertext F';

[0057] The verification process of the verification unit is as follows: after the main cloud platform completes the data-related analysis and calculation, the calculation result is sent to the verification unit in the form of ciphertext F'; after receiving the ciphertext F', the verification unit uses the private key QH to decrypt the previously encrypted symmetric key F2 to obtain the symmetric key H'=GeFrypt(F2,QH);

[0058] Use the decrypted symmetric key H' to decrypt the ciphertext F' of the calculation result to obtain the plaintext result G';

[0059] G'=GeFrypt(F',H'), send the plaintext result G' to the interaction module.

[0060] Preferably, the main cloud platform also includes a decryption unit, which is used to decrypt the data uploaded by the encryption unit using the public key PH, and to encrypt the calculated data using the public key PH.

[0061] Preferably, the interactive module enables users to view enterprise data and operating systems and receive alerts through mobile applications or computer web pages; enterprise data includes abnormal data marked by a data marking unit and data decisions calculated and input by a data analysis unit; the operating system includes recovering enterprise data through a data recovery module and initializing calculation parameters in the data analysis unit.

[0062] Compared with the prior art, the present invention has the following beneficial effects:

[0063] 1. The present invention realizes the computing and backup of cloud services respectively by adopting an architecture combining a main cloud platform with an auxiliary cloud platform. When the main cloud platform encounters problems due to unforeseen circumstances such as technical failures of the cloud service provider, service interruptions or adjustments to business strategies, the auxiliary cloud platform can quickly take over and store the data, thus reducing the risk of business stagnation. At the same time, it avoids the situation where backup data is damaged due to a disaster in a single data center, thus reducing the additional costs and workload of enterprises due to data problems.

[0064] 2. The present invention uses a deep belief network to perform data analysis and adopts real-time data collection to achieve accurate capture and dynamic analysis of real-time changes in enterprise data. It can generate decisions that fit the actual situation for the enterprise based on the real-time changes in enterprise data, so that the enterprise can adapt to the complex and changing market environment in a timely manner, avoid mistakes caused by inaccurate and untimely data support when making decisions, improve the scientificity and accuracy of enterprise decision-making, and enhance the competitiveness and adaptability of enterprises in the market. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 It is a work flow chart of the present invention;

[0066] Figure 2 Schematic diagram of switching between the main cloud platform and the subsidiary cloud platform of the present invention;

[0067] Figure 3 It is a data recovery operation flow chart of the present invention;

[0068] Figure 4 It is a schematic diagram of the data management process and data synchronization of the main cloud platform of the present invention. DETAILED DESCRIPTION

[0069] The following embodiments of the present invention are described in further detail in conjunction with the accompanying drawings and examples. The following examples are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

[0070] The present invention provides an enterprise data management system based on a cloud platform, comprising:

[0071] A data collection module, which is used to collect enterprise data information from the enterprise intranet by means of web crawlers or system interface docking. The enterprise data includes but is not limited to financial data, customer data, sales data, production data, market data and operation data, and enterprise historical decision-making or historical strategy information data;

[0072] A cloud computing platform, comprising a main cloud platform and one or more subsidiary cloud platforms, for analyzing and computing enterprise data;

[0073] A data synchronization module, which is used to perform data synchronization and transmission tasks between the main cloud platform and the subsidiary cloud platform in real time;

[0074] A security module, which includes an encryption unit and a verification unit, and is used to encrypt the enterprise data before analysis and verify the enterprise data after analysis;

[0075] An interactive module, which is used to provide users with analysis and calculation results of the cloud computing platform and an operation port, which allows relevant users to perform system settings and adjust calculation parameters in the cloud computing platform;

[0076] A data recovery module is used to extract data from the subsidiary cloud platform for recovery.

[0077] The interaction relationship between the modules of the enterprise data management system based on the cloud platform is as follows:

[0078] The data acquisition module collects enterprise data and transmits it to the encryption unit of the security module. The encryption unit encrypts the enterprise data and uploads it to the main cloud platform of the cloud computing platform. The main cloud platform analyzes and calculates the enterprise data. At the same time, the data synchronization module synchronizes the data of the main cloud platform to the subsidiary cloud platform to realize data backup and cloud platform data sharing. After the main cloud platform completes the analysis and calculation, it transmits the data to the verification unit. The verification unit completes the verification and decryption and transmits the data to the interactive module for relevant personnel to review and operate. The data recovery module is connected to the subsidiary cloud platform and the interactive module data, and is started when the relevant personnel trigger the recovery operation, and extracts the data in the subsidiary cloud platform for recovery.

[0079] Embodiment 1: Figure 1-4 As shown, in this embodiment, a certain e-commerce enterprise formally uses a cloud platform-based enterprise data management system in the present invention. In this embodiment, the main cloud platform works normally;

[0080] The data collection module introduces intelligent crawler technology to automatically capture data from various business systems, web pages, and document sources within e-commerce companies, and adopts a real-time data push mechanism to immediately collect new transactions, such as purchases, personnel changes, and other information into the system to ensure the timeliness of the data.

[0081] After the data acquisition module completes the acquisition, the original data G is transmitted to the encryption unit;

[0082] The encryption unit processes the data G collected and uploaded by the data collection module: first, a public-private key pair is generated using a random number generator. The public key PH is made public to the main cloud platform, and the private key QH is saved by the enterprise itself and uploaded to the verification module. AES is used to generate a random symmetric key H. Then, the symmetric key H is used to encrypt the data G to obtain the ciphertext F1=EnFrypt(G,H). Next, the encryption unit uses the public key PH to asymmetrically encrypt the symmetric key H to obtain F2=EnFrypt(H,PH);

[0083] Upload the ciphertext F1 and the encrypted symmetric key F2 to the main cloud platform;

[0084] The security unit of the main cloud platform decrypts the symmetric key F2 according to the public key PH to obtain the symmetric key H, and uses the symmetric key H to decrypt the ciphertext F1 to obtain the original data G for data management and other calculations;

[0085] The data storage unit is used to adopt a data hierarchical storage strategy for the information collected by the data collection module, calculate the enterprise data call frequency, obtain the median of the call frequency, store the data with an enterprise call frequency greater than the median frequency in a high-speed storage medium, and store the data with an enterprise call frequency less than the median frequency in a low-cost large-capacity storage device;

[0086] The data marking unit is used to perform anomaly detection on the stored data, mark and record the detected abnormal data, and then transmit it to the interaction module;

[0087] First, suppose there are d items of data in the enterprise data, and construct k isolated data trees based on the classification of the time series of the data; C represents the specific data, and its path length in the zth tree is ; The formula for calculating the average path length is as follows:

[0088]

[0089] Calculate whether data C is abnormal, and judge by calculating the abnormal score. First, calculate the correction value ;

[0090]

[0091] Among them, R is the harmonic value. The number of terms affects the correction value through the harmonic number. When the number of terms increases, the harmonic number will also increase. The smaller the average path length or the larger the number of terms, the larger the correction value will be. is a harmonic number, R is equal to ;

[0092] Then the anomaly score of data item C is;

[0093]

[0094] like The closer it is to 1, the more abnormal the data item C is. Compare the size of the abnormality, or sort by the size of the abnormality, mark and record the data item with the largest abnormality. For example, calculate the data of a certain customer. The consumption data is abnormal, which may include malicious order-swiping or abnormally high consumption. It will be marked and transmitted to the interaction module.

[0095] The data tagging unit can help companies detect abnormal customer data in a timely manner and take corresponding measures, such as further review, enhanced risk control or targeted marketing activities.

[0096] The data analysis unit uses deep belief networks to perform complex data analysis;

[0097] The deep belief network is composed of multiple restricted Boltzmann machines (RBMs) stacked together. For a single RBM, it has its own energy function , the energy function is expressed as:

[0098]

[0099] in, represents the amount of change in enterprise data, including but not limited to financial data, customer data, sales data, production data, market data, and operational data, Indicates Types of enterprise data, Represents the total number of categories of enterprise data collected; It means that different types of enterprise decisions or strategies are generated under the influence of different types of enterprise data. Each decision or strategy is manually quantitatively evaluated based on factors including importance, priority, implementation difficulty, expected benefits, resource investment and risk level, and converted into numerical values ​​for quantification; Indicates A decision or strategy based on the influence of enterprise data. Express the total number of decision or strategy types; is the bias of v; is the bias of u; is the connection weight between v and u, representing the strength of the association between them. The bias and connection weight are uniformly initialized and set, and the parameters are optimized through subsequent training;

[0100] Trained with a large amount of enterprise data to better optimize the deep belief network model; for each combination Calculate its energy function , constantly adjust the parameters , and , in order to minimize the energy function to achieve the training purpose; the parameter update formula of the training process is as follows:

[0101]

[0102]

[0103]

[0104] Where t represents the number of iterations. Represents the learning rate, which determines the step size of parameter updates. It can be selected and generated through grid search or adaptive adjustment. Grid search is an exhaustive parameter search method, and adaptive adjustment means adjusting according to the model training process.

[0105] Compute partial derivatives:

[0106]

[0107] represents the partial derivative, Represents enterprise data and decision making The degree of influence of , substitute the data after partial derivative calculation into the parameter update formula:

[0108]

[0109] After calculating all the training data, the parameters that ensure the minimization of the energy function are determined as the final parameters. The deep belief network after the final training is able to change according to the input of new enterprise data. , calculated with The energy function results ,for , choose the energy function Smallest The data analysis unit outputs corporate decisions, thereby providing decision-making basis for corporate managers.

[0110] For example, this e-commerce company inputs new company data:

[0111] Financial data: profit increased by 200,000 yuan; customer data: number of customers increased by 1,000; sales data: sales increased by 500 units;

[0112] Based on the impact of these data, possible decisions or strategies include: Decision 1: Expand production scale and increase output by 30%; Decision 2: Slightly expand production and increase output by 10%; Decision 3: Maintain the current production scale unchanged.

[0113] After calculation by the deep belief network, it is found that the energy function value of decision 2 is the smallest, so the corporate decision output by the data analysis unit is to slightly expand production and increase output by 10%.

[0114] The user rights management unit implements decentralized management of user rights based on blockchain technology, ensures that the rights information cannot be tampered with and has high security, and realizes automatic allocation and control of rights by setting and managing the access and operation rights of different users to the main cloud platform data.

[0115] Through the user rights management unit, in the enterprise's file access rights management, the employee's rights allocation records will be securely stored on the blockchain. No matter how the internal personnel of the enterprise change, or any external attack attempts, these rights information cannot be tampered with, ensuring that only authorized employees can access specific files.

[0116] After the calculation is completed, the data synchronization module will synchronize and transmit the relevant contents of the calculation to the affiliated cloud platform in real time;

[0117] The data synchronization module realizes the real-time synchronization and transmission of data between the main cloud platform and the subsidiary cloud platform, making up for the defects of untimely or error-prone data synchronization.

[0118] Then, the security unit of the main cloud platform encrypts the calculated data using the public key PH to obtain the ciphertext F', and transmits the ciphertext F' to the verification unit.

[0119] After receiving the ciphertext F', the verification unit uses the private key QH to decrypt the previously encrypted symmetric key F2 to obtain the symmetric key H'=GeFrypt(F2,QH), and uses the decrypted symmetric key H' to decrypt the calculation result ciphertext F' to obtain the plaintext result G'=GeFrypt(F',H'); and sends it to the interaction module.

[0120] Through the encryption unit and verification unit in the security module, enterprise data is strictly encrypted and verified, reducing the risk of data being stolen or leaked during cloud storage and transmission.

[0121] Relevant enterprise personnel can access and manage enterprise data through mobile phones anytime and anywhere through the mobile application provided by the interactive module.

[0122] Embodiment 2: Figure 1-3 As shown, in this embodiment, when the e-commerce enterprise in the first embodiment uses the system, the main cloud platform is suspended due to cloud service upgrades or other reasons, but the enterprise still needs the present invention to continue data management due to business requirements;

[0123] When the main cloud platform has not upgraded its cloud services, the data synchronization module performs data synchronization and transmission tasks between the main cloud platform and the subsidiary cloud platform in real time;

[0124] The monitoring unit monitors the status of the main cloud platform in real time. When it detects that the main cloud platform is undergoing a cloud service upgrade, it immediately issues an alarm to the interaction module and activates the emergency takeover unit.

[0125] After the emergency takeover unit is started, it takes over the data acceptance task of the main cloud platform and performs storage and backup, and maintains communication with the main cloud platform. When the main cloud platform returns to normal, the emergency takeover unit will synchronize the data received during the emergency takeover back to the main cloud platform through the data synchronization module.

[0126] When it is detected that the main cloud platform has completed the service upgrade, the data synchronization module synchronizes the data received by the subsidiary cloud platform to the main cloud platform.

[0127] The cloud computing platform including the main cloud platform and the subsidiary cloud platform effectively reduces the dependence on a single cloud service provider. Even if a cloud service provider encounters unforeseen circumstances such as technical failure, service interruption or business strategy adjustment, the enterprise can still rely on other cloud computing platforms to ensure data backup and storage. The data synchronization module realizes the real-time synchronization and transmission of data between the main cloud platform and the subsidiary cloud platform, making up for the defects of untimely or error-prone data synchronization.

[0128] Embodiment 3: Figure 1-3 As shown, in this embodiment, when the enterprise staff accesses the enterprise data through the interactive module, some data is lost due to a sudden power outage, and data recovery is now required;

[0129] Enterprise personnel issue recovery instructions to the data recovery module through the interactive module. After receiving the recovery instructions, the data recovery module sends the recovery instructions to the data recovery preparation unit in the subsidiary cloud platform. The data recovery preparation unit continuously monitors the integrity and availability of the backup data, and quickly responds to the need for recovery by pulling the calculation and analysis results synchronized in real time from the main cloud platform through the data synchronization module from the data backup unit.

[0130] The data backup unit receives data from the main cloud platform through the data synchronization module in real time and performs a complete backup, while recording the backup time point and version information; and transmits the data to be restored to the interactive module according to the recovery instructions.

[0131] The data backup and recovery mechanism of the subsidiary cloud platform and data recovery module ensures the security and availability of backup data. Even if the data center where the main cloud platform is located encounters natural disasters or other force majeure factors, the enterprise can still quickly restore the backup data through the subsidiary cloud platform, avoiding serious consequences caused by damage to the backup data.

[0132] The embodiments of the present invention are provided for the purpose of illustration and description. Although the embodiments of the present invention have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations of the present invention. Ordinary technicians in this field can change, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. An enterprise data management system based on a cloud platform, characterized in that: include: A data collection module, which is used to collect enterprise data information from the enterprise intranet by means of web crawlers or system interface docking. The enterprise data includes but is not limited to financial data, customer data, sales data, production data, market data and operation data, and enterprise historical decision-making or historical strategy information data; A cloud computing platform, comprising a main cloud platform and one or more subsidiary cloud platforms, for analyzing and computing enterprise data; A data synchronization module, which is used to perform data synchronization and transmission tasks between the main cloud platform and the subsidiary cloud platform in real time; A security module, which includes an encryption unit and a verification unit, and is used to encrypt the enterprise data before analysis and verify the enterprise data after analysis; An interactive module, which is used to provide users with analysis and calculation results of the cloud computing platform and an operation port, which allows relevant users to perform system settings and adjust calculation parameters in the cloud computing platform; A data recovery module, the data recovery module is used to extract data in the subsidiary cloud platform for recovery; The interaction relationship between the modules of the enterprise data management system based on the cloud platform is as follows: The data acquisition module collects enterprise data and transmits it to the encryption unit of the security module. The encryption unit encrypts the enterprise data and then uploads it to the main cloud platform of the cloud computing platform. The main cloud platform analyzes and calculates the enterprise data. At the same time, the data synchronization module synchronizes the data of the main cloud platform to the subsidiary cloud platform to realize data backup and cloud platform data sharing. After the main cloud platform completes the analysis and calculation, it transmits the data to the verification unit. The verification unit completes the verification and decryption and transmits the data to the interactive module for relevant personnel to review and operate. The data recovery module is connected with the subsidiary cloud platform and the interactive module data, and is started when the relevant personnel trigger the recovery operation, and extracts the data in the subsidiary cloud platform for recovery; The main cloud platform includes: A data storage unit, wherein the data storage unit stores the information collected by the data collection module by adopting a data partition storage strategy, calculates the call frequency of the enterprise data, obtains the median of the call frequency, stores the data whose enterprise call frequency is greater than the median frequency in a high-speed storage medium, and stores the data whose enterprise call frequency is less than the median frequency in a low-cost large-capacity storage device; A data marking unit, which is used to perform anomaly detection on the stored data, mark and record the detected abnormal data, and then transmit it to the interaction module; A data analysis unit, which is used to analyze enterprise data, analyze the degree of connection between data and decision-making based on the acquired enterprise data, calculate the strategy to be adopted after the enterprise data changes, and provide support for enterprise decision-making; A user rights management unit, which implements decentralized management of user rights based on blockchain technology, ensures that the rights information cannot be tampered with and is highly secure, and implements automated allocation and control of rights by setting and managing different users’ access and operation rights to the main cloud platform data; The data analysis unit uses a deep belief network to perform complex data analysis; The deep belief network is composed of multiple restricted Boltzmann machines (RBMs) stacked together. For a single RBM, it has its own energy function , the energy function is expressed as: in, represents the amount of change in enterprise data, including but not limited to financial data, customer data, sales data, production data, market data, and operational data, Indicates Types of enterprise data, Represents the total number of categories of enterprise data collected; It means that different types of enterprise decisions or strategies are generated under the influence of different types of enterprise data. Each decision or strategy is manually quantitatively evaluated based on factors including importance, priority, implementation difficulty, expected benefits, resource investment and risk level, and converted into numerical values ​​for quantification; Indicates A decision or strategy based on the influence of enterprise data. Express the total number of decision or strategy types; is the bias of v; is the bias of u; is the connection weight between v and u, representing the strength of the association between them. The bias and connection weight are uniformly initialized and set, and the parameters are optimized through subsequent training; Trained with a large amount of enterprise data to better optimize the deep belief network model; for each combination Calculate its energy function , constantly adjust the parameters , and , in order to minimize the energy function to achieve the training purpose; the parameter update formula of the training process is as follows: Where t represents the number of iterations. Represents the learning rate, which determines the step size of parameter updates. It can be selected and generated through grid search or adaptive adjustment. Grid search is an exhaustive parameter search method, and adaptive adjustment means adjusting according to the model training process. Compute partial derivatives: represents the partial derivative, Represents enterprise data and decision making The degree of influence of , substitute the data after partial derivative calculation into the parameter update formula: After calculating all the training data, the parameters that ensure the minimization of the energy function are determined as the final parameters. The deep belief network after the final training is able to change according to the input of new enterprise data. , calculated with The energy function results ,for , choose the energy function Smallest The data analysis unit outputs corporate decisions, thereby providing decision-making basis for corporate managers.

2. The cloud platform-based enterprise data management system according to claim 1, characterized in that: When the data collection module automatically grabs data from various business systems, web pages, and document sources within the enterprise through a web crawler, each newly collected data will be compared with all the data already in the system. For each data, the comparison is performed from multiple dimensions including the specific identifier of the data and key content fields. If it is found that the new data is completely consistent with an existing data in these dimensions, then the new data is determined to be duplicate data and is filtered out. A real-time data push mechanism is used, so that when new data is generated by the data source, it can be immediately collected into the system to ensure the timeliness of the data.

3. The cloud platform-based enterprise data management system according to claim 1, characterized in that: The data marking unit marking process is as follows: First, suppose there are d items of data in the enterprise data, and construct k isolated data trees based on the classification of the time series of the data; C represents the specific data, and its path length in the zth tree is ; The formula for calculating the average path length is as follows: Calculate whether data C is abnormal, and judge by calculating the abnormal score. First, calculate the correction value ; Among them, R is the harmonic value. The number of terms affects the correction value through the harmonic number. When the number of terms increases, the harmonic number will also increase. The smaller the average path length or the larger the number of terms, the larger the correction value will be. is a harmonic number, R is equal to ; Then the anomaly score of data item C is; like The closer it is to 1, the more abnormal the data item C is; compare the size of the abnormality, or sort according to the size of the abnormality, mark and record the data item with the largest abnormality.

4. The cloud platform-based enterprise data management system according to claim 1, characterized in that: The subsidiary cloud platform includes: A data backup unit, which receives data from the main cloud platform through a data synchronization module in real time and performs a complete backup, records the backup time point and version information, and uses a distributed database storage architecture to store different types of enterprise data in a distributed manner; A data recovery preparation unit, which continuously monitors the integrity and availability of backup data when the main cloud platform is operating normally, to ensure rapid response when recovery is needed; A monitoring unit, which monitors the status of the main cloud platform in real time and immediately issues an alarm when detecting that the main cloud platform is unavailable; The emergency takeover unit is activated after the monitoring unit issues an alarm, and then takes over the data acceptance task of the main cloud platform and performs storage and backup, and maintains communication with the main cloud platform. When the main cloud platform returns to normal, the emergency takeover unit will synchronize the data received during the emergency takeover back to the main cloud platform through the data synchronization module.

5. The cloud platform-based enterprise data management system according to claim 1, characterized in that: The encryption process of the encryption unit in the security module is as follows: Assume that the original data of the enterprise is G. The encryption unit processes the data G collected and uploaded by the data acquisition module: First, a pair of public and private keys are generated using a random number generator. The public key PH is made public to the main cloud platform, and the private key QH is saved by the enterprise itself and uploaded to the verification unit. AES is used to generate a random symmetric key H. The symmetric key H is then used to encrypt the data G to obtain the ciphertext F1=EnFrypt(G,H). Next, the encryption unit uses the public key PH to asymmetrically encrypt the symmetric key H to obtain F2=EnFrypt(H,PH). Upload the ciphertext F1 and the encrypted symmetric key F2 to the main cloud platform; The main cloud platform decrypts the symmetric key F2 according to the public key PH to obtain the symmetric key H, and uses the symmetric key H to decrypt the ciphertext F1 to obtain the original data G for data analysis and calculation; after completion, the calculated data is encrypted using the public key PH to obtain the ciphertext F'; The verification process of the verification unit is as follows: after the main cloud platform completes the data-related analysis and calculation, the calculation result is sent to the verification unit in the form of ciphertext F'; after receiving the ciphertext F', the verification unit uses the private key QH to decrypt the previously encrypted symmetric key F2 to obtain the symmetric key H'=GeFrypt(F2,QH); Use the decrypted symmetric key H' to decrypt the ciphertext F' of the calculation result to obtain the plaintext result G'; G'=GeFrypt(F',H'), send the plaintext result G' to the interaction module.

6. The cloud platform-based enterprise data management system according to claim 1, characterized in that: The main cloud platform also includes a decryption unit, which is used to decrypt the data uploaded by the encryption unit using the public key PH, and to encrypt the calculated data using the public key PH.

7. The cloud platform-based enterprise data management system according to claim 1, characterized in that: The interactive module allows users to view enterprise data and operating systems and receive alerts through mobile applications or computer web pages; the enterprise data includes abnormal data marked by the data marking unit and data decisions calculated and input by the data analysis unit; the operating system includes recovering enterprise data through the data recovery module and initializing the calculation parameters in the data analysis unit.

Images